Effects of body position, imaging plane, and observer on computed tomographic measurements of the lumbosacral intervertebral foraminal area in dogs

Brent M. Higgins Department of Musculoskeletal Biology, Leahurst Campus, University of Liverpool, Neston, Wirral, Cheshire, CH64 7TE, England.

Search for other papers by Brent M. Higgins in
Current site
Google Scholar
PubMed
Close
 BVSc
,
Peter J. Cripps Department of Epidemiology and Population Health, Leahurst Campus, University of Liverpool, Neston, Wirral, Cheshire, CH64 7TE, England.

Search for other papers by Peter J. Cripps in
Current site
Google Scholar
PubMed
Close
 BVSc, PhD
,
Martin Baker Department of Small Animal Teaching Hospital, Leahurst Campus, University of Liverpool, Neston, Wirral, Cheshire, CH64 7TE, England.

Search for other papers by Martin Baker in
Current site
Google Scholar
PubMed
Close
 MSc
,
Lee Moore Department of Veterinary Education in Institute of Teaching and Learning, Faculty of Health and Life Sciences, University of Liverpool, Neston, Wirral, Chesire, L69 7ZJ, England.

Search for other papers by Lee Moore in
Current site
Google Scholar
PubMed
Close
 MSc
,
Fay E. Penrose Department of Veterinary Education in Institute of Teaching and Learning, Faculty of Health and Life Sciences, University of Liverpool, Neston, Wirral, Chesire, L69 7ZJ, England.

Search for other papers by Fay E. Penrose in
Current site
Google Scholar
PubMed
Close
 BA
, and
James F. McConnell Department of Musculoskeletal Biology, Leahurst Campus, University of Liverpool, Neston, Wirral, Cheshire, CH64 7TE, England.

Search for other papers by James F. McConnell in
Current site
Google Scholar
PubMed
Close
 BVM&S

Abstract

Objective—To evaluate effects of imaging plane, flexion and extension, patient weight, and observer on computed tomographic (CT) image measurements of the area of the lumbosacral (L7-S1) intervertebral foramen (LSIF) in dogs.

Sample—12 dog cadavers (2 were excluded because of foraminal stenosis).

Procedures—In each cadaver, sagittal, sagittal oblique, transverse oblique, and double oblique CT images were obtained at 3 zones (entrance, middle, and exit zones) of the region of the lateral lumbar spinal canal that comprises the LSIF while the lumbosacral junction (LSJ) was positioned in flexion or extension. Barium-impregnated polymethylmethacrylate was used to fill the intervertebral foramina to aid boundary detection. Measurements of interest were obtained.

Results—Among the dog cadavers, there was large variability in LSIF cross-sectional areas (range, 0.12 to 0.44 cm2; SD, 0.1 cm2) and in foraminal angles required to obtain a double oblique plane in LSJ extension (SD, 8° to 9°). For LSIF area measurements in standard sagittal CT images, interobserver variability was 23% to 44% and intraobserver variability was 4% to 5%. Sagittal oblique images obtained during LSJ extension yielded smaller mean LSIF areas (0.30 cm2), compared with findings in sagittal images (0.37 to 0.52 cm2). The exit and middle zone areas were smaller than the entrance zone area in sagittal images obtained during LSJ extension.

Conclusions and Clinical Relevance—Repeated measurements of the LSIF area in images obtained during LSJ extension may be unreliable as a result of interobserver variability and the effects of dog positioning and CT slice orientation.

Abstract

Objective—To evaluate effects of imaging plane, flexion and extension, patient weight, and observer on computed tomographic (CT) image measurements of the area of the lumbosacral (L7-S1) intervertebral foramen (LSIF) in dogs.

Sample—12 dog cadavers (2 were excluded because of foraminal stenosis).

Procedures—In each cadaver, sagittal, sagittal oblique, transverse oblique, and double oblique CT images were obtained at 3 zones (entrance, middle, and exit zones) of the region of the lateral lumbar spinal canal that comprises the LSIF while the lumbosacral junction (LSJ) was positioned in flexion or extension. Barium-impregnated polymethylmethacrylate was used to fill the intervertebral foramina to aid boundary detection. Measurements of interest were obtained.

Results—Among the dog cadavers, there was large variability in LSIF cross-sectional areas (range, 0.12 to 0.44 cm2; SD, 0.1 cm2) and in foraminal angles required to obtain a double oblique plane in LSJ extension (SD, 8° to 9°). For LSIF area measurements in standard sagittal CT images, interobserver variability was 23% to 44% and intraobserver variability was 4% to 5%. Sagittal oblique images obtained during LSJ extension yielded smaller mean LSIF areas (0.30 cm2), compared with findings in sagittal images (0.37 to 0.52 cm2). The exit and middle zone areas were smaller than the entrance zone area in sagittal images obtained during LSJ extension.

Conclusions and Clinical Relevance—Repeated measurements of the LSIF area in images obtained during LSJ extension may be unreliable as a result of interobserver variability and the effects of dog positioning and CT slice orientation.

Degenerative lumbosacral stenosis is a common cause of back pain and neurologic dysfunction in dogs, and large-breed dogs are predisposed to the condition.1–3 Degenerative lumbosacral stenosis has a multifactorial etiopathogenesis including LS instability, soft-tissue hypertrophy, intervertebral disk protrusion, and osteophytosis, all of which can result in compression of the cauda equina and L7 nerve roots.4,5 Clinical signs are variable; signs of pain in the caudal portion of the back can progress to paraparesis, hind limb ataxia, and urinary or fecal incontinence as the compression increases.1,4,5

Experimental studies6–8 in dogs have revealed that increasing the mechanical compression of the cauda equina causes relative increases in neurologic and histologic abnormalities. However, results of studies3,9 in clinical patients have indicated that the relationship between the degree of compression detected via imaging methods and the clinical and surgical findings is more complex. Preoperative diagnosis of DLSS is usually based on results of clinical and neurologic examinations and diagnostic imaging. Computed tomography and MRI are the imaging techniques of choice for the diagnosis of DLSS because cross-sectional images can be obtained and image contrast is greater than that achieved via radiography. However, with regard to pathologic changes within the LS region, results of imaging procedures do not always correlate well with clinical signs or surgical findings.3,9–12 One possible reason for the lack of agreement between diagnostic imaging findings and clinical signs and surgical findings in dogs with DLSS is that the degree of neural compression within the LSIF is not accurately represented in images obtained from standard planes of view. An accurate, repeatable, and noninvasive method of measurement of the LSIF area in dogs may improve clinicians' ability to diagnose DLSS by allowing objective assessment of LS foraminal stenosis. This would also be of value in evaluating the effects of surgical treatments for foraminal stenosis.

Presently, there is no widely accepted standard imaging protocol for examination of the LS region, but MRI procedures commonly include acquisition of sagittal and transverse images through the L7-S1 intervertebral disk. By use of these imaging planes, views of the LSIF in a sagittal plane (parallel to the median plane) and in a transverse oblique plane (parallel to the L7-S1 intervertebral disk and perpendicular to the long axis of the vertebral column) are obtained. The LSIF and seventh lumbar (L7) nerve run in an oblique direction, with the nerve traversing caudally, ventrally, and laterally through the foramen. Therefore, the LSIF or L7 nerve do not appear in true cross section in either a sagittal or oblique transverse plane because the L7 nerve runs obliquely through the imaging plane. This results in partial volume averaging artifacts, which could potentially affect quantification of LSIF dimensions and assessment of foraminal stenosis. Computed tomographic images are usually obtained in a transverse plane13 and reformatted into other planes, the quality of which is dependent upon voxel size. By use of multislice helical CT scanners and 3-D MRI pulse sequences, isotropic voxels can be acquired. Isotropic voxels can be reformatted into any plane (multiplanar reconstruction) with minimal artifacts, thereby allowing images to be created in any plane and obliquity. Oblique plane images could be advantageous for evaluating the LSIF in dogs given that use of such images improves confidence in the diagnosis of foraminal stenosis in humans. Oblique plane images may also reveal lesions that are not clearly visible in standard imaging planes.14–19

Compression of the L7 nerve within the LSIF may be more likely to occur at the point at which the cross-sectional area of the foramen is smallest. By use of oblique imaging planes, it may be possible to obtain CT images of the foramen at the point of smallest cross-sectional area and to visualize the L7 nerve in true cross section, which may improve diagnostic accuracy. A double oblique plane is required to obtain images of the foramen and L7 nerve in true cross section, and to our knowledge, the imaging plane that would provide views of the LSIF (and therefore L7 nerve) in true cross section (ie, perpendicular to the long axis of the L7 nerve) has not been described. The angulation required to obtain images in oblique planes may vary depending on whether the LS junction of an animal is positioned in flexion or extension.

The LSIF is anatomically complex. Its boundaries are formed dorsally by the articular processes of L7 and S1 and their joint capsules and the ligamentum flavum, cranially and caudally by the L7 and S1 vertebral pedicles, and ventrally by the L7 and S1 vertebral bodies and annulus fibrosis of the L7-S1 intervertebral disk.20 The LSIF has been divided into 3 functional zones: the entrance, middle, and exit.11,21 The entrance zone is defined as the zone at the medial portion of the L7 vertebral pedicle. The middle zone is located at the center of the pedicle, and the exit zone is located at the lateral portion of the vertebral pedicle.11 The shape of the LSIF is not well understood because it is a dynamic space that changes according to the relatively large range of LS joint20 movement during flexion and extension. A morphological study22 of the LS region of the vertebral column in human cadavers revealed that the foraminal size increases during flexion and decreases during extension; studies21,23 in dogs have yielded similar findings. Variation in the shape of the LSIF among dogs and the role of LSIF conformation in foraminal stenosis have not been evaluated, to our knowledge.

The purpose of the study of this report was to evaluate effects of an oblique imaging plane, flexion and extension, patient size, and observer on CT image measurements of the area of the LSIF in dog cadavers. Our intent was to measure the angle of the CT imaging plane in which views of the LSIF were obtained in true cross section (ie, the imaging plane that results in views of the smallest cross-sectional area of the LSIF), compare the intra- and interobserver variability of CT measurements of foraminal area between traditional and novel imaging planes, investigate the available space for the L7 nerve as it passes through the LSIF in cadavers of dogs without foraminal stenosis as the LS region of the vertebral column was positioned in physiologic angles of flexion and extension, identify the effects of animal variables on foraminal size measurements obtained by use of traditional and novel imaging planes, and describe variations in foraminal shape among the dogs examined. We hypothesized that the LS angle in the dog cadavers would not be significantly different before and after dissection and plastication, that LSIF area measurements in CT images obtained by use of oblique imaging planes would be smaller than values in images obtained by use of a sagittal imaging plane, that intra- and interobserver foraminal area measurement variability (determined by use of CT imaging software) derived from CT images obtained in oblique planes would be less than the variability in measurements derived from images obtained in a standard sagittal plane, that the available space for the L7 nerve as it courses through the LSIF would be less than that outlined by the traditionally measured foraminal bony periphery and would differ with flexion and extension of the LS region, that body weight and L7 vertebral body length would be predictive of foraminal area measurements, and that foraminal shape would vary among dog cadavers.

Materials and Methods

Sample—Cadavers of 12 adult medium-large—breed dogs that were euthanized via IV administration of pentobarbital for reasons unrelated to the study were used. After euthanasia, the cadavers were immediately chilled at 4° to 5°C. Cadavers were excluded from the study when there was evidence of pathologic changes within the LS joint or the LSIF (eg, transitional vertebrae, L7 sacralization, obvious vertebral canal stenosis, LSIF osteophytosis, or L7-S1 intervertebral disk disease) or when the cadavers were not scanned and sectioned within 14 days of euthanasia. The cadavers that were not excluded from the study underwent CT scanning of the LS region of the vertebral column; scanning was performed with the LS junction in flexion and in extension.

Scanning protocol 1—Dog cadavers were positioned in dorsal recumbency, and CT scanning was performed by use of a 4-slice helical CT scannera (collimation, 0.5 mm; table feed, 1 mm). Images were acquired in the transverse plane and reconstructed by use of a sharp algorithm with 0.5-mm slice thickness. Images were viewed on a bone window setting (window level, 700 HU; window width, 4,000 HU) and reformatted into a sagittal plane by use of the CT workstation.b

Measurement of LS angle—The L7-S1 (LS) angle was determined in sagittal images obtained when the LS junction was positioned in flexion and extension. In those images, a line was drawn along the dorsal surface of the L7 vertebral body and another line was drawn along the dorsal surface of the S1 vertebral body; the angle formed by the intersection of the 2 lines was recorded as the LS angle (Figure 1).

Figure 1—
Figure 1—

Reformatted sagittal plane CT image of the LS region of the vertebral column of a dog cadaver to illustrate measurement of the L7-S1 (LS) angle. The LS angle was measured at the intersection of a line drawn on the dorsal surface of the L7 vertebral body and a line drawn on the dorsal surface of the S1 vertebral body (both lines shown in red). This image was obtained during extension of the LS junction (the other lines in this image relate to image processing and are not relevant to the measurement process).

Citation: American Journal of Veterinary Research 72, 7; 10.2460/ajvr.72.7.905

Sectioning protocol—Following completion of scanning protocol 1 and the calculation of LS angles, the cadavers were sectioned in the median plane and the spinal cord and soft tissues inside the LSIF were manually removed from the median surface. Half of each cadaver was positioned so that the LS junction was flexed, and the other half was positioned so that the LS junction was extended. The half of the cadaver (ie, left or right) that was placed in flexion or extension was randomly selected via removal of markers from a bag. To mimic physiologic positioning, the LS junction of each cadaver half was positioned in flexion or extension at the same angulation (a protractor was used for measurement) as that determined from the individual cadaver's presectioning CT images. This LS angle was held in position by use of Backhaus towel clamps, which were clamped through vertebral bone (to avoid relaxation of soft tissue) and fixed together with non-elastic ties.

Plastication protocol—To aid identification of the foramen on oblique plane images and to help define the foraminal boundaries when CT images were viewed on a bone window setting, the foramina were filled with a plastic material; the CT attenuation of this material is intermediate between attenuations of bone and soft tissues. Presence of this material allowed the area occupied by soft tissues within the LSIF to be viewed clearly on a bone window setting (Figure 2). A preliminary investigation (unpublished data) indicated that barium diluted (ratio of 1:2) with calcium carbonate powder and then mixed with PMMA (1:10 ratio) produces a plastic material that can easily be differentiated from bone and soft tissues on CT images viewed with bone window settings. The LSIF and vertebral canal were filled with the barium-impregnated PMMA at room temperature (approx 21°C); when the PMMA had set (after an interval of approx 30 minutes), the cadavers were frozen at −17°C.

Figure 2—
Figure 2—

Photograph of the LSIF sectioned in a sagittal plane at the midpedicle level in a dog cadaver (A) and a sagittal plane CT image though the LSIF at the level of the middle zone in another dog cadaver (B). In panel A, the foramen is viewed from the lateral aspect and barium-impregnated PMMA that has been injected into the foramen is visible (arrowheads). The L7-S1 intervertebral disk (asterisk) and articular process of L7 (arrow) are identifiable. In panel B, the PMMA-filled foramen is evident (star). The injection of the plastic material into the foramen allows delineation of soft tissue boundaries and helps to identify the orientation of the foramen in oblique plane CT reconstructions.

Citation: American Journal of Veterinary Research 72, 7; 10.2460/ajvr.72.7.905

Scanning protocol 2—After thawing at room temperature, CT scanning was repeated for each cadaver half; the scanning settings were similar to those used in scanning protocol 1. A plastic syringe filled with barium-impregnated PMMA was placed in an oblique plane adjacent to the dorsum of the cadaver half and included in the scan field of view. This provided a means by which accuracy of area measurements and effect of reformatting images on area measurement could be assessed. Image data were viewed in the CT 3-D workstationb and initially reformatted into 3 orthogonal planes (dorsal, sagittal [Figure 2], and transverse). Cadavers with foramina that were incompletely filled (as assessed from the CT images) were reinjected with PMMA and rescanned. Images obtained in sagittal, dorsal, and transverse planes were saved and viewed on the CT workstation to allow length, angle, and area measurements to be performed.

Measurement of the length of the L7 vertebral body—For each cadaver half, the length of the L7 vertebral body was measured along the middle part of the centrum in sagittal plane reconstruction images. The value was recorded.

Variability and accuracy assessments of CT measurement—To estimate the errors attributable to reformatting of the images and errors in measurement of PMMA in CT images, a syringe was filled with PMMA to create a PMMA mold of it. The CT images of this mold were compared with its physical measurements. The CT images of the syringe mold were reformatted into transverse and sagittal images on the same CT workstation as that used during the aforementioned imaging procedures. By use of the same method as that used for measurement of foraminal area in cadaver images, the areas and diameter of the syringe mold were measured on the CT console and compared with measurements of the same plastic mold made with vernier calipers that had an accuracy of 0.02 mm. Each CT measurement was made 3 times by the same 2 observers (JFM and MB) who performed the LSIF area measurements. The physical measurements were made by a third observer (BMH). From measurements made with the vernier calipers, areas were calculated by use of the formula for area of a circle (area = πr2, where r is the radius).

Foraminal angulation—The orientation of the LSIF relative to the median plane (the spinous process of L7 used as a landmark) was measured from the transverse plane reconstructions by manually fitting a straight line through the center of the PMMA within the LSIF so that the line was equidistant between the dorsal and ventral boundaries of the foramen and bisected the foramen at its narrowest point (Figure 3). This provided an angle of the foramen relative to the median plane (designated as the transverse plane angle). The orientation of the LSIF relative to the median plane from dorsal plane reconstructions was measured by manually fitting a straight line through the center of the PMMA (designated as the dorsal plane angle; Figure 4). These 2 angles (transverse and dorsal plane angles) describe the orientation of the soft tissue component of the LSIF as it runs caudolaterally from the vertebral canal.

Figure 3—
Figure 3—

Transverse CT image of the LSIF in a dog cadaver to illustrate the method used to calculate the transverse plane angle (a). The median plane was identified by use of the spinous process and vertebral body of L7 as landmarks (red line) so that the image plane was parallel to the spinous process. On dorsal plane images (not shown), the median plane was aligned so that it was parallel to the vertebral pedicles and perpendicular to the LS vertebral end plates. The orientation of the foramen was determined with the PMMA (star) as a guide. A straight line was manually fitted through the center of the PMMA within the LSIF (blue line) so that the line was equidistant between the dorsal and ventral boundaries of the foramen and bisected the foramen at its narrowest point. This provides an angle of the foramen relative to the median plane (the transverse plane angle). The syringe used to inject the foramen with PMMA was refilled with PMMA (asterisk) and used as a calibration phantom in the CT image.

Citation: American Journal of Veterinary Research 72, 7; 10.2460/ajvr.72.7.905

Figure 4—
Figure 4—

Dorsal plane CT image of the LSIF in a dog cadaver to illustrate the method used to calculate the dorsal plane angle (a). The median plane was identified by placing a line parallel to the vertical pedicles and perpendicular to the LS vertebral end plates. On transverse plane images (not shown), the median plane was placed so that it was parallel with the L6, L7, and sacral spinous processes. The orientation of the foramen was determined with the PMMA (star) as a guide. A straight line was manually fitted through the center of the PMMA within the LSIF (blue line) so that the line was equidistant between the dorsal and ventral boundaries of the foramen and bisected the foramen at its narrowest point. This provides an angle of the foramen relative to the median plane (the dorsal plane angle).

Citation: American Journal of Veterinary Research 72, 7; 10.2460/ajvr.72.7.905

Standard paramedian sagittal and transverse images and novel transverse oblique, sagittal oblique, and double oblique images—At the start of the study, both observers (JFM and MB) jointly created the multiplanar reconstructions of the images, and anatomic landmarks and angulation of imaging plane relative to landmarks were agreed by consensus. The transverse and dorsal plane angles were used to create specific oblique images of the LSIF. Sagittal oblique images were created by angulating standard sagittal images dorsolateroventromedially so that they were perpendicular to the transverse plane angle at the foramen's narrowest diameter (Figure 5). Transverse oblique images were created by angulating standard transverse images craniolaterocaudomedially so that they were perpendicular to the dorsal plane angle (Figure 6). A double oblique image was created by positioning slices so that they were perpendicular to both the transverse plane and dorsal plane angles centered at the middle zone of the foramen. In addition, conventional sagittal paramedian images were created at the entrance, middle, and exit zones (the pedicles used as a landmark) as described in a previous study.21 Dorsal plane reformatted images were used to obtain the paramedian sagittal images, entrance zone images were obtained by placing the cursor over the cortical bone on the medial aspect of the L7 pedicle, middle zone images (Figure 2) were obtained by placing the cursor in the middle of the pedicle, and exit zone images were obtained by placing the cursor over the cortical bone on the lateral aspect of the L7 pedicle. Sagittal reformatted images were parallel with the spinous processes of L7, as assessed from the transverse images. The reformatted images were saved electronically with a slice thickness of 0.5 mm and viewed with a bone window setting (window level, 700 HU; window width, 4,000 HU).

Figure 5—
Figure 5—

Transverse CT image of the LSIF in a dog cadaver to illustrate the method used to obtain an image in a sagittal oblique plane (blue line). The image plane was placed perpendicular to the transverse plane angle (red line drawn parallel to the PMMA [star]). The image plane crossed the narrowest part of the foramen. The syringe used to inject the foramen with PMMA was refilled with PMMA (right side of image) and used as a calibration phantom in the CT image.

Citation: American Journal of Veterinary Research 72, 7; 10.2460/ajvr.72.7.905

Figure 6—
Figure 6—

Dorsal plane CT image of the LSIF in a dog cadaver to illustrate the method used to obtain an image in a transverse oblique plane (blue line). The image plane was placed perpendicular to the dorsal plane angle (red line drawn parallel to the PMMA [star]). The image plane crossed the narrowest part of the foramen.

Citation: American Journal of Veterinary Research 72, 7; 10.2460/ajvr.72.7.905

CT measurement of foraminal area—The area of the LSIF was measured on a computer workstationb by manually tracing around the bony margins in CT images. The cursor was placed on the foraminal surface of the cortical bone. The PMMA mold of the soft tissue component was smaller than the bone, and the outer surface of the PMMA mold was also manually traced. There was a difference between the 2 measurements because the plastic material was separated from the bone by a narrow hypodense rim. For some transverse and double oblique images, the caudal portion of the foramen was not visible because of the alignment of the imaging plane; the caudal portion of the image was medial to the pedicle and facets of S1. If the foramen was not bounded by bone on all sides, only the cranial portion of the foramen was measured. The caudal boundary was created by drawing a line that was vertically aligned with the caudal end plate of L7 (Figure 7); the area cranial to that caudal boundary was then measured. To allow comparison of the transverse and double oblique images in which the caudal portion of the foramen was not visible with sagittal plane images, the cranial parts of the foramen in the latter image planes were also measured in the aforementioned manner. The LSIF areas determined by use of the bone and the periphery of the plastic material were recorded for each double oblique, transverse oblique, sagittal oblique, and parasagittal image at the entry, middle, and exit zones of the LSIF.21 Two observers (1 board-certified radiologist [JFM] and 1 radiographer with 1 year's experience in CT [MB]) independently created oblique and sagittal images; each observer measured all variables 3 times, and each was unaware of the other's findings. In this way, the PMMA measurements were used to represent the LSIF cross-sectional area available for the L7 nerve, and conventional bone measurements around the periphery of the foramen were used to measure the size of the LSIF.

Figure 7—
Figure 7—

Transverse oblique CT image of the LSIF (in extension) in a dog cadaver to illustrate measurement of the cranial portion of the LSIF, which was performed when the foramen did not appear bounded by bone on all sides in an image because of the alignment of the imaging plane. In this view, the transverse process of the L7 vertebral body (arrowhead), lateral part of the L7 facet (arrow), and the L7-S1 disk (asterisk) are evident; the foramen is filled with PMMA (star). To measure the cranial portion of the foramen, a caudal boundary was created by drawing a line that was vertically aligned with the caudal endplate of L7 (red line and dotted red line, respectively); the area of the foramen cranial to the line was measured. Dashed gray line in this picture is related to image processing and is not relevant to the measurement process.

Citation: American Journal of Veterinary Research 72, 7; 10.2460/ajvr.72.7.905

Foraminal shape—The reconstructed images were subjectively evaluated by a board-certified radiologist (JFM) for common shape characteristics and variability in foraminal appearance in traditional and oblique imaging planes.

Data and statistical analyses—Data and statistical analyses were supervised by a statistician (PJC). Data were entered into a spreadsheetc and analyzed by use of proprietary software.d,e Data were checked for obvious errors, and basic descriptive analysis was performed. The suitability of the data for ANOVA and t tests was confirmed by testing the residuals for normality via quantile plots and the Anderson-Darling normality test; plots of residuals against fitted values were examined to check for systematic trends and unequal variances.

Preliminary analyses revealed a strong effect of the individual dog cadaver, and most comparisons involved the means of variables (mean values for individual dogs). The paired t test was used to account for matching and was used to compare the mean LS angle before and after dissection; CT measurements of foraminal area during flexion versus extension, in different imaging planes, and in different zones; and PMMA measurement versus the bony margin measurement.

A fixed-effect ANOVA was used to obtain estimates of intra- and interobserver variation for foraminal area measurements. Dog cadaver and observer identity were used as predictors, each zone was considered separately, and the effect of each observer (each of whom provided 3 replicate observations for each zone) was nested within dog. Differences between estimates derived from CT measurement and physical measurements were examined by use of an ANOVA for fixed effects; comparisons between individual means were performed by use of t tests.

Least squares regression analysis was used to examine the ability of cadaver weight, L7 measurement, and LS angle to predict foraminal area measurements. The data were collapsed so that each dog cadaver provided a single mean value for each site, and the predictors were then offered to the model; R2 values were used to assess the proportion of the variation (sum of squares) explained by each individual predictor and also for all 3 predictors used together in the model.

Statistical tests examined a null hypothesis of no difference between means, and significance was set at a value of P < 0.05 with a 2-sided alternative hypothesis. Corrections for multiple comparisons were not applied. The coefficient of variation was defined as the mean divided by the SD.

Results

Two dog cadavers were excluded after initial CT imaging evaluations because they had osteophytic LSIF stenosis. The study was performed with cadavers of 4 Staffordshire Bull Terrier crosses, 2 German Shepherd Dog crosses, 1 German Shepherd Dog, 1 Labrador Retriever cross, and 2 crossbred dogs. Mean cadaver weight was 20.8 kg (range, 14.5 to 27.4 kg); there were equal numbers of male and female dogs. For 2 cadavers, the vertebral column was not maintained in extension for collection of all of the required data. Any extension data obtained from these 2 cadavers were excluded from the extension data analysis; however, flexion data were included in the flexion data analysis.

LS angle measurements—For each of the 10 cadavers, the LS angles were measured on CT images obtained prior to and after dissection. With the LS junction in flexion, the mean ± SD predissection angle was 178 ± 4° (range, 172° to 185°) and the postdissection angle was 172 ± 6° (range, 164° to 179°); these angles were significantly (P = 0.001) different. With the LS junction in extension, the mean predissection angle was 142 ± 6° (range, 129° to 154°) and the postdissection angle was 144 ± 7° (range, 132° to 152°); these angles did not differ significantly (P = 0.19).

Foraminal angulation—To obtain double oblique images of the LSIF (ie, images in which the foramen appeared in true cross section), transverse and dorsal plane foraminal angles were assessed in transverse and dorsal plane CT reconstructions. Half of each cadaver was placed in flexion (n = 10), and the other half was placed in extension (8; data not available for 2 cadaver halves). The mean transverse plane angles in flexion and extension were 120 ± 4° (range, 106° to 137°) and 126 ± 9° (range, 118° to 139°), respectively. The mean dorsal plane angles in flexion and extension were 135 ± 4° (range, 128° to 143°) and 134 ± 8° (range, 119° to 146°), respectively.

Variability and accuracy assessments of CT measurement—Analysis of variance revealed significant differences between reconstructed CT measurements and physical measurements (made by use of vernier calipers) of the length and area of the cast PMMA cylinder (Table 1). Length measurements derived from the reconstructed images were slightly lower (by approx 1%; P < 0.05) than those derived by use of the vernier calipers. The observers' measurements differed from each other by < 1% (P > 0.3). Area measurements derived from the reconstructed images were greater (by 3% to 9%) than the physical area measurements. The difference between the CT and physical measurements was significant (P < 0.05) for observer 2 but not for observer 1. The difference in mean CT area measurements between the 2 observers was 5.5% of their combined mean (P < 0.001).

Table 1—

Mean ± SD (range) measurements of length and cross-sectional area of a PMMA-filled syringe determined by 2 observers from measurements of CT images and corresponding physical measurements of the syringe mold.

Type of measurementLength (cm)Area (cm2)
CT
   Observer 14.18 ± 0.012 (4.17–4.19)1.19 ± 0.025 (1.19–1.22)
   Observer 24.17 ± 0.017 (4.15–4.18)1.26 ± 0.010 (1.25–1.27)
Physical4.24 ± 0.005 (4.23–4.24)1.15 ± 0.002 (1.153–1.57)

Area was measured by manually tracing the mold boundary in a 2-D cross-sectional image generated on the CT workstation. The physical measurements (determined with vernier calipers by a third person) of the mold were acquired in an attempt to estimate the errors in measurement attributable to reformatting of CT images or the method of physical measurement. Physical measurement of the cross-sectional area was calculated by use of the formula for area of a circle following measurement of the mold diameter.

Intra- and interobserver variation in foraminal area measurements determined by use of bony margins in CT images obtained during LS junction extension—Each observer provided 3 LSIF area measurements (3 for each LSIF zone) for each of the 8 cadaver halves that were positioned in extension. Intra- and interobserver variation for foraminal area measurements determined from sagittal, sagittal oblique, and double oblique CT images were assessed by use of a fixed-effect ANOVA (Table 2). Via least squares regression analysis, R2 values were calculated to assess the proportion of the variation explained by cadaver weight, length of the L7 vertebral body, and LS angle and also for all 3 of those variables together. For area measurements of the 3 LSIF zones in standard sagittal images, there was considerable interobserver variability (interobserver coefficient of variation, 24% to 44%) but intraobserver variability was smaller (5.0% to 5.3%).

Table 2—

Percentage of variance (attributable to dog cadaver weight, L7 vertebral body length, or LS angle) and intra- and interobserver variability in LSIF area measurements made in CT images of the LSIF (obtained in various imaging planes during extension of the LS junction) in 8 dog cadavers.*

  Adjusted R2 value   
Imaging plane and regionDog cadaver weightL7 vertebral body lengthLS angleDog cadaver weight, L7 length, and LS angleCoefficient of variation (%) for interobserver variabilityVariability (%) explained by differences between the 2 observersCoefficient of variation (%) for intraobserver variability
Sagittal entrance zone66a76a06724.335.8b5.3
Sagittal middle zone65c73a0743734.6b5.0
Sagittal exit zone66a76a07744.150.2b5.0
Sagittal oblique1555c06923.250.6b4.4
Double oblique1101123.6b5.8

Sagittal images of each of the 3 zones of the LSIF were obtained as follows: the entrance zone is defined as the zone at the medial portion of the L7 vertebral pedicle, the middle zone is located at the center of the pedicle, and the exit zone is located at the lateral portion of the vertebral pedicle. Adjusted R2 values represent the proportion of the variance in foraminal area measurements that was explained by the named predictor; values were obtained via regression analysis with a single mean estimate of area for each dog cadaver.

Although 10 cadavers were included in the study, the vertebral columns of 2 cadavers were not maintained in extension during data collection and those values were excluded from extension data analyses.

Percentage of variance in mean foraminal areas that was not explained by a model including the cadaver weight, L7 length, and LS angle as predictors.

Value obtained via an ANOVA with observer identity nested within dog cadaver.

Significant values are indicated by superscripted letters as follows: a = P < 0.01; b = P ≤ 0.001; and c = P < 0.05.

CT measurement of foraminal area—From the various CT images, LSIF areas were determined by use of bony margins and the percentage reduction in foraminal areas in extension versus flexion was calculated (Table 3). There was a reduction in mean bony-margin area measurements of the LSIF for all planes obtained when the LS junction was extended, compared with area measurements obtained when the LS junction was flexed (P ≤ 0.001), except in the double oblique, transverse oblique, and cranial portion of the entrance zone (P > 0.2). Foraminal areas were reduced from 26% to 70% during extension, compared with areas during flexion, for the sagittal and sagittal oblique planes (bone used as the boundary). Mean sagittal oblique foraminal areas were significantly (P < 0.001) smaller than the mean conventional sagittal planes in both flexion and extension (measured by use of bony margins) for the entrance zone only. When only the cranial portion of the foramen was measured, foraminal areas measured in the double oblique and transverse oblique planes were consistently smaller than areas measured in the standard planes during flexion (P ≤ 0.013) but not during extension. The middle and exit zones were smaller than the entrance zone of the LSIF; this difference was significant (P < 0.005) in all imaging planes during extension and for the middle zones during flexion. As expected, the LSIF area measurements derived by use of the PMMA were consistently smaller (P < 0.005) than the area measurements derived by use of the bony margin (smaller by a mean value of 41% during extension and 32% during flexion; Table 4).

Table 3—

Mean ± SD (range) measurements of LSIF areas (cm2) derived by use of the bony margins of the foramen in CT images obtained in various imaging planes during flexion and extension of the LS junction in 10 dog cadavers.*

Imaging plane and regionFlexionExtensionReduction in foraminal area (flexion vs extension [%])
Sagittal entrance zone0.74 ± 0.16 (0.53–1.02)0.52 ± 0.12 (0.30–0.67)32 ± 9.5 (26 to 43)
Sagittal middle zone0.66 ± 0.18 (0.43–0.92)0.35 ± 0.13 (0.24–0.64)48.6 ± 13.3 (31 to 70)
Sagittal exit zone0.67 ± 0.19 (0.36–0.95)0.37 ± 0.15 (0.22–0.69)46.4 ± 14.3 (23 to 61)
Sagittal oblique0.63 ± 0.21 (0.34–0.97)0.30 ± 0.06 (0.24–0.40)55.9 ± 7.6 (45 to 67)
Cr-double oblique0.20 ± 0.06 (0.11–0.32)0.22 ± 0.11 (0.12–0.44)−8.3 ± 42.5 (−106 to 21)
Cr-transverse oblique0.23 ± 0.07 (0.15–0.36)0.24 ± 0.08 (0.15–0.4)−13.3 ± 44.6 (−113 to 32)
Cr-sagittal entrance zone0.32 ± 0.11 (0.21–0.55)0.34 ± 0.11 (0.18–0.49)−12.1 ± 21.2 (−38 to 22)
Cr-sagittal middle zone0.26 ± 0.1 (0.17–0.48)0.23 ± 0.07 (0.13–0.32)5.0 ± 22.8 (−37 to 36)
Cr-sagittal exit zone0.30 ± 0.08 (0.16–0.48)0.21 ± 0.04 (0.15–0.27)24.9 ± 14.7 (6 to 45)

Cr = Only the cranial portion of the foramen was measured.

See Table 2 for key.

Table 4—

Mean ± SD (range) measurements of LSIF areas (cm2) derived by use of the visible margins of PMMA (previously injected into the foramen) in CT images obtained in various imaging planes during flexion and extension of the LS junction in 10 dog cadavers.*

Imaging plane and regionFlexionExtension
Sagittal entrance zone0.54 ± 0.15 (0.37–0.790.31 ± 0.09 (0.19–0.44)
Sagittal middle zone0.42 ± 0.12 (0.28–0.62)0.19 ± 0.07 (0.13–0.34)
Sagittal exit zone0.38 ± 0.11 (0.22–0.54)0.17 ± 0.08 (0.11–0.31)
Sagittal oblique0.39 ± 0.13 (0.19–0.59)0.15 ± 0.04 (0.1–0.21)
Double oblique0.39 ± 0.13 (0.16–0.60)0.17 ± 0.07 (0.09–0.34)

See Table 2 for key.

Predictors of foraminal area—Least squares regression analysis was used to investigate the ability of cadaver weight, L7 measurement, and LS angle to predict foraminal area measurements. The analyses revealed R2 values between 1% and 77%, indicating that between 33% and 99% of the variation could not be explained by these predictors (Table 2).

Foraminal shape—Large variation in the CT appearance of the LSIF was evident and attributed primarily to variation in shape of the cranial portion of the foramen. In middle-zone sagittal plane images, there were 2 broad types of foraminal conformation: a vertically orientated hook-shaped bony margin and an ovoid shape of the cranial portion of the foramen (Figure 8). The hook shape was created by a groove within the cranioventral portion of the foramen within L7. The hook shape became more marked in the exit zones. For middle-zone sagittal images, the foramina were evenly divided into ovoid and hook-shaped foramina, although there was gradation between the 2 types.

Figure 8—
Figure 8—

Representative reformatted CT images of the LSIF (in flexion) in 2 dog cadavers to illustrate the 2 typical conformations of the middle zone of the foramen viewed in a sagittal plane. The vertebral body and pedicle of L7 and sacral body (at the right of the image) are visible in this image plane; the foramen is shaded black for illustrative purposes. The foramen had a hook-shaped bony margin cranially in some cadavers (A) and an ovoid shape in the other cadavers (B). Compared with the more vertical orientation of the cranial portion of the hook-shaped foramen, the cranial portion of the ovoid foramen has a more horizontal orientation.

Citation: American Journal of Veterinary Research 72, 7; 10.2460/ajvr.72.7.905

Sagittal oblique images of the LSIF obtained during LS junction extension differed from middle-zone sagittal images in that more of the pedicle and facet of S1 and less of the L7 facet joint were visible (Figure 9). During flexion of the LS junction, the sagittal oblique image was similar to middle-zone sagittal images. In transverse oblique images, the cranial portion of the foramen was visible; it appeared ovoid in all dog cadavers and was similar during flexion and extension. In double and transverse oblique images, the facet joints and pedicle of S1 were not visible but the transverse process of L7 was seen cranial to the foramen (Figure 10). The LSIF shape in the double oblique plane was similar to its shape in the transverse oblique plane during flexion but not during extension (Figure 11); during extension of the LS junction, the double oblique plane resulted in an image in which the craniolateral aspect of the foramen, part of the transverse process of L7, and the L7 facets were visible.

Figure 9—
Figure 9—

Representative sagittal oblique CT images of the LSIF in a dog cadaver during extension (A) and flexion (B) of the LS junction. Notice that the foraminal shape in flexion is similar to that observed in a standard middle-zone sagittal image. When the LS junction is extended, the L7 facet is not as clearly seen and more of the pedicle and facet of S1 is visible. The area of the bone margin and PMMA of the LSIF is greatly reduced in extension compared with flexion. In these images, the foramen is filled with PMMA (star). Dashed gray line in this picture is related to image processing and is not relevant to the measurement process.

Citation: American Journal of Veterinary Research 72, 7; 10.2460/ajvr.72.7.905

Figure 10—
Figure 10—

Representative transverse oblique CT image of the plane of the LSIF in a dog cadaver. In this view, the transverse process of the L7 vertebral body (arrowhead), lateral part of the L7 facet (arrow), and the L7-S1 disk (asterisk) are evident; the foramen is filled with PMMA (star). Notice that the caudal portion of the foramen is not visible in this imaging plane.

Citation: American Journal of Veterinary Research 72, 7; 10.2460/ajvr.72.7.905

Figure 11—
Figure 11—

Representative double oblique CT images of the LSIF in a dog cadaver during extension (A) and flexion (B) of the LS junction. When the LS junction is flexed, the shape of the foramen is similar to that observed in a transverse oblique image (the caudal portion of the foramen is not visible). In panel A, the transverse process of L7 (arrowhead), the articular process of L7 (arrow), and the L7-S1 intervertebral disk (asterisk) are evident. In panel B, the articular process of L7 (small arrow), the lateral aspect of the articular process of S1 (large arrow), the L7-S1 intervertebral disk (asterisk), and the transverse process of L7 (arrowhead) are evident. In these images, the foramen is filled with PMMA (star).

Citation: American Journal of Veterinary Research 72, 7; 10.2460/ajvr.72.7.905

Discussion

In the present study, the degrees of flexion and extension of the LS junction were similar to those found in another CT investigation11 of the LSIF area and LS angle in dogs with LS disease, which indicated that the dog cadavers were positioned within a physiologic range of extension and flexion. The angles measured during LS junction extension did not significantly change before and after dissection, except for findings in 2 cadavers in which there was failure to maintain extension, probably because of failure of the clamps. Although there was a significant difference between pre- and postdissection angles measured during LS junction flexion, the difference was small (6°).

Accurate measurement of the LSIF area from CT images is difficult. Manual tracing of the boundaries of the foramen may not be accurate if the soft tissue components cannot be visualized, and systematic errors may exist because of the CT technique and artifacts. Attempts to improve accuracy of foraminal area measurements by use of a step-window display resulted in difficulty in assessing bony margins.11 Linear measurements made on CT images are accurate, but manual tracing of areas leads to errors and considerable inaccuracy.24 Quantification of the foraminal area has been undertaken in dogs11,21 and involves freehand tracing around the margins of the foramen in CT images. This process may be inaccurate because of errors in positioning the cursor and in accurately defining the boundaries of the foramen. It is not known how accurately bony-margin tracing reflects the space available within the foramen for the L7 nerve and periradicular soft tissues. It is possible that the available area is smaller than that measured because of the presence of the periosteum and inaccuracies in defining the exact position of the disk margins, ligamentum flavum, and bony margins. We attempted to improve boundary detection and to model the soft tissue component by injecting a plastic material into the foramen of the study cadavers. Comparison of areas derived by use of the bony margin of the foramen with those derived by use of the injected PMMA revealed that the PMMA-derived areas were consistently smaller. A linear halo was present between the bone and surface of the PMMA. The origin of this line was probably largely artifactual; it may have represented a rebound artifact,25 although there may have been some contribution from the periosteum and other soft tissues. The rebound effect is a consequence of the use of thin CT slices and a high-frequency image reconstruction algorithm. In a recent study,25 this artifact only appeared with a pitch of 0, whereas images were obtained with a pitch of 0.5 in the present study. Thin CT slices and a bone reconstruction algorithm were used in the present study because it was thought that this would result in greatest accuracy when reformatting the images into other planes.

Physical measurements of a plastic mold made from a syringe filled with PMMA were used as a standard in the present study to evaluate errors in the CT image reformatting. Linear measurements were accurate, although the CT measurements underestimated the true length by < 1%, which was significant. This may have been a result of the high image window level used, as the use of higher window levels results in underestimation of length measured in CT images.26,27 Measurements made in CT images are most accurate when the window level is set at a value that is half the difference in HU between the object and the background.26,27 Lower window levels result in over-estimation of size, and higher window levels result in underestimation of size.26,27 The optimal window to accurately assess foraminal area has not been described, to our knowledge. Previously, a bone window or a step-window display has been used in an attempt to reduce bias due to window settings.11 In the present study, area measurements made on CT images overestimated the syringe area by 3% to 9%. This finding differs from results of a previous study24 (in which similar methods were used), which indicated that CT measurements underestimated area and had an accuracy of only 85%. The improved accuracy in the present study may have been attributable to the use of more modern CT equipment and the CT protocol. The coefficient of variation for area measurements was 6% in the present study, which likely represented the minimum interobserver variability that may be expected with manual CT area measurements because the syringe had distinct margins and a regular shape that allowed easy boundary detection.

A recent study11 to quantify foraminal area in dogs with LS disease did not reveal a direct correlation between foraminal area and clinical signs. Attempts to correlate objective measures of LS stenosis determined via MRI with clinical signs in dogs with degenerative LS stenosis have also found poor agreement.3 In humans, measurement of the cross-sectional area of the dural sac is considered more effective for diagnosis of central stenosis than are measurements of the osseous canal.28 Cross-sectional area measurements of the vertebral canal in humans with DLSS may be misleading because patients with severe stenosis may have larger cross-sectional areas than do healthy people as a result of the large variability within and between patients; thus, it is unlikely that a purely objective value could predict stenosis.29 Subjective assessment of DLSS severity via MRI in humans has revealed that there may be poor interobserver agreement and that intraobserver variability is only moderate.29 However, good intra- and interobserver agreement and good agreement between surgical and MRI findings have been identified when a clearly defined grading scheme is used.30 In humans, objective assessment of the LS foramina is reportedly difficult because small differences in section position can markedly affect foraminal diameters.31 A subjective grading scheme for assessment of foraminal stenosis and nerve root compromise in humans has been described,30–32 but no such scheme has been generated for use in dogs with DLSS, to our knowledge. There is conflicting evidence for the accuracy of the human grading scheme; good interobserver agreement for the grading scheme has been reported by Wildermuth et al,31 and yet poor sensitivity and specificity of the grading scheme have been reported by Attias et al.33

In the present study, there was marked variability in foraminal size among dog cadavers, and the significant interobserver variability suggested that absolute measurement of foraminal area as a means of quantifying stenosis may be unreliable and poorly repeatable. The intraobserver coefficient of variation for foraminal area measurements ranged from 4.4% to 5.3% when the standard sagittal images and sagittal oblique plane images were used; in a previous study11 of the LSIF area and LS angle in dogs with LS disease, the intraobserver coefficient of variation for foraminal area measurements was approximately 6.5%. Between the 2 observers involved in the present study, there was a significant difference in area measurements among all zones of the foramen, which were determined in sagittal plane images obtained during extension of the LS junction. Observer variability contributed up to 51% of the overall variation, although the actual differences in measurements were small (from 0.063 to 0.093 cm2). When the LS junction was in flexion, intraobserver variability was less and there was no significant difference between area determinations made by the 2 observers for the entrance and middle zones. The lower variability in area measurements obtained during LS junction flexion is probably simply a function of the fact that when tracing areas freehand on those CT images, there is less distance traced and perhaps less opportunity for error; however, errors may have a greater relative effect with small cross-sectional areas. Observer variability when the LS junction was flexed contributed to 8% to 10% of the overall variability, with mean difference in measurement of 0.04 cm2. Although observer variability may be lower when the LS junction is in flexion, it may be better to measure foraminal areas during extension because foraminal areas are smaller and extension generally exacerbates clinical signs. In the present study, interobserver variability was significant, and the interobserver coefficient of variation for measurements determined from standard sagittal plane images obtained during LS junction extension was 23.2% to 44.1%. Comparison of area measurements in different studies should be interpreted with caution, particularly if one is trying to quantify the effects of surgical techniques on foraminal area.

The use of oblique plane CT and MRI to assess the intervertebral foramina in the cervical portion of the vertebral column in humans has been described.15–19 The use of sagittal oblique images may improve confidence in diagnosis of foraminal stenosis16,18 and reveal pathologic changes that are not visible in images obtained in standard sagittal and axial planes. The use of oblique planes for evaluating the LSIF in humans or dogs has not been previously described, to our knowledge. One of the aims of the present study was to determine which imaging plane revealed the smallest cross-sectional area of the LSIF. We found that a sagittal oblique plane revealed a smaller cross-sectional area (derived by use of the bony margins), compared with areas in any of the standard sagittal plane images. One explanation for this smaller area was reduction in partial volume averaging, although the slices were only 0.5 mm and unlikely to have major volume averaging artifacts. In addition to providing a smaller cross-sectional area, the L7 nerve should appear in true cross section in the sagittal oblique plane images, unlike the oblique cross-sectional view obtained in standard sagittal plane images. The appearance of LSIF in the sagittal oblique images was similar to that in the conventional sagittal plane images. Interpretation of any abnormalities would be easier in sagittal oblique images than it would be in double oblique images; moreover, the shape of the LSIF in double oblique images differed markedly from the shape in parasagittal plane images in some dogs.

The areas of the LSIF at entrance, middle, and exit zones were measured because this had been done in a previous study.21 It has been stated that the entrance zone is an area of particular clinical interest because it is the zone that is most easily accessed via dorsal-approach foraminotomies and closest to the spinal (dorsal horn) ganglion.11 In the present study, we found that the middle and exit zones had much smaller cross-sectional areas, compared with that of the entrance zone; as a consequence, the middle and exit zones may be more likely to be the site of compression. It is likely that the part of the foraminal canal with the smallest cross-sectional area or diameter is likely to be most at risk for development of stenosis because there is less soft tissue surrounding the nerve and small degrees of stenosis may be more likely to result in nerve compression.

Our hypothesis that a double-oblique plane would provide an image in which the LSIF had a smaller cross-sectional area than it did in sagittal images was proven correct, but only for the cranial portion of the foramen. By manipulating (by use of the dorsal plane image as a localizer) the imaging plane so that it is perpendicular to the path of the L7 nerve, the caudal portion of the image slice does not section the caudal boundary of the foramen. This may not be a problem when acquiring images of clinical cases because the L7 nerve runs within a groove in the cranial portion of the foramen, which is visible in the double oblique view. The variation in foraminal angulation prevented identification of a standard imaging plane for obtaining images of the LSIF that would be applicable in any dog.

Results of the present study were in agreement with findings of a previous study11 in that the LSIF area was smaller when the LS junction was extended, rather than flexed, but that there was significant interobserver variability in area measurement. The LSIF cross-sectional area was smaller in sagittal oblique and double oblique images than it was in standard sagittal images, and use of the former may provide increased sensitivity in the diagnosis of foraminal stenosis and nerve root compression. Further clinical studies are required to determine whether oblique imaging planes improve sensitivity or accuracy in the diagnosis of foraminal stenosis in dogs.

Imaging measurements (eg, renal length or heart size) are often correlated with standard anatomic features (eg, vertebrae) in an attempt to reduce the effect of anatomic conformation and to normalize for body weight. In the present study, there was a mild association between the length of the L7 vertebra and body weight and also foraminal cross-sectional areas. This was not surprising because dogs of larger breeds have longer L7 vertebrae and greater body weight, compared with dogs of small breeds, and would be expected to have relatively larger foramina. This has been shown to be true in humans, and taller people have larger foramina.34 In humans, weight is negatively correlated with foraminal size, possibly because of more severe pathological changes that can result in foraminal stenosis in patients of greater body weight.34 This may not be true in dogs because of differences in posture, compared with humans. We had hoped that the length of the L7 vertebral body would correlate with foraminal area, thereby allowing a ratio to be derived that could enable determination of foraminal stenosis to be made independent of body size. The results of the present study indicated that foraminal area is complex and not solely related to body size or L7 length; the correlation between either of those variables and foraminal area was only moderate. Unfortunately, there was also potential bias in the breeds available for the present study, and Staffordshire Bull Terriers crosses, a German Shepherd Dog, and German Shepherd Dog crosses composed most of the study group. The sample size was small, and further studies are required to examine a correlation between body size and foraminal area.

The foraminal shape in the sagittal images varied markedly among the dog cadavers. This may explain why the individual dog accounted for most of the variation in foraminal area measurements. The differences in shape appeared to be a result of differences in depth of the groove through which the L7 nerve runs. Although the sagittal planes were positioned on the basis of previously described landmarks, the variation in foraminal shape may be attributable to the fact that middle-zone sagittal planes were positioned further laterally in some dogs. This could have been a result of observer error or variations in pedicle thickness and position relative to the groove within the foramen. There are documented breed variations with regard to orientation of the LS facet joints,35 but there are no reports describing breed variation in foraminal conformation, to our knowledge. Simply measuring foraminal areas does not account for shape variation because different foraminal shapes may have the same cross-sectional area. It is possible that foraminal conformation has a role in whether nerve compression develops. Similar variation in foraminal shape has been detected in humans, and it has been suggested that shape differences may affect the likelihood of nerve root compression.36

The present study has several limitations. The number of dog cadavers used was small, and the number of breeds represented in the groups was limited; further studies are required to describe variations in foraminal shape, and the results of our statistical analyses should not be overinterpreted. There was no gold standard for measurement of foraminal area, and the true accuracy of CT measurements of foraminal area still needs to be determined. The results of this cadaver study suggested that there is considerable interobserver variability in LSIF area measurements, and this fact should be taken into account in subsequent investigations. Oblique imaging planes provided images in which the LSIF is visualized at its smallest cross section, which potentially may be of benefit in the diagnosis of LS foraminal stenosis. Additional studies are required to test this hypothesis in clinical cases.

ABBREVIATIONS

CT

Computed tomography

DLSS

Degenerative lumbosacral stenosis

HU

Hounsfield units

LS

Lumbosacral

LSIF

Lumbosacral (L7-S1) intervertebral foramen

MRI

Magnetic resonance imaging

PMMA

Polymethylmethacrylate

a.

Siemens Somatom Volume Zoom, Siemens AG, Munich, Germany.

b.

Syngo, Siemens AG, Munich, Germany.

c.

Excel, Microsoft Corp, Redmond, Wash.

d.

Minitab, version 15, State College, Pa.

e.

STATA, version 10, Statacorp, College Station, Tex.

References

  • 1.

    Ness MG. Degenerative lumbosacral stenosis in the dog: a review of 30 cases. J Small Anim Pract 1994; 35: 185190.

  • 2.

    Suwankong N, Meij BP, Voorhout G, et al. Review and retrospective analysis of degenerative lumbosacral stenosis in 156 dogs treated by dorsal laminectomy. Vet Comp Orthop Traumatol 2008; 21: 285293.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 3.

    Mayhew PD, Kapatkin AS, Wortman JA, et al. Association of cauda equina compression on magnetic resonance images and clinical signs in dogs with degenerative lumbosacral stenosis. J Am Anim Hosp Assoc 2002; 38: 555562.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 4.

    Sharp NJH, Wheeler S. Lumbosacral disease. In: Sharp NJH, Wheeler S, eds. Small animal spinal disorders: diagnosis and surgery. 2nd ed. Rome: Elsevier Mosby, 2005; 181209.

    • Search Google Scholar
    • Export Citation
  • 5.

    Sjostrom L. Lumbosacral disorders. In: Slatter D, ed. Textbook of small animal surgery. 3rd ed. Philadelphia: Saunders, 2003; 12271237.

    • Search Google Scholar
    • Export Citation
  • 6.

    Kim NH, Yang IH. A study of motor and sensory evoked potentials in chronic cauda equina compression of the dog. Eur Spine J 1996; 5: 338344.

  • 7.

    Delamarter RB, Bohlman HH, Dodge LD, et al. Experimental lumbar spinal stenosis. Analysis of the cortical evoked potentials, microvasculature, and histopathology. J Bone Joint Surg Am 1990; 72: 110120.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8.

    Bodner DR, Delamarter RB, Bohlman HH, et al. Urologic changes after cauda equina compression in dogs. J Urol 1990; 143: 186190.

  • 9.

    Jones JC, Banfield CM, Ward DL. Association between postoperative outcome and results of magnetic resonance imaging and computed tomography in working dogs with degenerative lumbosacral stenosis. J Am Vet Med Assoc 2000; 216: 17691774.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 10.

    Jones JC, Inzana KD. Subclinical CT abnormalities in the lumbosacral spine of older large-breed dogs. Vet Radiol Ultrasound 2000; 41: 1926.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11.

    Jones JC, Davies SE, Werre SR, et al. Effects of body position and clinical signs on L7-S1 intervertebral foraminal area and lumbosacral angle in dogs with lumbosacral disease as measured via computed tomography. Am J Vet Res 2008; 69: 14461454.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12.

    Steffen F, Hunold K, Scharf G, et al. A follow-up study of neurologic and radiographic findings in working German Shepherd Dogs with and without degenerative lumbosacral stenosis. J Am Vet Med Assoc 2007; 231: 15291533.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 13.

    Jones JC, Wilson ME, Bartels JE. A review of high resolution computed tomography and a proposed technique for regional examination of the canine lumbosacral spine. Vet Radiol Ultrasound 1994; 35: 339346.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 14.

    Volle E, Kern A, Stoltenburg G, et al. CT reconstruction technique in lumbar intraneuroforaminal disc herniation. Neuroradiology 1988; 30: 138144.

  • 15.

    Humphreys SC, An HS, Eck JC, et al. Oblique MRI as a useful adjunct in evaluation of cervical foraminal impingement. J Spinal Disord 1998; 11: 295299.

    • Search Google Scholar
    • Export Citation
  • 16.

    Goodman BS, Geffen JF, Mallempati S, et al. MRI images at a 45-degree angle through the cervical neural foramina: a technique for improved visualization. Pain Physician 2006; 9: 327332.

    • Search Google Scholar
    • Export Citation
  • 17.

    Modic MT, Masaryk TJ, Ross JS, et al. Cervical radiculopathy: value of oblique MR imaging. Radiology 1987; 163: 227231.

  • 18.

    Roberts CC, McDaniel NT, Krupinski EA, et al. Oblique reformation in cervical spine computed tomography: a new look at an old friend. Spine 2003; 28: 167170.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 19.

    Yenerich DO, Haughton VM. Oblique plane MR imaging of the cervical spine. J Comput Assist Tomogr 1986; 10: 823826.

  • 20.

    Evans HE, deLahunta A. Miller's guide to the dissection of the dog. 4th ed. Philadelphia: WB Saunders Co, 1996.

  • 21.

    Wood BC, Lanz OI, Jones JC, et al. Endoscopic-assisted lumbosacral foraminotomy in the dog. Vet Surg 2004; 33: 221231.

  • 22.

    Inufusa A, An HS, Lim TH, et al. Anatomic changes of the spinal canal and intervertebral foramen associated with flexion-extension movement. Spine 1996; 21: 24122420.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 23.

    Henninger W, Werner G. CT of the canine lumbosacral spine in extension—flexion rotation. Part I: bony window. Wien Tierarztl Monatsschr 2002; 89: 142151.

    • Search Google Scholar
    • Export Citation
  • 24.

    Jones JC, Wright JC, Bartels JE. Computed tomographic morphometry of the lumbosacral spine of dogs. Am J Vet Res 1995; 56: 11251132.

  • 25.

    Drees R, Dennison SE, Keuler NS, et al. Computed tomographic imaging protocol for the canine cervical and lumbar spine. Vet Radiol Ultrasound 2009; 50: 7479.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 26.

    Auriemma E, Voorhout G, Barthez PY. Determination of optimal window width and level for measurement of the canine pituitary gland height on computed tomographic images using a phantom. Vet Radiol Ultrasound 2007; 48: 113117.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 27.

    Baxter BS, Sorenson JA. Factors affecting the measurement of size and CT number in computed tomography. Invest Radiol 1981; 16: 337341.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 28.

    Sirvanci M, Bhatia M, Ganiyusufoglu KA, et al. Degenerative lumbar spinal stenosis: correlation with Oswestry Disability Index and MR imaging. Eur Spine J 2008; 17: 679685.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 29.

    Speciale AC, Pietrobon R, Urban CW, et al. Observer variability in assessing lumbar spinal stenosis severity on magnetic resonance imaging and its relation to cross-sectional spinal canal area. Spine 2002; 27: 10821086.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 30.

    Pfirrmann CW, Dora C, Schmid MR, et al. MR image-based grading of lumbar nerve root compromise due to disk herniation: reliability study with surgical correlation. Radiology 2004; 230: 583588.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 31.

    Wildermuth S, Zanetti M, Duewell S, et al. Lumbar spine: quantitative and qualitative assessment of positional (upright flexion and extension) MR imaging and myelography. Radiology 1998; 207: 391398.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 32.

    Weishaupt D, Schmid MR, Zanetti M, et al. Positional MR imaging of the lumbar spine: does it demonstrate nerve root compromise not visible at conventional MR imaging? Radiology 2000; 215: 247253.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 33.

    Attias N, Hayman A, Hipp JA, et al. Assessment of magnetic resonance imaging in the diagnosis of lumbar spine foraminal stenosis—a surgeon's perspective. J Spinal Disord Tech 2006; 19: 249256.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 34.

    Cramer GD, Cantu JA, Dorsett RD, et al. Dimensions of the lumbar intervertebral foramina as determined from the sagittal plane magnetic resonance imaging scans of 95 normal subjects. J Manipulative Physiol Ther 2003; 26: 160170.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 35.

    Seiler GS, Hani H, Busato AR, et al. Facet joint geometry and intervertebral disk degeneration in the L5-S1 region of the vertebral column in German Shepherd Dogs. Am J Vet Res 2002; 63: 8690.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 36.

    Stephens MM, Evans JH, O'Brien JP. Lumbar intervertebral foramens. An in vitro study of their shape in relation to intervertebral disc pathology. Spine 1991; 16: 525529.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Figure 1—

    Reformatted sagittal plane CT image of the LS region of the vertebral column of a dog cadaver to illustrate measurement of the L7-S1 (LS) angle. The LS angle was measured at the intersection of a line drawn on the dorsal surface of the L7 vertebral body and a line drawn on the dorsal surface of the S1 vertebral body (both lines shown in red). This image was obtained during extension of the LS junction (the other lines in this image relate to image processing and are not relevant to the measurement process).

  • Figure 2—

    Photograph of the LSIF sectioned in a sagittal plane at the midpedicle level in a dog cadaver (A) and a sagittal plane CT image though the LSIF at the level of the middle zone in another dog cadaver (B). In panel A, the foramen is viewed from the lateral aspect and barium-impregnated PMMA that has been injected into the foramen is visible (arrowheads). The L7-S1 intervertebral disk (asterisk) and articular process of L7 (arrow) are identifiable. In panel B, the PMMA-filled foramen is evident (star). The injection of the plastic material into the foramen allows delineation of soft tissue boundaries and helps to identify the orientation of the foramen in oblique plane CT reconstructions.

  • Figure 3—

    Transverse CT image of the LSIF in a dog cadaver to illustrate the method used to calculate the transverse plane angle (a). The median plane was identified by use of the spinous process and vertebral body of L7 as landmarks (red line) so that the image plane was parallel to the spinous process. On dorsal plane images (not shown), the median plane was aligned so that it was parallel to the vertebral pedicles and perpendicular to the LS vertebral end plates. The orientation of the foramen was determined with the PMMA (star) as a guide. A straight line was manually fitted through the center of the PMMA within the LSIF (blue line) so that the line was equidistant between the dorsal and ventral boundaries of the foramen and bisected the foramen at its narrowest point. This provides an angle of the foramen relative to the median plane (the transverse plane angle). The syringe used to inject the foramen with PMMA was refilled with PMMA (asterisk) and used as a calibration phantom in the CT image.

  • Figure 4—

    Dorsal plane CT image of the LSIF in a dog cadaver to illustrate the method used to calculate the dorsal plane angle (a). The median plane was identified by placing a line parallel to the vertical pedicles and perpendicular to the LS vertebral end plates. On transverse plane images (not shown), the median plane was placed so that it was parallel with the L6, L7, and sacral spinous processes. The orientation of the foramen was determined with the PMMA (star) as a guide. A straight line was manually fitted through the center of the PMMA within the LSIF (blue line) so that the line was equidistant between the dorsal and ventral boundaries of the foramen and bisected the foramen at its narrowest point. This provides an angle of the foramen relative to the median plane (the dorsal plane angle).

  • Figure 5—

    Transverse CT image of the LSIF in a dog cadaver to illustrate the method used to obtain an image in a sagittal oblique plane (blue line). The image plane was placed perpendicular to the transverse plane angle (red line drawn parallel to the PMMA [star]). The image plane crossed the narrowest part of the foramen. The syringe used to inject the foramen with PMMA was refilled with PMMA (right side of image) and used as a calibration phantom in the CT image.

  • Figure 6—

    Dorsal plane CT image of the LSIF in a dog cadaver to illustrate the method used to obtain an image in a transverse oblique plane (blue line). The image plane was placed perpendicular to the dorsal plane angle (red line drawn parallel to the PMMA [star]). The image plane crossed the narrowest part of the foramen.

  • Figure 7—

    Transverse oblique CT image of the LSIF (in extension) in a dog cadaver to illustrate measurement of the cranial portion of the LSIF, which was performed when the foramen did not appear bounded by bone on all sides in an image because of the alignment of the imaging plane. In this view, the transverse process of the L7 vertebral body (arrowhead), lateral part of the L7 facet (arrow), and the L7-S1 disk (asterisk) are evident; the foramen is filled with PMMA (star). To measure the cranial portion of the foramen, a caudal boundary was created by drawing a line that was vertically aligned with the caudal endplate of L7 (red line and dotted red line, respectively); the area of the foramen cranial to the line was measured. Dashed gray line in this picture is related to image processing and is not relevant to the measurement process.

  • Figure 8—

    Representative reformatted CT images of the LSIF (in flexion) in 2 dog cadavers to illustrate the 2 typical conformations of the middle zone of the foramen viewed in a sagittal plane. The vertebral body and pedicle of L7 and sacral body (at the right of the image) are visible in this image plane; the foramen is shaded black for illustrative purposes. The foramen had a hook-shaped bony margin cranially in some cadavers (A) and an ovoid shape in the other cadavers (B). Compared with the more vertical orientation of the cranial portion of the hook-shaped foramen, the cranial portion of the ovoid foramen has a more horizontal orientation.

  • Figure 9—

    Representative sagittal oblique CT images of the LSIF in a dog cadaver during extension (A) and flexion (B) of the LS junction. Notice that the foraminal shape in flexion is similar to that observed in a standard middle-zone sagittal image. When the LS junction is extended, the L7 facet is not as clearly seen and more of the pedicle and facet of S1 is visible. The area of the bone margin and PMMA of the LSIF is greatly reduced in extension compared with flexion. In these images, the foramen is filled with PMMA (star). Dashed gray line in this picture is related to image processing and is not relevant to the measurement process.

  • Figure 10—

    Representative transverse oblique CT image of the plane of the LSIF in a dog cadaver. In this view, the transverse process of the L7 vertebral body (arrowhead), lateral part of the L7 facet (arrow), and the L7-S1 disk (asterisk) are evident; the foramen is filled with PMMA (star). Notice that the caudal portion of the foramen is not visible in this imaging plane.

  • Figure 11—

    Representative double oblique CT images of the LSIF in a dog cadaver during extension (A) and flexion (B) of the LS junction. When the LS junction is flexed, the shape of the foramen is similar to that observed in a transverse oblique image (the caudal portion of the foramen is not visible). In panel A, the transverse process of L7 (arrowhead), the articular process of L7 (arrow), and the L7-S1 intervertebral disk (asterisk) are evident. In panel B, the articular process of L7 (small arrow), the lateral aspect of the articular process of S1 (large arrow), the L7-S1 intervertebral disk (asterisk), and the transverse process of L7 (arrowhead) are evident. In these images, the foramen is filled with PMMA (star).

  • 1.

    Ness MG. Degenerative lumbosacral stenosis in the dog: a review of 30 cases. J Small Anim Pract 1994; 35: 185190.

  • 2.

    Suwankong N, Meij BP, Voorhout G, et al. Review and retrospective analysis of degenerative lumbosacral stenosis in 156 dogs treated by dorsal laminectomy. Vet Comp Orthop Traumatol 2008; 21: 285293.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 3.

    Mayhew PD, Kapatkin AS, Wortman JA, et al. Association of cauda equina compression on magnetic resonance images and clinical signs in dogs with degenerative lumbosacral stenosis. J Am Anim Hosp Assoc 2002; 38: 555562.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 4.

    Sharp NJH, Wheeler S. Lumbosacral disease. In: Sharp NJH, Wheeler S, eds. Small animal spinal disorders: diagnosis and surgery. 2nd ed. Rome: Elsevier Mosby, 2005; 181209.

    • Search Google Scholar
    • Export Citation
  • 5.

    Sjostrom L. Lumbosacral disorders. In: Slatter D, ed. Textbook of small animal surgery. 3rd ed. Philadelphia: Saunders, 2003; 12271237.

    • Search Google Scholar
    • Export Citation
  • 6.

    Kim NH, Yang IH. A study of motor and sensory evoked potentials in chronic cauda equina compression of the dog. Eur Spine J 1996; 5: 338344.

  • 7.

    Delamarter RB, Bohlman HH, Dodge LD, et al. Experimental lumbar spinal stenosis. Analysis of the cortical evoked potentials, microvasculature, and histopathology. J Bone Joint Surg Am 1990; 72: 110120.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8.

    Bodner DR, Delamarter RB, Bohlman HH, et al. Urologic changes after cauda equina compression in dogs. J Urol 1990; 143: 186190.

  • 9.

    Jones JC, Banfield CM, Ward DL. Association between postoperative outcome and results of magnetic resonance imaging and computed tomography in working dogs with degenerative lumbosacral stenosis. J Am Vet Med Assoc 2000; 216: 17691774.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 10.

    Jones JC, Inzana KD. Subclinical CT abnormalities in the lumbosacral spine of older large-breed dogs. Vet Radiol Ultrasound 2000; 41: 1926.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11.

    Jones JC, Davies SE, Werre SR, et al. Effects of body position and clinical signs on L7-S1 intervertebral foraminal area and lumbosacral angle in dogs with lumbosacral disease as measured via computed tomography. Am J Vet Res 2008; 69: 14461454.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12.

    Steffen F, Hunold K, Scharf G, et al. A follow-up study of neurologic and radiographic findings in working German Shepherd Dogs with and without degenerative lumbosacral stenosis. J Am Vet Med Assoc 2007; 231: 15291533.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 13.

    Jones JC, Wilson ME, Bartels JE. A review of high resolution computed tomography and a proposed technique for regional examination of the canine lumbosacral spine. Vet Radiol Ultrasound 1994; 35: 339346.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 14.

    Volle E, Kern A, Stoltenburg G, et al. CT reconstruction technique in lumbar intraneuroforaminal disc herniation. Neuroradiology 1988; 30: 138144.

  • 15.

    Humphreys SC, An HS, Eck JC, et al. Oblique MRI as a useful adjunct in evaluation of cervical foraminal impingement. J Spinal Disord 1998; 11: 295299.

    • Search Google Scholar
    • Export Citation
  • 16.

    Goodman BS, Geffen JF, Mallempati S, et al. MRI images at a 45-degree angle through the cervical neural foramina: a technique for improved visualization. Pain Physician 2006; 9: 327332.

    • Search Google Scholar
    • Export Citation
  • 17.

    Modic MT, Masaryk TJ, Ross JS, et al. Cervical radiculopathy: value of oblique MR imaging. Radiology 1987; 163: 227231.

  • 18.

    Roberts CC, McDaniel NT, Krupinski EA, et al. Oblique reformation in cervical spine computed tomography: a new look at an old friend. Spine 2003; 28: 167170.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 19.

    Yenerich DO, Haughton VM. Oblique plane MR imaging of the cervical spine. J Comput Assist Tomogr 1986; 10: 823826.

  • 20.

    Evans HE, deLahunta A. Miller's guide to the dissection of the dog. 4th ed. Philadelphia: WB Saunders Co, 1996.

  • 21.

    Wood BC, Lanz OI, Jones JC, et al. Endoscopic-assisted lumbosacral foraminotomy in the dog. Vet Surg 2004; 33: 221231.

  • 22.

    Inufusa A, An HS, Lim TH, et al. Anatomic changes of the spinal canal and intervertebral foramen associated with flexion-extension movement. Spine 1996; 21: 24122420.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 23.

    Henninger W, Werner G. CT of the canine lumbosacral spine in extension—flexion rotation. Part I: bony window. Wien Tierarztl Monatsschr 2002; 89: 142151.

    • Search Google Scholar
    • Export Citation
  • 24.

    Jones JC, Wright JC, Bartels JE. Computed tomographic morphometry of the lumbosacral spine of dogs. Am J Vet Res 1995; 56: 11251132.

  • 25.

    Drees R, Dennison SE, Keuler NS, et al. Computed tomographic imaging protocol for the canine cervical and lumbar spine. Vet Radiol Ultrasound 2009; 50: 7479.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 26.

    Auriemma E, Voorhout G, Barthez PY. Determination of optimal window width and level for measurement of the canine pituitary gland height on computed tomographic images using a phantom. Vet Radiol Ultrasound 2007; 48: 113117.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 27.

    Baxter BS, Sorenson JA. Factors affecting the measurement of size and CT number in computed tomography. Invest Radiol 1981; 16: 337341.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 28.

    Sirvanci M, Bhatia M, Ganiyusufoglu KA, et al. Degenerative lumbar spinal stenosis: correlation with Oswestry Disability Index and MR imaging. Eur Spine J 2008; 17: 679685.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 29.

    Speciale AC, Pietrobon R, Urban CW, et al. Observer variability in assessing lumbar spinal stenosis severity on magnetic resonance imaging and its relation to cross-sectional spinal canal area. Spine 2002; 27: 10821086.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 30.

    Pfirrmann CW, Dora C, Schmid MR, et al. MR image-based grading of lumbar nerve root compromise due to disk herniation: reliability study with surgical correlation. Radiology 2004; 230: 583588.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 31.

    Wildermuth S, Zanetti M, Duewell S, et al. Lumbar spine: quantitative and qualitative assessment of positional (upright flexion and extension) MR imaging and myelography. Radiology 1998; 207: 391398.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 32.

    Weishaupt D, Schmid MR, Zanetti M, et al. Positional MR imaging of the lumbar spine: does it demonstrate nerve root compromise not visible at conventional MR imaging? Radiology 2000; 215: 247253.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 33.

    Attias N, Hayman A, Hipp JA, et al. Assessment of magnetic resonance imaging in the diagnosis of lumbar spine foraminal stenosis—a surgeon's perspective. J Spinal Disord Tech 2006; 19: 249256.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 34.

    Cramer GD, Cantu JA, Dorsett RD, et al. Dimensions of the lumbar intervertebral foramina as determined from the sagittal plane magnetic resonance imaging scans of 95 normal subjects. J Manipulative Physiol Ther 2003; 26: 160170.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 35.

    Seiler GS, Hani H, Busato AR, et al. Facet joint geometry and intervertebral disk degeneration in the L5-S1 region of the vertebral column in German Shepherd Dogs. Am J Vet Res 2002; 63: 8690.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 36.

    Stephens MM, Evans JH, O'Brien JP. Lumbar intervertebral foramens. An in vitro study of their shape in relation to intervertebral disc pathology. Spine 1991; 16: 525529.

    • Crossref
    • Search Google Scholar
    • Export Citation

Advertisement