Evaluation of diffusion-weighted magnetic resonance imaging at 3.0 Tesla for differentiation between intracranial neoplastic and noninfectious inflammatory lesions in dogs

Megan J. Maclellan 1Department of Clinical Sciences, College of Veterinary Medicine, University of Minnesota, Saint Paul, MN 55108.

Search for other papers by Megan J. Maclellan in
Current site
Google Scholar
PubMed
Close
 DVM
,
Christopher P. Ober 1Department of Clinical Sciences, College of Veterinary Medicine, University of Minnesota, Saint Paul, MN 55108.

Search for other papers by Christopher P. Ober in
Current site
Google Scholar
PubMed
Close
 DVM, PhD
,
Daniel A. Feeney 1Department of Clinical Sciences, College of Veterinary Medicine, University of Minnesota, Saint Paul, MN 55108.

Search for other papers by Daniel A. Feeney in
Current site
Google Scholar
PubMed
Close
 DVM, MS
, and
Carl R. Jessen 1Department of Clinical Sciences, College of Veterinary Medicine, University of Minnesota, Saint Paul, MN 55108.

Search for other papers by Carl R. Jessen in
Current site
Google Scholar
PubMed
Close
 DVM, PhD

Abstract

OBJECTIVE

To evaluate the utility of apparent diffusion coefficient (ADC) and fractional anisotropy (FA) values obtained by diffusion-weighted MRI (DWI) at 3.0 T for differentiating intracranial neoplastic lesions from noninfectious inflammatory lesions (NIILs) in dogs.

ANIMALS

54 dogs that met inclusion criteria (ie, had a histologically confirmed intracranial lesion and DWI of the brain performed) with 5 lesion types: meningioma [n = 18], glioma [14], metastatic hemangiosarcoma [3], other metastatic neoplasms [5], and NIIL [14]).

PROCEDURES

Two observers, who were blinded to the histologic diagnoses, independently determined the mean ADC and FA values for each evaluated intracranial lesion on the basis of 3 circular regions of interest on DWI images. Findings were compared among the 5 lesion types, between all neoplasms combined and NIILs, and between the 5 legion types and previously determined values for corresponding locations for neurologically normal dogs.

RESULTS

The mean ADC and FA values did not differ significantly among the 5 lesion types or between all neoplasms combined and NIILs. However, 35% (14/40) of the neoplastic lesions had an ADC value ≥ 1.443 × 10−3 mm2/s, whereas all NIILs had ADC values < 1.443 × 10−3 mm2/s. Meningiomas and NIILs had FA values that were significantly lower than those for neurologically normal dogs.

CONCLUSIONS AND CLINICAL RELEVANCE

In this population of dogs, the FA values for meningiomas and NIILs differed significantly from those previously reported for neurologically normal dogs. In addition, an ADC cutoff value of 1.443 × 10−3 mm2/s appeared to be highly specific for diagnosing neoplastic lesions (vs NIILs), although the sensitivity and accuracy were low.

Abstract

OBJECTIVE

To evaluate the utility of apparent diffusion coefficient (ADC) and fractional anisotropy (FA) values obtained by diffusion-weighted MRI (DWI) at 3.0 T for differentiating intracranial neoplastic lesions from noninfectious inflammatory lesions (NIILs) in dogs.

ANIMALS

54 dogs that met inclusion criteria (ie, had a histologically confirmed intracranial lesion and DWI of the brain performed) with 5 lesion types: meningioma [n = 18], glioma [14], metastatic hemangiosarcoma [3], other metastatic neoplasms [5], and NIIL [14]).

PROCEDURES

Two observers, who were blinded to the histologic diagnoses, independently determined the mean ADC and FA values for each evaluated intracranial lesion on the basis of 3 circular regions of interest on DWI images. Findings were compared among the 5 lesion types, between all neoplasms combined and NIILs, and between the 5 legion types and previously determined values for corresponding locations for neurologically normal dogs.

RESULTS

The mean ADC and FA values did not differ significantly among the 5 lesion types or between all neoplasms combined and NIILs. However, 35% (14/40) of the neoplastic lesions had an ADC value ≥ 1.443 × 10−3 mm2/s, whereas all NIILs had ADC values < 1.443 × 10−3 mm2/s. Meningiomas and NIILs had FA values that were significantly lower than those for neurologically normal dogs.

CONCLUSIONS AND CLINICAL RELEVANCE

In this population of dogs, the FA values for meningiomas and NIILs differed significantly from those previously reported for neurologically normal dogs. In addition, an ADC cutoff value of 1.443 × 10−3 mm2/s appeared to be highly specific for diagnosing neoplastic lesions (vs NIILs), although the sensitivity and accuracy were low.

Magnetic resonance imaging is a vital component of the diagnosis, therapeutic planning, and monitoring of intracranial lesions in both human and veterinary medicine.1 Biopsy is the gold standard for brain tumor diagnosis; however, brain biopsy carries a considerable risk of morbidity and death as well as incorrect diagnosis with insufficient sampling.2 In dogs, intracranial neoplastic and nonneoplastic lesions were previously differentiated solely on the basis of evaluation of multiple conventional MRI findings (eg, shape, location, and number of lesions; involvement of dura matter; evidence of a mass effect; and contrast medium enhancement).3 Advances in MRI, such as DWI, have improved the utility of MRI findings for characterizing intracranial lesions, rather than simply evaluating structural abnormalities and their related complications.1,4 These advances have expanded the morphology-based imaging capabilities of MRI to include functional, cellular, metabolic, cytoarchitectural, and hemodynamic information, which aids in the diagnosis, grading, therapeutic planning, and monitoring of brain tumors.1

Diffusion imaging is a physiology-based type of MRI that uses water movement through tissues as an inherent contrast medium.5,6 When unrestricted, water molecules move (diffuse) equally in all directions with Brownian motion.7,8 In the neurologically normal brain, the diffusion of intracellular water is more limited than that of extracellular water because of restriction by cell membranes.1,4 This water movement is usually assessed with DWI and by ADC measurement. In addition, the directionality of water diffusion can be evaluated for structures with a highly directional cytoarchitecture (eg, axons) by FA measurement.6,8-10

In human medicine, the utility of brain diffusion imaging has been demonstrated for the diagnosis, therapeutic guidance, and monitoring of ischemic infarcts, multiple sclerosis and other neurodegenerative diseases, infectious conditions (diffuse disease or abscesses), neoplasia, and metabolic lesions.2,5 Apparent diffusion coefficient maps are useful in humans and dogs for assessing the cellularity of a lesion to predict tumor grade and distinguishing between benign and malignant meningiomas; such assessments typically include comparisons with values for neurologically normal brain tissue, resulting in higher positive predictive values for predicting tumor grade.1,2,8,11,12 Measurement of intracranial ADC values is also used to monitor for postoperative injuries and to distinguish between tumor recurrence and surgical injury at follow-up.1

Fractional anisotropy maps are used for surgical planning and monitoring of intracranial lesion progression (eg, distinguishing between white-matter tracts that have been destroyed or invaded vs those that are intact but simply displaced or distorted as the result of the lesion).1 Although ADC values are available from diffusion imaging studies5 in human medicine, the few studies7,13-15 of dogs with intracranial lesions have mostly involved few cases, small subsets of cases, or nonconfirmed cases and included no comparisons with neurologically normal dogs. Given the utility of DWI in human medicine, the authors believe that DWI has the potential to be a beneficial, noninvasive diagnostic tool in veterinary patients and could also be used to study brain lesions in dogs in an effort to better understand similar lesions in humans.

The objective of the retrospective study reported here was to measure ADC and FA values via DWI at 3.0 T in dogs with histologically confirmed intracranial neoplastic or noninfectious inflammatory lesions. Our hypothesis was that these values would differ significantly from previously determined values for neurologically normal dogs.16 We also hypothesized that ADC and FA values could be used to differentiate between neoplastic and noninfectious inflammatory lesions and among certain types of neoplasms and noninfectious inflammatory lesions.

Materials and Methods

Case selection

The Picture Archive and Communication system at the University of Minnesota Veterinary Medical Center was searched to identify dogs that underwent brain MRI including DWI and DTI in the dorsal plane between January 1, 2008, and December 31, 2015, and that had a histologically confirmed intracranial lesion. For each dog included in the study, data on reproductive status, age, body weight, breed, and number, type, and location (affected brain lobe or site) of intracranial lesions were obtained from the medical record. Type of intracranial lesion was determined on the basis of the histologic diagnosis.

Procedures

Because all dogs were clinical patients, the anesthetic protocol, area to be imaged, image sequences (eg, T2- and T1-weighted, fluid-attenuated inversion recovery, fat-saturated, and postcontrast), image plane (eg, dorsal, axial, or sagittal), and duration of imaging session were determined on the basis of each dog's status by the attending radiologist, clinician, and anesthesiologist.

Magnetic resonance imaging was performed with a 3.0-T MRI scannera and knee coil.b All diffusion images were obtained with the following settings: mean repetition time, 10,000 milliseconds; echo time, 90 milliseconds; slice thickness, 3 mm; and slice gap, 0.3 mm. The DTI sequences were obtained by use of 25 imaging directions with a matrix size of 160 × 160 and 1 excitation. The DWI sequences were obtained by use of 3 imaging directions with a matrix size of 128 × 128 and 2 excitations. A diffusion sensitivity value (b value) of 1,000 s/mm2 was used. The ADC and FA map values were calculated from the DTI sequence. The ADC values were expressed in square millimeters per second, and FA was expressed as a scalar value that ranged from 0 to 1.

Intracranial lesions were typically first identified on the conventional MRI images, including the T2-and T1-weighted pre- and postcontrast images, then identified on the diffusion images (ie, the diffusion tensor sequences and the ADC and FA maps). All diffusion measurements, including ADC and FA values, for each lesion were obtained with commercial softwarec that allowed for simultaneous evaluation of all diffusion images for that lesion.

Areas for evaluation were selected on the basis of all available image sequences to avoid potential measurement errors (eg, partial volume averaging or inclusion of cystic regions or necrotic centers). For each dog, ADC and FA measurements were calculated for 1 lesion. When a dog had multiple lesions, 1 lesion was selected on the basis of size, appearance, and regional parenchymal change to decrease the risk of including regions with edema, necrosis, or normal parenchyma in the calculation of ADC and FA measurements. For each lesion, the mean ADC and FA values were calculated on the basis of 3 circular ROIs. Each ROI was sized and positioned to maximize the amount of lesioned tissue and to minimize the inclusion of nonlesioned tissue (eg, CSF in regional sulci) within the region. The area of the ROIs was dependent on lesion size (range, 5 to 30 mm2). For each dog, the ADC and FA measurements were made independently by 2 observers (MJM and CPO) who were blinded to the histologic diagnoses. Standardization of slices for each ROI was not performed, and choice of slice and area within the lesion for ADC and FA measurements was left to the observer's discretion (and not random selection). All sequences were used to determine the areas to be measured by each observer because this provided a more realistic clinical application. The location (ie, lobe or site) of each measured lesion was recorded to allow comparison of ADC and FA measurements with those from corresponding locations in neurologically normal dogs.

Statistical analysis

Because a previous study16 revealed no significant differences in ADC and FA values between the right and left hemispheres, brain hemisphere of the lesions was not taken into consideration for statistical analysis. Descriptive statistics (ie, mean, median, SD, and quartiles) were calculated for ADC and FA values by the type of neoplasm, for all neoplasms combined, and for noninfectious inflammatory lesions. The normality of the ADC and FA values was assessed by means of the Shapiro-Wilk test. Because the data were not normally distributed, nonparametric analysis methods were used. The Kruskal-Wallis test was used to compare median ADC and FA values among specific types of lesions; the Mann-Whitney U test was used to compare the values for all neoplasms combined with those for noninfectious inflammatory lesions. The Wilcoxon signed rank test was used to compare the ADC and FA values for each dog with a given lesion type (except dogs with intraventricular lesions) with previously reported median ADC and FA values for the corresponding lobe or site in neurologically normal dogs.16 The ADC and FA values for intraventricular lesions in the dogs of the present study were compared with previously determined, but unreported, values for the corresponding site in neurologically normal dogs (median intraventricular ADC and FA values in neurologically normal dogs, 2.889 × 10−3 mm2/s and 0.136 × 10−3 mm2/s, respectively). Because different areas of the brain have dissimilar diffusion values,16 when comparing diffusion values between dogs with intracranial lesions and those that are neurologically normal, it is important to compare values from the corresponding lobe or site (eg, diffusion values for a lesion in the frontal lobe should only be compared with previously established values in the corresponding lobe).

A potential ADC cutoff value for differentiating between neoplastic lesions (all neoplasms combined) and noninfectious inflammatory lesions was identified by examination of ADC values for all lesions combined by quartile. An ROC curve was then used to evaluate the accuracy of this cutoff for differentiating between neoplastic and noninfectious inflammatory lesions.

All statistical analyses were performed by use of commercially available statistical software.d-f Two-tailed tests were used for all analyses, and values of P < 0.05 were considered significant.

Results

Sixty-two dogs were identified that met the study inclusion criteria; 8 of these were excluded because they had lesions that were represented by ≤ 2 dogs (choroid plexus tumor [n = 2], infarct [2], and cavernoma, fungal disease, hematoma, and epidermoid cyst [1 each]). The remaining 54 dogs included 28 spayed females, 20 neutered males, 5 sexually intact males, and 1 sexually intact female, with a mean age and body weight of 8.3 years (range, 2.5 to 15 years) and 21.2 kg (46.6 lb; range, 2.5 to 62.1 kg [5.5 to 136.6 lb]), respectively. The 3 most common breeds were Boxer (n = 7), Labrador Retriever (7), and German Shepherd Dog (5). The overall median number of intracranial lesions per patient was 1 (range, 1 to 25). These patient characteristics were also summarized by type of lesion (Table 1).

Table 1—

Characteristics of 54 dogs with various types of histologically confirmed intracranial lesions that underwent DWI from 2008 through 2015.

CharacteristicMeningioma (n = 18)Glioma (n = 14)Noninfectious inflammation (n = 14)Metastatic neoplasia (n = 5)Hemangiosarcoma (n = 3)
Reproductive status (No. of dogs)     
 Sexually intact female10000
 Spayed female126622
 Sexually intact male12110
 Neutered male46721
Mean (range) age (y)10 (2–15)9.4 (5–13)6.7 (3–11.8)7.9 (2.5–11)9.3 (8–11)
Mean (range) body weight (kg)26.7 (9.9–37.5)21.4 (4.2–40)15.5 (3.5–62.1)20.4 (2.5–45)28.3 (13.6–45.2)
Median (range) No. of lesions per dog1 (1–1)1 (1–2)3 (1–7)1 (1–4)12 (11–25)
Breed (No. of dogs)     
 Beagle00011
 Boxer34000
 Dachshund00200
 French Bulldog01110
 German Shepherd Dog21101
 Golden Retriever11110
 Labrador Retriever61000
 Rottweiler00110
 Yorkshire Terrier00110
 Other6*6701§

Australian Cattle Dog, Field Spaniel, Rat Terrier, Siberian Husky, Standard Schnauzer, and Weimaraner.

Bichon Frise, Boston Terrier, unspecified Corgi breed, English Springer Spaniel, Scottish Terrier, and Standard Poodle.

Chihuahua, unspecified Coonhound breed, Miniature Pinscher, Norwich Terrier, Pug, Pekingese, and Toy Poodle.

Bullmastiff.

Types of lesions in the 54 dogs included meningioma (n = 18), glioma (14), noninfectious inflammatory (ie, granulomatous, lymphoplasmacytic, and histiocytic; 14) lesion, metastatic hemangiosarcoma (3), and other metastatic neoplasia (ie, carcinoma, histiocytic sarcoma, round cell tumor, anaplastic sarcoma, and primitive neuroectodermal tumor; 5). Because hemangiosarcomas had a markedly different (ie, more hypointense) appearance than other metastatic neoplasias on conventional MRI and on ADC and FA maps, this type of lesion was evaluated separately. The distribution of the various types of lesions within the brain was summarized (Table 2).

Table 2—

Number of dogs in Table 1 with various types of intracranial neoplastic and noninfectious inflammatory lesions by brain lobe or site.

Type of lesionFrontal (n = 25)Temporal (n = 8)Parietal (n = 7)Other site* (n = 14)
Meningioma13122
Glioma8510
Metastatic hemangiosarcoma0030
Other metastatic neoplasias1112
Noninfectious inflammation31010

Includes lesions in thalamus (5 noninfectious inflammatory lesions), pons (1 meningioma and 3 noninfectious inflammatory lesions), ventricle (2 other metastatic neoplasia [intraventricular]), occipital (1 meningioma and 1 noninfectious inflammatory lesion), and cerebellum (1 noninfectious inflammatory lesion).

Representative MRI images of a meningioma (Figure 1), noninfectious inflammatory lesion (Figure 2), and metastatic hemangiosarcoma (Figure 3) were obtained. When compared with the corresponding brain lobe or site, the FA values for all neoplasms combined and for noninfectious inflammatory lesions were significantly (P < 0.001 and P = 0.01, respectively) lower than FA values of neurologically normal dogs; only 2 of the 5 types of evaluated lesions, meningioma and noninfectious inflammatory lesions, had FA values that were significantly (P = 0.001 and P = 0.009, respectively) lower than FA values for neurologically normal dogs (Table 3). When compared with the corresponding brain lobe or site, the ADC values for all neoplasms combined and for noninfectious inflammatory lesions did not differ significantly (P = 0.15 and P = 0.11, respectively) from the ADC values for neurologically normal dogs; in addition, no significant differences were found between the ADC values for any of the 5 types of lesions and the ADC values for neurologically normal dogs.

Figure 1—
Figure 1—

Dorsal plane T2-weighted (A), postcontrast T1-weighted (B), and diffusion-weighted (C) MRI images and ADC (D) and FA (E) maps of a meningioma (arrows) associated with the left side of the brainstem in a 4-year-old spayed female Boxer, resulting in rightward displacement of the brainstem. The mass is markedly hyperintense to the brainstem in panel C and is markedly and homogeneously contrast enhancing. Note the markedly lower signal intensity of the mass relative to the brainstem in panel E.

Citation: Journal of the American Veterinary Medical Association 255, 1; 10.2460/javma.255.1.71

Figure 2—
Figure 2—

Dorsal plane T2-weighted (A), postcontrast T1-weighted (B), and diffusion-weighted (C) MRI images and ADC (D) and FA (E) maps of a focal inflammatory lesion (arrows) in the left occipital lobe of an 8-year-old spayed female Pug. The lesion is mildly hyperintense on the T2-weighted image and heterogeneously contrast enhancing. There is a mild mass effect associated with the lesion. Note that in panel C, the mass is slightly less distinguishable from the surrounding tissue than the meningioma in Figure 1.

Citation: Journal of the American Veterinary Medical Association 255, 1; 10.2460/javma.255.1.71

Figure 3—
Figure 3—

Dorsal plane T2-weighted (A), postcontrast T1-weighted (B), and diffusion-weighted (C) MRI images and ADC (D) and FA (E) maps of a hemangiosarcoma (arrows) in the left temporal lobe of a 9-year-old spayed female Bullmastiff. There is medial extension of the mass into the left lateral ventricle. Note the distortion of anatomic features and marked differences in signal intensity in panels C, D, and E due to the presence of blood.

Citation: Journal of the American Veterinary Medical Association 255, 1; 10.2460/javma.255.1.71

Table 3—

Summary data and comparison of median ADC values (× 10−3 mm2/s) and FA values* obtained by DWI for the dogs in Table 1 with previously determined14 median values from neurologically normal dogs.

 Dogs with intracranial lesionsNeurologically normal dogs
Variable, by type of lesionMean ± SDRangeIQRMedianMedianP value
ADC value (× 10−3 mm2/s)      
 Meningioma1.229 ± 0.7890.565–3.2900.696–1.6450.9390.8490.18
 Glioma1.212 ± 0.6490.381–2.5030.727–1.7480.8780.8490.20
 Metastatic hemangiosarcoma0.872 ± 0.6540.183–1.4830.182–1.4830.9510.8331.00
 Other metastatic neoplasia1.130 ± 0.4360.610–1.6400.741–1.5781.0120.8610.35
 Noninfectious inflammation0.922 ± 0.2460.502–1.3620.734–1.1640.8790.7800.11
 All neoplasms combined1.184 ± 0.6790.183–3.2900.722–1.6040.9330.8490.15
FA value      
 Meningioma0.250 ± 0.0940.083–0.4200.171–0.3140.2780.3500.001
 Glioma0.241 ± 0.1550.071–0.5580.138–0.3340.1780.3500.08
 Metastatic hemangiosarcoma0.402 ± 0.2370.237–0.6740.237–0.6740.2950.3791.00
 Other metastatic neoplasias0.179 ± 0.0810.095–0.2910.108–0.2620.1560.2940.31
 Noninfectious inflammation0.219 ± 0.1270.129–0.5860.139–0.2650.1720.3500.009
 All neoplasms combined0.250 ± 0.1340.071–0.6740.140–0.3090.2400.350< 0.001

Data represent values for all lobes or sites of the brain combined.

The Wilcoxon signed rank test was used for these comparisons: the value for each dog with a given type of lesion was compared with the median value for the same brain lobe or site in neurologically normal dogs.

Scalar value (range, 0–1).

IQR = Interquartile (25th to 75th percentile) range.

The median ADC and FA values did not differ significantly among the individual diagnoses (P = 0.93 and P = 0.30, respectively) or between all neoplasms combined and noninfectious inflammatory lesions (P = 0.49 and P = 0.47, respectively). The fourth quartile of ADC values for all lesions combined (≥ 1.443 × 10−3 mm2/s) included only neoplastic lesions, representing 35% (14/40) of the neoplasms evaluated (Table 4). The neoplastic lesions with ADC values ≥ 1.443 × 10−3 mm2/s included meningiomas (5/18), gliomas (6/14), metastatic neoplasms (2/5), and metastatic hemangiosarcoma (1/3).

Table 4—

Number of dogs in Table 1 with neoplastic or noninfectious inflammatory lesions, grouped by quartile of overall* ADC values.

QuartileRange in ADC values (× 10−3 mm2/s)Neoplasia (n = 40)Noninfectious inflammation (n = 14)
1st0.182–0.714103
2nd0.747–0.90086
3rd0.909–1.36285
4th1.443–3.290140

For all (both neoplastic and noninfectious inflammatory) lesions combined.

The ROC analysis indicated that the use of an ADC cutoff value of 1.443 × 10−3 mm2/s for differentiating neoplastic lesions from noninfectious inflammatory lesions yielded the highest value for sensitivity - (1 - specificity). The sensitivity and specificity of this cutoff value for diagnosing neoplastic (vs noninfectious inflammatory) lesions were 35% (95% CI, 21% to 52%) and 100% (95% CI, 77% to 100%), respectively. However, the area under the ROC curve (56% [95% CI, 42% to 70%]) indicated that the accuracy of this ADC cutoff value for differentiating between neoplastic and noninfectious inflammatory lesions was low.

Discussion

The similar diffusion (ie, ADC and FA) values among the various types of intracranial lesions evaluated in the dogs of the present study were presumably due to similar disruptions of normal brain architecture among these lesions. However, although assessment of ADC and FA values alone appeared to be insufficient for differentiating among specific types of intracranial lesions in dogs, an ADC value ≥ 1.443 × 10−3 mm2/s may be suggestive of the presence of neoplastic rather than noninfectious inflammatory lesions. In the present study, all 14 dogs (26% of all dogs with intracranial lesions) with an ADC value ≥ 1.443 × 10−3 mm2/s had neoplastic lesions.

The results of the present study indicated that FA values may be more sensitive than ADC values for identification of pathological changes in the brains of dogs. When compared with the corresponding brain lobe or site, none of the ADC values for the individual types of intracranial lesions differed significantly from previously determined16 values for neurologically normal dogs. However, the FA values for dogs with meningioma and noninfectious inflammatory lesions were significantly lower than the FA values from the corresponding brain lobe or site of neurologically normal dogs. Although not significantly different, the FA values for the other types of intracranial lesions were also lower than those for the neurologically normal dogs. The lower FA values associated with the lesions evaluated in the present study indicated that disruption of normal brain tissue architecture by a surrounding or invading lesion resulted in similar alterations in water diffusion patterns in affected tissue.

The sample size and available histologic detail for the meningioma lesions evaluated in the present study were insufficient to evaluate the utility of diffusion values for distinguishing between benign lesions (eg, noninfectious inflammatory lesion vs benign meningioma) or between malignant neoplasms (eg, meningioma vs glioma). Reports1,8,11,12 in the human medical literature indicate that diffusion values can be used to assess tumor grade (ie, degree of malignancy), especially for meningioma, by comparing these values with those obtained by DWI of corresponding tissue from clinically normal patients.

The metastatic hemangiosarcoma lesions in the dogs of the present study were noted to have a dramatically altered appearance on DWI images, compared with the appearance of brain tissue in the corresponding lobes or sites of neurologically normal dogs.14 Although the FA values for dogs with hemangiosarcoma lesions did not differ significantly from those for the neurologically normal dogs, the sample size was small (only 3 dogs with hemangiosarcoma). Diffusion imaging is highly sensitive to magnetic field inhomogeneities, such as those from paramagnetic substances typically found in hemorrhagic lesions (malignant or benign).17 A study17 of humans shows that such susceptibility artifacts are associated with a decrease in ADC values, resulting in inaccurate and unreliable measurements. Taken together with the results of the present study, it appears that specific diffusion measurements may be less valuable than visual evaluation of diffusion images for identification of intracranial hemorrhagic lesions (eg, hemangiosarcoma).

A limitation to the present study was the lack of extensive histopathologic subtyping and grading for the evaluated lesions. This information, which is more readily available in human medicine than in veterinary medicine, would have allowed for further categorization and possible separation between the tumor types. Another limitation was the small sample size. For the dogs of the present report, comparisons by lobe or site were limited by the small number (or absence) of each lesion type in each lobe or site. Additional studies with larger sample sizes and more extensive information on the grading and classification of tumors might be useful for identification of meaningful differences in diffusion values.

In conclusion, for the dogs with intracranial lesions in the present study, DWI alone was not sufficient for differentiating between neoplastic and noninfectious inflammatory lesions or for distinguishing between noninfectious inflammatory lesions and the various types of neoplastic lesions. However, the ROC analysis indicated that an ADC cutoff value of 1.443 × 10−3 mm2/s was highly specific (ie, no false-positive results) for diagnosing neoplastic lesions (vs noninfectious inflammatory lesions), although the proportion of patients with neoplastic lesions that were identified by use of this cutoff was low.

Acknowledgments

Supported in part by the Small Companion Animal Grant from the University of Minnesota and the American College of Veterinary Radiology Research Grant.

The authors declare that there were no conflicts of interest.

Presented as an oral presentation at the Annual American College of Veterinary Radiology Scientific Conference, Saint Paul, October 2015.

ABBREVIATIONS

ADC

Apparent diffusion coefficient

CI

Confidence interval

DTI

Diffusion tensor imaging

DWI

Diffusion-weighted MRI

FA

Fractional anisotropy

ROC

Receiver operating characteristic

ROI

Region of interest

Footnotes

a.

Sigma HDx 3.0-T MRI scanner, GE Healthcare, Waukesha, Wis.

b.

HD T/R quad extremity, Invivio, Gainesville, Fla.

c.

Func Tool, GE Healthcare, Waukesha, Wis.

d.

SPSS Statistics, version 20, IBM Corp, Armonk, NY.

e.

MedCalc, version 18.0, MedCalc Software, Ostend, Belgium.

f.

JMP Pro, version 13.1.0, SAS Institute Inc, Cary, NC.

References

  • 1. Cha S. Update on brain tumor imaging: from anatomy to physiology. Am J Neuroradiol 2006;27:475487.

  • 2. Kono K, Inoue Y, Nakayama K, et al. The role of diffusion-weighted imaging in patients with brain tumors. Am J Neuroradiol 2001;22:10811088.

    • Search Google Scholar
    • Export Citation
  • 3. Cherubini GB, Mantis P, Martinez TA, et al. Utility of MRI for distinguishing neoplastic from non-neoplastic brain lesions in dogs and cats. Vet Radiol Ultrasound 2005;46:384387.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 4. Hartmann A, Soffler C, Failing K, et al. Diffusion-weighted magnetic resonance imaging of the normal canine brain. Vet Radiol Ultrasound 2014;55:592598.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 5. Sener RN. Diffusion MRI: apparent diffusion coefficient (ADC) values in the normal brain and a classification of brain disorders based on ADC values. Comput Med Imaging Graph 2001;25:299326.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 6. Neil JJ. Diffusion imaging concepts for clinicians. J Magn Reson Imaging 2008;27:17.

  • 7. Sutherland-Smith J, King R, Faissler D, et al. Magnetic resonance imaging apparent diffusion coefficients for histologically confirmed intracranial lesions in dogs. Vet Radiol Ultrasound 2011;52:142148.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8. Price SJ, Tozer DJ, Gillard JH. Methodology of diffusion-weighted, diffusion tensor and magnetisation transfer imaging. Br J Radiol 2011;84:S121S126.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 9. Hecht S, Adams WH. MRI of brain disease in veterinary patients part 1: basic principles and congenital brain disorders. Vet Clin North Am Small Anim Pract 2010;40:2138.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 10. Anaya García MS, Hernández-Anaya JS, Meléndez OM, et al. In vivo study of cerebral white matter in the dog using diffusion tensor tractography. Vet Radiol Ultrasound 2015;56:188195.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11. Nagar VA, Ye JR, Ng WH, et al. Diffusion-weighted MR imaging: diagnosing atypical or malignant meningiomas and detecting tumor dedifferentiation. Am J Neuroradiol 2008;29:11471152.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12. Hakyemez B, Yildirim N, Gokalp G, et al. The contribution of diffusion-weighted MR imaging to distinguishing typical from atypical meningiomas. Neuroradiology 2006;48:513520.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 13. Garosi L, McConnell JF, Platt SR, et al. Clinical and topographic magnetic resonance characteristics of suspected brain infarction in 40 dogs. J Vet Med Intern Med 2006;20:311321.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 14. Cervera V, Mai W, Vite CH, et al. Comparative magnetic resonance imaging findings between gliomas and presumed cerebrovascular accidents in dogs. Vet Radiol Ultrasound 2011;52:3340.

    • Search Google Scholar
    • Export Citation
  • 15. McConnell JF, Garosi L, Platt SR. Magnetic resonance imaging findings of presumed cerebellar cerebrovascular accident in twelve dogs. Vet Radiol Ultrasound 2005;46:110.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 16. MacLellan MJ, Ober CP, Feeney DA, et al. Diffusion-weighted magnetic resonance imaging of the brain of neurologically normal dogs. Am J Vet Res 2017;78:601608.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 17. Kang BK, Dong GN, Jae WR. at al. Diffusion-weighted MR imaging of intracerebral hemorrhage. Korean J Radiol 2001;2:183191.

  • Figure 1—

    Dorsal plane T2-weighted (A), postcontrast T1-weighted (B), and diffusion-weighted (C) MRI images and ADC (D) and FA (E) maps of a meningioma (arrows) associated with the left side of the brainstem in a 4-year-old spayed female Boxer, resulting in rightward displacement of the brainstem. The mass is markedly hyperintense to the brainstem in panel C and is markedly and homogeneously contrast enhancing. Note the markedly lower signal intensity of the mass relative to the brainstem in panel E.

  • Figure 2—

    Dorsal plane T2-weighted (A), postcontrast T1-weighted (B), and diffusion-weighted (C) MRI images and ADC (D) and FA (E) maps of a focal inflammatory lesion (arrows) in the left occipital lobe of an 8-year-old spayed female Pug. The lesion is mildly hyperintense on the T2-weighted image and heterogeneously contrast enhancing. There is a mild mass effect associated with the lesion. Note that in panel C, the mass is slightly less distinguishable from the surrounding tissue than the meningioma in Figure 1.

  • Figure 3—

    Dorsal plane T2-weighted (A), postcontrast T1-weighted (B), and diffusion-weighted (C) MRI images and ADC (D) and FA (E) maps of a hemangiosarcoma (arrows) in the left temporal lobe of a 9-year-old spayed female Bullmastiff. There is medial extension of the mass into the left lateral ventricle. Note the distortion of anatomic features and marked differences in signal intensity in panels C, D, and E due to the presence of blood.

  • 1. Cha S. Update on brain tumor imaging: from anatomy to physiology. Am J Neuroradiol 2006;27:475487.

  • 2. Kono K, Inoue Y, Nakayama K, et al. The role of diffusion-weighted imaging in patients with brain tumors. Am J Neuroradiol 2001;22:10811088.

    • Search Google Scholar
    • Export Citation
  • 3. Cherubini GB, Mantis P, Martinez TA, et al. Utility of MRI for distinguishing neoplastic from non-neoplastic brain lesions in dogs and cats. Vet Radiol Ultrasound 2005;46:384387.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 4. Hartmann A, Soffler C, Failing K, et al. Diffusion-weighted magnetic resonance imaging of the normal canine brain. Vet Radiol Ultrasound 2014;55:592598.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 5. Sener RN. Diffusion MRI: apparent diffusion coefficient (ADC) values in the normal brain and a classification of brain disorders based on ADC values. Comput Med Imaging Graph 2001;25:299326.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 6. Neil JJ. Diffusion imaging concepts for clinicians. J Magn Reson Imaging 2008;27:17.

  • 7. Sutherland-Smith J, King R, Faissler D, et al. Magnetic resonance imaging apparent diffusion coefficients for histologically confirmed intracranial lesions in dogs. Vet Radiol Ultrasound 2011;52:142148.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8. Price SJ, Tozer DJ, Gillard JH. Methodology of diffusion-weighted, diffusion tensor and magnetisation transfer imaging. Br J Radiol 2011;84:S121S126.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 9. Hecht S, Adams WH. MRI of brain disease in veterinary patients part 1: basic principles and congenital brain disorders. Vet Clin North Am Small Anim Pract 2010;40:2138.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 10. Anaya García MS, Hernández-Anaya JS, Meléndez OM, et al. In vivo study of cerebral white matter in the dog using diffusion tensor tractography. Vet Radiol Ultrasound 2015;56:188195.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11. Nagar VA, Ye JR, Ng WH, et al. Diffusion-weighted MR imaging: diagnosing atypical or malignant meningiomas and detecting tumor dedifferentiation. Am J Neuroradiol 2008;29:11471152.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12. Hakyemez B, Yildirim N, Gokalp G, et al. The contribution of diffusion-weighted MR imaging to distinguishing typical from atypical meningiomas. Neuroradiology 2006;48:513520.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 13. Garosi L, McConnell JF, Platt SR, et al. Clinical and topographic magnetic resonance characteristics of suspected brain infarction in 40 dogs. J Vet Med Intern Med 2006;20:311321.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 14. Cervera V, Mai W, Vite CH, et al. Comparative magnetic resonance imaging findings between gliomas and presumed cerebrovascular accidents in dogs. Vet Radiol Ultrasound 2011;52:3340.

    • Search Google Scholar
    • Export Citation
  • 15. McConnell JF, Garosi L, Platt SR. Magnetic resonance imaging findings of presumed cerebellar cerebrovascular accident in twelve dogs. Vet Radiol Ultrasound 2005;46:110.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 16. MacLellan MJ, Ober CP, Feeney DA, et al. Diffusion-weighted magnetic resonance imaging of the brain of neurologically normal dogs. Am J Vet Res 2017;78:601608.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 17. Kang BK, Dong GN, Jae WR. at al. Diffusion-weighted MR imaging of intracerebral hemorrhage. Korean J Radiol 2001;2:183191.

Advertisement