• View in gallery
    Figure 1—

    Time-attenuation curves of the aorta (line a) and the pancreatic parenchyma (line b) obtained from a healthy adult Beagle during CT perfusion analysis. Blood flow was calculated from the ratio of maximum slope of the tissue time-attenuation curve (line c) to maximum arterial enhancement (point d). Blood volume was calculated from the ratio of maximum tissue enhancement (point e) to maximum arterial enhancement (point d). Time to the start of pancreatic enhancement (interval f) was defined as the time from the onset of aortic enhancement to the onset of pancreatic parenchyma enhancement (ie, time to the start of tissue perfusion). Time to peak pancreatic enhancement (interval g) was defined as the time from the onset of pancreatic parenchyma enhancement to the highest enhancement value in an ROI for the pancreas (ie, peak of tissue perfusion).

  • View in gallery
    Figure 2—

    Perfusion CT images of the pancreas in a healthy adult Beagle. Images were obtained with 4.8-mm (A, C, E, G, I, K, and M) and 2.4-mm (B, D, F, H, J, L, and N) slice thicknesses. An ROI (dotted circle) was placed over the pancreatic body on the maximum intensity projection images (A and B). Simultaneous calculation of perfusion variables, including blood flow (C and D), blood volume (E and F), time to the start of pancreatic enhancement (G and H), time to peak pancreatic enhancement (I and J), permeability (K and L), and blood volume assessed by Patlak plot analysis (M and N), was performed, and data are expressed as color maps. BF = Blood flow. BV = Blood volume (derived by the maximum slope method). MIP = Maximum intensity projection. PB = Permeability. PBV = Blood volume determined by Patlak plot analysis. TP = Time to peak pancreatic enhancement. TS = Time to start of pancreatic enhancement.

  • 1. Kambadakone AR, Sahani DV. Body perfusion CT: technique, clinical applications, and advances. Radiol Clin North Am 2009;47:161178.

  • 2. Tsuji Y, Yamamoto H, Yazumi S, et al. Perfusion computerized tomography can predict pancreatic necrosis in early stages of severe acute pancreatitis. Clin Gastroenterol Hepatol 2007;5:14841492.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 3. Kishimoto M, Tsuji Y, Katabami N, et al. Measurement of canine pancreatic perfusion using dynamic computed tomography: influence of input-output vessels on deconvolution and maximum slope methods. Eur J Radiol 2011;77:175181.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 4. Miles KA, Griffiths MR. Perfusion CT: a worthwhile enhancement? Br J Radiol 2003;76:220231.

  • 5. Petralia G, Bonello L, Viotti S, et al. CT perfusion in oncology: how to do it. Cancer Imaging 2010;10:819.

  • 6. Ropers D, Baum U, Pohle K, et al. Detection of coronary artery stenoses with thin-slice multi-detector row spiral computed tomography and multiplanar reconstruction. Circulation 2003;107:664666.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7. Schwarz T, O'Brien R. CT acquisition principles. In: Schwarz T, Saunders J, eds. Veterinary computed tomography. Ames, Iowa: John Wiley & Sons, 2011;7576.

    • Search Google Scholar
    • Export Citation
  • 8. Martínez-de-Alegría A, García-Figueiras R, Baleato-González S, et al. Effect of slice thickness reconstruction on quantitative vascular parameters in perfusion CT. Open J Radiol 2014;4:5359.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 9. Penninck DG, Zeyen U, Taeymans ON, et al. Ultrasonographic measurement of the pancreas and pancreatic duct in clinically normal dogs. Am J Vet Res 2013;74:433437.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 10. Bize PE, Platon A, Becker CD, et al. Perfusion measurement in acute pancreatitis using dynamic perfusion MDCT. AJR Am J Roentgenol 2006;186:114118.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11. Cicchetti DV. Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychol Assess 1994;6:284290.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12. Bennink E, Oosterbroek J, Horsch AD, et al. Influence of thin slice reconstruction on CT brain perfusion analysis. PLoS One 2015;10:e0137766.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 13. Martuscelli E, Romagnoli A, D'Eliseo A, et al. Accuracy of thin-slice computed tomography in the detection of coronary stenoses. Eur Heart J 2004;25:10431048.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 14. Retico A, Delogu P, Fantacci ME, et al. Lung nodule detection in low-dose and thin-slice computed tomography. Comput Biol Med 2008;38:525534.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 15. Zwingenberger AL, Shofer FS. Dynamic computed tomographic quantitation of hepatic perfusion in dogs with and without portal vascular anomalies. Am J Vet Res 2007;68:970974.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 16. Nabavi DG, Cenic A, Craen RA, et al. CT assessment of cerebral perfusion: experimental validation and initial clinical experience. Radiology 1999;213:141149.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 17. Longnecker DS. Anatomy and histology of the pancreas. Pancreapedia: exocrine pancreas knowledge base. Miami, Fla: American Pancreatic Association. Available at: pancreapedia.org/?q=node/8098. Accessed Nov 5, 2019.

    • Search Google Scholar
    • Export Citation
  • 18. Larvin M, Chalmers AG, McMahon MJ. Dynamic contrast enhanced computed tomography: a precise technique for identifying and localising pancreatic necrosis. BMJ 1990;300:14251428.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 19. Tsushima Y, Kusano S. Age-dependent decline in parenchymal perfusion in the normal human pancreas: measurement by dynamic computed tomography. Pancreas 1998;17:148152.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 20. Evans HE, de Lahunta A. The digestive apparatus and abdomen. In: Evans HE, de Lahunta A, eds. Miller's anatomy of the dog. 4th ed. St Louis: Elsevier Saunders, 2013;281337.

    • Search Google Scholar
    • Export Citation
  • 21. Schwarz T. Artifacts in CT. In: Schwarz T, Saunders J, eds. Veterinary computed tomography. Ames, Iowa: John Wiley & Sons, 2011;3556.

    • Search Google Scholar
    • Export Citation
  • 22. Konstas AA, Goldmakher GV, Lee TY, et al. Theoretic basis and technical implementations of CT perfusion in acute ischemic stroke, part 1: theoretic basis. AJNR Am J Neuroradiol 2009;30:662668.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 23. Lee S-K, Kim C, Yoon S, et al. Effect of contrast medium injection rate on computed tomography-derived renal perfusion estimates obtained with the maximum slope method in healthy Beagles. Am J Vet Res 2019;80:168173.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 24. Zussman BM, Boghosian G, Gorniak RJ, et al. The relative effect of vendor variability in CT perfusion results: a method comparison study. AJR Am J Roentgenol 2011;197:468473.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 25. Zussman B, Jabbour P, Talekar K, et al. Sources of variability in computed tomography perfusion: implications for acute stroke management. Neurosurg Focus 2011;30:E8.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 26. Ng CS, Wei W, Ghosh P, et al. Observer variability in CT perfusion parameters in primary and metastatic tumors in the lung. Technol Cancer Res Treat 2018;17:1533034618769767.

    • Search Google Scholar
    • Export Citation
  • 27. Ng QS, Goh V, Fichte H, et al. Lung cancer perfusion at multi-detector row CT: reproducibility of whole tumor quantitative measurements. Radiology 2006;239:547553.

    • Crossref
    • Search Google Scholar
    • Export Citation

Advertisement

Effect of slice thickness on computed tomographic perfusion analysis of the pancreas in healthy dogs

Seungjo Park1Department of Veterinary Medical Imaging, College of Veterinary Medicine, and BK21 Plus Project Team, Chonnam National University, Gwangju 61186, Republic of Korea.

Search for other papers by Seungjo Park in
Current site
Google Scholar
PubMed
Close
 DVM, MS
,
Jin-Woo Jung1Department of Veterinary Medical Imaging, College of Veterinary Medicine, and BK21 Plus Project Team, Chonnam National University, Gwangju 61186, Republic of Korea.

Search for other papers by Jin-Woo Jung in
Current site
Google Scholar
PubMed
Close
 DVM
,
Hyejin Je1Department of Veterinary Medical Imaging, College of Veterinary Medicine, and BK21 Plus Project Team, Chonnam National University, Gwangju 61186, Republic of Korea.

Search for other papers by Hyejin Je in
Current site
Google Scholar
PubMed
Close
 DVM
,
Youjung Jang1Department of Veterinary Medical Imaging, College of Veterinary Medicine, and BK21 Plus Project Team, Chonnam National University, Gwangju 61186, Republic of Korea.

Search for other papers by Youjung Jang in
Current site
Google Scholar
PubMed
Close
 DVM
, and
Jihye Choi1Department of Veterinary Medical Imaging, College of Veterinary Medicine, and BK21 Plus Project Team, Chonnam National University, Gwangju 61186, Republic of Korea.

Search for other papers by Jihye Choi in
Current site
Google Scholar
PubMed
Close
 DVM, PHD

Abstract

OBJECTIVE

To evaluate the effect of slice thickness on CT perfusion analysis of the pancreas in healthy dogs.

ANIMALS

12 healthy Beagles.

PROCEDURES

After precontrast CT scans, CT perfusion scans of the pancreatic body were performed every second for 30 seconds by sequential CT scanning after injection of contrast medium (iohexol; 300 mg of 1/kg) at a rate of 3 mL/s. Each dog underwent CT perfusion scans twice in a crossover-design study with 2 different slice thicknesses (2.4 and 4.8 mm). Computed tomographic pancreatic perfusion variables, including blood flow, blood volume determined with the maximum slope model, times to the start of enhancement and peak enhancement, permeability, and blood volume determined by Patlak plot analysis, were measured independently by 2 reviewers. The CT perfusion variables were compared between slice thicknesses. Interoperator reproducibility was determined by ICC calculation.

RESULTS

Interoperator reproducibility of CT perfusion variable measurements was excellent on 2.4-mm (mean ± SD ICC, 0.81 ± 0.17) and 4.8-mm (0.90 ± 0.07) slice thicknesses, except for time to peak pancreatic enhancement on 2.4-mm-thick slices, which had moderate reproducibility (intraclass correlation coefficient, 0.473). There was no significant difference in measurements of blood flow, blood volume by either method, times to the start and peak of pancreatic enhancement, or permeability between slice thicknesses.

CONCLUSIONS AND CLINICAL RELEVANCE

Results supported that a thin slice thickness of 2.4 mm can be used for assessment of pancreatic perfusion variables in healthy dogs.

Abstract

OBJECTIVE

To evaluate the effect of slice thickness on CT perfusion analysis of the pancreas in healthy dogs.

ANIMALS

12 healthy Beagles.

PROCEDURES

After precontrast CT scans, CT perfusion scans of the pancreatic body were performed every second for 30 seconds by sequential CT scanning after injection of contrast medium (iohexol; 300 mg of 1/kg) at a rate of 3 mL/s. Each dog underwent CT perfusion scans twice in a crossover-design study with 2 different slice thicknesses (2.4 and 4.8 mm). Computed tomographic pancreatic perfusion variables, including blood flow, blood volume determined with the maximum slope model, times to the start of enhancement and peak enhancement, permeability, and blood volume determined by Patlak plot analysis, were measured independently by 2 reviewers. The CT perfusion variables were compared between slice thicknesses. Interoperator reproducibility was determined by ICC calculation.

RESULTS

Interoperator reproducibility of CT perfusion variable measurements was excellent on 2.4-mm (mean ± SD ICC, 0.81 ± 0.17) and 4.8-mm (0.90 ± 0.07) slice thicknesses, except for time to peak pancreatic enhancement on 2.4-mm-thick slices, which had moderate reproducibility (intraclass correlation coefficient, 0.473). There was no significant difference in measurements of blood flow, blood volume by either method, times to the start and peak of pancreatic enhancement, or permeability between slice thicknesses.

CONCLUSIONS AND CLINICAL RELEVANCE

Results supported that a thin slice thickness of 2.4 mm can be used for assessment of pancreatic perfusion variables in healthy dogs.

Computed tomographic perfusion analysis is an imaging procedure that allows functional evaluation of tissue vascularity on the basis of temporal changes in tissue attenuation after IV injection of contrast medium. In human patients with acute pancreatitis, CT perfusion analysis allows detection of pancreatic ischemia with excellent sensitivity (100%) and specificity (95.3%).1,2 Ischemic pancreatic tissue has lower perfusion variable values than normal pancreatic tissue, and there is a high probability that tissue ischemia will lead to necrosis; consequently, this finding is associated with a poor prognosis.2 Early detection of pancreatic ischemia allows early intensive treatment to help prevent severe complications and improve prognosis.2 To the authors' knowledge, only 1 study3 has investigated canine pancreatic perfusion by means of CT perfusion analysis. That study3 investigated the applicability of CT perfusion algorithms such as the maximum slope and deconvolution methods for evaluation of pancreatic perfusion in dogs; pancreatic blood flow, blood volume, and mean transit time were measured in Beagles by use of CT with a 5-mm slice thickness.3 We are aware of no other studies on the protocol for CT perfusion analysis of the pancreas in dogs.

The maximum slope method is a compartmental analytic method used for quantification of CT perfusion.1,4 This mathematical modeling technique obtains time-attenuation data from arterial input and estimates perfusion from the maximum slope on the basis of a single compartment model.1,4 Because CT perfusion analysis estimates tissue perfusion with theoretical assumptions, there can be a difference between CT perfusion analysis-derived data and actual tissue perfusion owing to technical factors.4 Image noise and patient motion can affect the reliability of CT perfusion data.1,4 Slice thickness is an important technical factor influencing CT perfusion analysis-derived data. In general, a slice thickness of ≤ 5 mm is recommended in most CT perfusion analysis protocols to provide an ideal spatial resolution and signal-to-noise ratio in humans.5 In practice, slice thickness is set to 5 to 10 mm for CT perfusion, regardless of the patient's size.1,6 However, slice thickness is proportional to the magnitude of partial volume artifact, and small anatomic structures need a < 5-mm slice thickness for CT perfusion.7 Results of a previous study8 on the application of CT perfusion analysis in the diagnosis of lung, rectal, or renal cancer in human patients reveal that different slice-thickness reconstructions induce significant variability in perfusion variables, particularly transit time. However, variability was smaller when the difference between slice thickness was smaller (eg, 1.25 vs 2.5 mm or 2.5 vs 5 mm) than when it was larger (eg, 1.25 vs 5 mm).

Because the pancreas in dogs has a thin, elongated conformation and is < 10 mm thick, CT perfusion analysis with 5- to 10-mm slice thicknesses can induce a marked partial volume effect.9 The purpose of the study reported here was to investigate the effect of slice thickness on pancreatic CT perfusion analysis in healthy dogs to assess the applicability of a thin slice (selected as 2.4 mm) for CT perfusion analysis of the pancreas in healthy dogs. We hypothesized that there would be no significant difference in pancreatic perfusion variables obtained with 2.4-mm and 4.8-mm slice thicknesses.

Materials and Methods

Animals

A crossover experimental design that included 12 purpose-bred Beagles (5 males and 7 females) was used. The median age of the dogs was 2 years (range, 1 to 3 years), and the median weight was 9.8 kg (range, 7.7 to 14.5 kg). All the dogs were deemed healthy on the basis of results of a physical examination, CBC, serum biochemical analysis, canine pancreatic lipase immunoreactivity testing,a electrolyte evaluation, and radiographic and abdominal ultrasonographic examination. The dogs were housed individually and were fed commercial dry food with tap water available ad libitum. This study was approved by the Institutional Animal Care and Use Committee of Chonnam National University, and the animals were cared for in accordance with the Guidelines for Animal Experiments of Chonnam National University (CNU IACUC-YB-2018-82).

CT perfusion

After food was withheld for 24 hours, general anesthesia was induced by IM injection of a combination of zolazepam hydrochloride-tiletamine hydrochlorideb (0.75 mg/kg) and medetomidine hydrochloridec (0.03 mg/kg). An endotracheal tube was placed, and anesthesia was maintained with isofluraned (1% to 2%) in oxygen (1 L/min).

A 20-gauge catheter was placed in a cephalic vein for bolus injection of contrast medium. The CT scans were conducted by use of a 16-channel multidetector CT scannere with each dog positioned in ventral recumbency. A precontrast CT scan was performed from the level of the diaphragm to a level caudal to the left kidney to include the pancreatic body with the following settings: slice thickness, 2 mm; pitch, 0.8; rotation time, 1 second; tube voltage, 110 kVp; and tube current, 100 mA. Subsequently, CT perfusion imaging was performed at the level of the pancreatic body with the following settings: rotation time, 1 second; tube voltage, 120 kVp; and tube current, 110 mA. The CT perfusion images were acquired every second for 30 seconds by sequential CT scanning with the initiation of contrast medium injection. Iohexolf (300 mg of I/kg) was administered via the cephalic vein catheter at a rate of 3 mL/s by an investigator using a power injector.f Each dog underwent CT perfusion scanning twiceg in the same order (once with a slice thickness of 4.8 mm and once with a slice thickness of 2.4 mm) with a 1-week interval between the scans. For the acquisition of CT images with a slice thickness of 4.8 mm, twelve 1.2-mm wide detectors were used (12 × 1.2 mm), and for CT images with a slice thickness of 2.4 mm, sixteen 0.6-mm wide detectors were used (16 × 0.6 mm). With each slice thickness, 4 consecutive images were obtained. Apnea was induced by hyperventilation immediately before the CT scan to minimize motion artifacts, and CT images were obtained by use of an adaptive smoothing filter. After CT perfusion imaging, the general condition of each dog, including possible adverse effects such as vomiting, signs of depression, and anorexia, was monitored for 5 days.

Image analysis

All CT perfusion images were sent to a workstation and reviewed through the use of an installed software programh by a veterinarian who was a fourth-year PhD student (SP) and a veterinarian with 2 years of radiology experience (J-WJ). A window width of 400 HU and window level of 100 HU were used for image review. Each reviewer performed measurements independently for determination of the inter-reviewer ICC; for comparison of data between CT slice thicknesses, measurements obtained by 1 reviewer (SP) were used.

The CT perfusion variables were derived by use of the maximum slope method for blood flow and blood volume and by use of a 2-compartmental model for permeability and for blood volume estimation by Patlak plot analysis.10 Time to the start of contrast enhancement and time to peak enhancement were measured on the basis of time-density curves. Among the 4 perfusion images, 1 representative image was selected by the 2 evaluators for analysis on the basis of image quality by evaluation of variables such as a wide area showing the pancreas, a low distribution of blood vessels within the image, and a low partial volume artifact.

First, a circular ROI with a diameter of 4 to 6 mm was placed over the aorta at the same level as that of the chosen region with the pancreatic body to determine the start of aortic enhancement and obtain the peak enhancement measurement for the aorta. Second, after a color map was generated automatically, circular ROIs with a diameter of 1 to 3 mm were placed in the pancreatic body. An effort was made to exclude blood vessels passing through the pancreatic parenchyma from the ROIs to the extent possible so that perfusion data would be obtained only from the pancreatic parenchyma. Then, a time-density curve for contrast enhancement of the pancreas was obtained automatically with the software. The CT perfusion analysis-derived variables for the pancreas, including blood flow (mL/100 mL of tissue/min), blood volume (mL/100 mL of tissue), time to the start of enhancement, time to peak enhancement, permeability, and blood volume determined by Patlak plot analysis, were obtained (Figure 1). Blood flow was calculated as the volume of blood entering or exiting the vasculature of a measured amount of tissue (100 mL) over 1 minute. Blood volume was determined as the amount of blood contained within the vasculature of 100 mL of tissue.1 Permeability was assessed as the flow of molecules through the capillary membranes of 100 mL of tissue over a minute. Time to the start of pancreatic enhancement was defined as the interval between the onset of aortic enhancement and the onset of tissue (pancreatic parenchyma) enhancement. Time to peak pancreatic enhancement was defined as the time from the start of pancreatic enhancement until the highest enhancement value in an ROI of the pancreas was measured.1

Figure 1—
Figure 1—

Time-attenuation curves of the aorta (line a) and the pancreatic parenchyma (line b) obtained from a healthy adult Beagle during CT perfusion analysis. Blood flow was calculated from the ratio of maximum slope of the tissue time-attenuation curve (line c) to maximum arterial enhancement (point d). Blood volume was calculated from the ratio of maximum tissue enhancement (point e) to maximum arterial enhancement (point d). Time to the start of pancreatic enhancement (interval f) was defined as the time from the onset of aortic enhancement to the onset of pancreatic parenchyma enhancement (ie, time to the start of tissue perfusion). Time to peak pancreatic enhancement (interval g) was defined as the time from the onset of pancreatic parenchyma enhancement to the highest enhancement value in an ROI for the pancreas (ie, peak of tissue perfusion).

Citation: American Journal of Veterinary Research 81, 9; 10.2460/ajvr.81.9.732

Statistical analysis

Statistical analyses were performed by 1 investigator (SP) using a commercial software package.i Reproducibility of CT perfusion variables between the 2 reviewers was determined by calculating the ICC and was defined as poor if ICC was < 0.4, moderate if it was 0.40 to 0.59, good if it was 0.60 to 0.74, and excellent if it was ≥ 0.75.11 Differences between the 2 slice thicknesses for blood flow, blood volume assessed by the maximum slope method, time to the start of pancreatic enhancement, permeability, and blood volume determined by Patlak plot analysis were evaluated with paired t tests. A Wilcoxon signed rank test was used to compare the differences in time to peak pancreatic enhancement between slice thicknesses. The data were assessed with Kolmogorov-Smirnov and Shapiro-Wilk tests, and normal distribution was confirmed. All data were expressed as mean ± SD. Values of P < 0.05 were considered significant.

Results

Twenty-four pancreatic CT perfusion images were obtained from the 12 dogs for analysis (1 each/dog at slice thicknesses of 2.4 and 4.8 mm). Measurement of CT perfusion analysis-derived variables was successfully performed for all images; an example is provided (Figure 2). There were no complications related to the CT perfusion procedures, such as vomiting, signs of depression, or anorexia, in any dog.

Figure 2—
Figure 2—

Perfusion CT images of the pancreas in a healthy adult Beagle. Images were obtained with 4.8-mm (A, C, E, G, I, K, and M) and 2.4-mm (B, D, F, H, J, L, and N) slice thicknesses. An ROI (dotted circle) was placed over the pancreatic body on the maximum intensity projection images (A and B). Simultaneous calculation of perfusion variables, including blood flow (C and D), blood volume (E and F), time to the start of pancreatic enhancement (G and H), time to peak pancreatic enhancement (I and J), permeability (K and L), and blood volume assessed by Patlak plot analysis (M and N), was performed, and data are expressed as color maps. BF = Blood flow. BV = Blood volume (derived by the maximum slope method). MIP = Maximum intensity projection. PB = Permeability. PBV = Blood volume determined by Patlak plot analysis. TP = Time to peak pancreatic enhancement. TS = Time to start of pancreatic enhancement.

Citation: American Journal of Veterinary Research 81, 9; 10.2460/ajvr.81.9.732

The ICCs for CT perfusion variables measured by the 2 reviewers at each slice thickness were summarized (Table 1). Mean reproducibility as assessed by ICC data for all CT perfusion variables at a slice thickness of 2.4 mm was excellent (mean ± SD ICC, 0.81 ± 0.17; range, 0.47 to 0.93), with similar results for a slice thickness of 4.8 mm (mean ± SD ICC, 0.90 ± 0.07; range, 0.76 to 0.96). The only variable that did not meet criteria for excellent reproducibility was time to peak pancreatic enhancement at a slice thickness of 2.4 mm, which had moderate reproducibility (ICC, 0.473).

Table 1—

Results of ICC analysis for 2 reviewers (a veterinarian with 2 years of radiology experience and a veterinarian who was a fourth-year PhD student) with regard to CT perfusion variables for the pancreas in images obtained at 2 slice thicknesses (2.4 and 4.8 mm; 1/slice thickness/dog) for 12 healthy Beagles.

Slice thickness (mm)BFBVTSTPPermeabilityPBV
2.40.9110.9240.7880.4730.930.812
4.80.910.9310.940.760.9120.955

Coefficients were assessed as follows: values < 0.4 were considered poor, values between 0.40 and 0.59 were considered moderate, values between 0.60 and 0.74 were considered good, and values ≥ 0.75 were considered excellent.

BF = Blood flow. BV = Blood volume (derived by the maximum slope method). PBV = Blood volume determined by Patlak plot analysis. TP = Time to peak pancreatic enhancement. TS = Time to start of pancreatic enhancement.

Mean ± SD peak attenuation of the aorta was measured as 419.2 ± 70.69 HU and 414.81 ± 66.88 HU with 2.4- and 4.8-mm slice thicknesses, respectively. The CT perfusion variables for the pancreas are shown (Table 2). There were no significant differences between slice thicknesses for blood flow (P = 0.776), blood volume derived by the maximum slope method (P = 0.869), times to start of (P = 0.826) and peak (P = 0.131) pancreatic enhancement, permeability (P = 0.970), or blood volume determined with Patlak plot analysis (P = 0.554).

Table 2—

Mean ± SD pancreatic perfusion variables measured on CT images obtained at 2 slice thicknesses (2.4 and 4.8 mm; 1/slice thickness/dog) for 12 healthy Beagles.

Slice thickness (mm)BF (mL/100 mL/min)BV (mL/100 mL)TS (× 10−1 s)TP (× 10−1 s)Permeability (mL/100 mL/min)PBV (mL/100 mL)
2.437.4 ± 14.55.6 ± 2.435.3 ± 18.5121.9 ± 32.958.9 ± 27.32.5 ± 1.5
4.838.2 ± 12.35.7 ± 1.939.9 ± 13.6134.4 ± 14.960.4 ± 15.42.6 ± 1.1

Measurements used for this comparison were made by 1 reviewer (SP).

See Table 1 for key.

The volume CT dose index for perfusion scans with 2.4-mm slice thickness was 288.47 mGy/dog. The volume CT dose index for perfusion scans with 4.8-mm slice thickness was 259.62 mGy/dog.

Discussion

The results of the present study indicated that CT perfusion analysis of the pancreatic body in dogs could be performed with a thinner slice thickness of 2.4 mm with no significant differences in measured variables, compared with results for the generally used slice thickness of 4.8 mm. The greatest apparent difference in measured variables was for the time to peak pancreatic enhancement (121.9 × 10−1 vs 134.4 × 10−1 for thicknesses of 2.4 and 4.8 mm, respectively). The ICC analysis results indicated excellent reproducibility of measurements between reviewers for most (5/6) variables when the 2.4-mm slice thickness was used, which was similar to the reproducibility obtained when the 4.8-mm slice thickness was used (excellent reproducibility for 6/6 variables).

In CT perfusion studies of people, thick-slice (> 5-mm) reconstruction is almost universally recommended for a balance between the requirements of spatial resolution and signal-to-noise ratio.8 Even if CT perfusion scans are performed with thinner slices, reconstruction into thick slices is performed to reduce the amount of data and processing time.12 However, obtaining thick slices for CT perfusion analysis is not always possible, even in human patients when there is a need for visualizing small anatomic structures such as those in the brain, because partial volume artifacts can affect the CT perfusion variables.12 Multidetector CT scanners allow dynamic acquisition of thin slices, and many studies6,13,14 have investigated the effect of thin (0.625- to 1-mm) slices on CT perfusion variables of the brain, pulmonary nodules, and stenotic coronary arteries. In veterinary medicine, a few reports15,16 of CT perfusion analysis have been published, and these studies involved the use of 2- to 7-mm slice thicknesses for imaging of the liver, pancreas, stomach, and brain.

The human pancreas is approximately 14 to 20 cm long and 1 to 3 cm thick, and it consists of a head, neck, body, and tail.17 The head of the human pancreas is on the right side of the abdomen, and the tail extends on the left side of the body.17 This means that the pancreas is elongated along the x-axis of the CT gantry, which allows the acquisition of wider images. In people, CT perfusion imaging has been successfully used to evaluate tissue perfusion of the pancreatic head, body, and tail with slice thicknesses of 8 to 10 mm.2,18,19

Unlike the human pancreas, the canine pancreas is a long, narrow, V-shaped organ, and it consists of a short, thick body with long and narrow left and right lobes.20 In a previous study3 on CT perfusion imaging of the pancreas in Beagles, a 5-mm slice thickness was used and ROIs were set over the center of pancreas (which includes the pancreatic body and right lobes). The thickness of the pancreas in dogs is reported to be < 10 mm.9 The thickness of all pancreatic lobes increased significantly with an increase in body weight; however, when the pancreatic thickness was measured by ultrasonography, mean ± SD value for the left lobe was 6.5 ± 1.7 mm, body was 6.3 ± 1.6 mm, and right lobe was 8.1 ± 1.8 mm in dogs with body weights ranging from 1.4 to 55 kg.9 Partial volume artifact is the result of mean attenuation along the path of the beam, and it occurs when an object is partially intruding into the width of the x-ray beam or is smaller than the scanning plane.21 The left lobe is located caudal to the greater curvature of the stomach, and it continues medially to the spleen along the x-axis of the gantry. Thus, the left lobe is likely to have partial volume artifact when thick slices are used for CT perfusion. The right lobe lies dorsomedial to the duodenum, ventral to the right kidney, and lateral to the portal vein. The right lobe is elongated along the right lateral aspect of the abdomen (z-axis) and is relatively free from partial volume artifact. The pancreatic body is located at the pyloroduodenal junction, ventral to the portal vein. Thus, in our study, the pancreatic body was selected for CT perfusion analysis after considering several factors. First, the pancreatic body is generally the widest portion where the ROI can be set appropriately. Second, the location of the pancreatic body is more consistent than that of the other 2 lobes, with the location of the stomach or small intestines as a reference. Third, the pancreatic body extends caudally to the right lobe, which reduces the likelihood of partial volume artifacts, and the left lobe extends to a transverse plane of the abdomen, which can lead to substantial partial volume artifacts. In an unpublished pilot study by our group that involved selection of a location for CT perfusion imaging of the pancreas in dogs, it was difficult to obtain 4 consecutive images of the left pancreatic lobe with a 4.8-mm slice thickness, and most of the CT perfusion images showed obvious partial volume artifacts. In the right pancreatic lobe, the pancreatic parenchyma included in 1 CT slice was too small to set the ROI with confidence because of the position of the right lobe (elongated along the abdominal wall).

In CT perfusion analysis, measurement of perfusion and permeability can be performed on CT images by postprocessing with a compartmental model such as a maximum slope model or with a deconvolution model.3 The deconvolution technique is theoretically superior to the maximum slope model; however, this method requires more complicated and time-consuming processing.1 Moreover, patient motion and partial volume averaging can significantly affect the computation of blood flow with the deconvolution method.1 The maximum slope method has advantages of simplicity, a high speed of calculation for the perfusion variables, and low susceptibility to respiratory motion artifact because of relatively few data acquisitions.1 However, this method assumes that there is no outflow and no delay in the time from arterial input to tissue enhancement during CT perfusion analysis, and there is a possibility of underestimation of blood flow.3 There have been many investigations on the interchangeability or differences in CT perfusion variables derived by the maximum slope and deconvolution methods, and significant differences in perfusion variables have been reported between the 2 methods.3 Thus, CT perfusion variables calculated by different methods are not directly interchangeable; however, the pairs of perfusion variables were significantly correlated with each other. In the present study, pancreatic perfusion variables including blood flow and blood volume were assessed by an investigator using the maximum slope method, which requires rapid injection of contrast medium because a longer duration of injection theoretically results in an underestimation of perfusion.22 However, in our previous study23 to investigate the effect of flow rate on renal CT perfusion variables in dogs, blood flow and blood volume were satisfactorily determined with 3 different rates of contrast agent injection (1.5, 3.0, and 4.5 mL/s) with the maximum slope method. We determined the dose of contrast medium on the basis of body weight, and the volume required for small patients is small enough to permit a low injection rate while providing a sufficiently short injection duration for determination of CT perfusion variables with the maximum slope method.23 Therefore, an injection rate of 3 mL/s was used for CT perfusion imaging of the pancreas in dogs of the present study.

In a previous CT perfusion study3 of the pancreas in 9 Beagles, blood flow, blood volume, and mean transit time were evaluated with the maximum slope method; CT perfusion variables were determined after placing ROIs for 3 different arteries including the abdominal aorta and splenic and celiac arteries, and the pancreatic CT perfusion variables differed according to the vessel selection. When the abdominal aorta was selected and the maximum slope model and 5-mm slice thickness were used, mean ± SD blood flow was 46.1 ± 14.5 mL/100 g/min and blood volume was 2.5 ± 0.09 mL/100 g. In the present study, mean ± SD blood flow was 38.2 ± 12.3 mL/100 mL/min and blood volume with the maximum slope model was 5.7 ± 1.9 mL/100 mL when the abdominal aorta was selected and a 4.8-mm slice thickness was used. Even though ROI placement and slice thickness were similar in the 2 studies, the perfusion variables could have differed according to the pancreatic region used for the injection rate of contrast medium, CT image acquisition, number of images obtained by CT scan per unit of time, software operational modes, and different vendor software applications.24,25

In the present study, nearly all CT perfusion variables were measured with excellent reproducibility regardless of slice thickness; however, time to peak pancreatic enhancement determined with thin slices had a moderate ICC. Observer variability is an important factor in evaluating the reproducibility of CT perfusion variables for the canine pancreas on thin slices.26 The main observer-dependent factors that can affect the computation of CT perfusion variables are setting an ROI on the target tissue and defining an arterial input function. The CT perfusion data sets typically consist of contiguous transverse slices, and we set ROIs on 1 representative image/dog. In studies26,27 of CT perfusion analysis in human patients with cancer, ROIs are typically set on each slice in which the tumor is visualized. Unlike CT perfusion images of the brain, CT perfusion images of abdominal organs such as the pancreas are susceptible to motion. In our study, CT perfusion imaging was performed in dogs under general anesthesia, apnea was induced by hyperventilation, and CT data were obtained for only 30 seconds to minimize motion artifacts; however, motion related to breathing and cardiac movement could lead to a high degree of variance in ROI placement by 2 reviewers, particularly when thin slices are used.

There were some limitations to the study reported here. First, only a small number of dogs of the same breed were used. Second, histologic examination was not performed to confirm that the pancreas in these dogs was histologically normal. Although the dogs were determined to be healthy on the basis of medical history and results for many laboratory tests, the possibility of minor lesions could not be completely ruled out. Third, pancreatic CT perfusion variables of the left and right lobes of the pancreas were not evaluated. Evaluation of these lobes with the 4.8-mm slice thickness would be difficult because of partial volume artifacts; however, evaluation with the 2.4-mm slice thickness might be possible. Finally, the CT perfusion analysis was not performed in dogs with known pancreatic abnormalities.

Our results indicated that estimation of the pancreatic perfusion variables in clinically normal dogs is feasible with a thin (2.4-mm) slice thickness and suggested that slice thickness has no significant effect on CT perfusion analysis variables for the pancreas in healthy dogs. When thin CT slices were used, blood flow, blood volume with the maximum slope method, time to the start of pancreatic enhancement, permeability, and blood volume by Patlak plot analysis were determined with excellent reproducibility; however, time to peak showed had moderate reproducibility. Overall, the findings suggested a slice thickness of 2.4 mm can replace the generally used slice thickness for evaluation of pancreatic perfusion variables in healthy dogs. Further research to assess the use of thin CT slices to examine pancreatic perfusion in dogs with pancreatic abnormalities is warranted.

Acknowledgments

This research was supported by the Animal Medical Institute of Chonnam National University and by the Basic Science Research Program through the National Research Foundation of Korea, funded by the Ministry of Science, ICT, and Future Planning (NRF-2018R1A2B6006775).

The authors declare that there were no conflicts of interest.

ABBREVIATIONS

ICC

Intraclass correlation coefficient

ROI

Region of interest

Footnotes

a.

SNAP cPL test, Idexx Laboratories Inc, Westbrook, Me.

b.

Zoletil, Virbac, Carros, France.

c.

Domitor, Orion Corp, Espoo, Finland.

d.

Terrell, Piramal Critical Care Inc, Bethlehem, Pa.

e.

Emotion 16, Siemens, Forchheim, Germany.

f.

Omnipaque 300, GE Healthcare, Oslo, Norway.

g.

Optivantage DH, Liebel-Flarsheim Co, Cincinnati, Ohio.

h.

Syngo body perfusion CT, Siemens, Forchheim, Germany.

i.

SPSS Statistics, version 23, IBM Corp, Armonk, NY.

References

  • 1. Kambadakone AR, Sahani DV. Body perfusion CT: technique, clinical applications, and advances. Radiol Clin North Am 2009;47:161178.

  • 2. Tsuji Y, Yamamoto H, Yazumi S, et al. Perfusion computerized tomography can predict pancreatic necrosis in early stages of severe acute pancreatitis. Clin Gastroenterol Hepatol 2007;5:14841492.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 3. Kishimoto M, Tsuji Y, Katabami N, et al. Measurement of canine pancreatic perfusion using dynamic computed tomography: influence of input-output vessels on deconvolution and maximum slope methods. Eur J Radiol 2011;77:175181.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 4. Miles KA, Griffiths MR. Perfusion CT: a worthwhile enhancement? Br J Radiol 2003;76:220231.

  • 5. Petralia G, Bonello L, Viotti S, et al. CT perfusion in oncology: how to do it. Cancer Imaging 2010;10:819.

  • 6. Ropers D, Baum U, Pohle K, et al. Detection of coronary artery stenoses with thin-slice multi-detector row spiral computed tomography and multiplanar reconstruction. Circulation 2003;107:664666.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7. Schwarz T, O'Brien R. CT acquisition principles. In: Schwarz T, Saunders J, eds. Veterinary computed tomography. Ames, Iowa: John Wiley & Sons, 2011;7576.

    • Search Google Scholar
    • Export Citation
  • 8. Martínez-de-Alegría A, García-Figueiras R, Baleato-González S, et al. Effect of slice thickness reconstruction on quantitative vascular parameters in perfusion CT. Open J Radiol 2014;4:5359.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 9. Penninck DG, Zeyen U, Taeymans ON, et al. Ultrasonographic measurement of the pancreas and pancreatic duct in clinically normal dogs. Am J Vet Res 2013;74:433437.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 10. Bize PE, Platon A, Becker CD, et al. Perfusion measurement in acute pancreatitis using dynamic perfusion MDCT. AJR Am J Roentgenol 2006;186:114118.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11. Cicchetti DV. Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychol Assess 1994;6:284290.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12. Bennink E, Oosterbroek J, Horsch AD, et al. Influence of thin slice reconstruction on CT brain perfusion analysis. PLoS One 2015;10:e0137766.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 13. Martuscelli E, Romagnoli A, D'Eliseo A, et al. Accuracy of thin-slice computed tomography in the detection of coronary stenoses. Eur Heart J 2004;25:10431048.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 14. Retico A, Delogu P, Fantacci ME, et al. Lung nodule detection in low-dose and thin-slice computed tomography. Comput Biol Med 2008;38:525534.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 15. Zwingenberger AL, Shofer FS. Dynamic computed tomographic quantitation of hepatic perfusion in dogs with and without portal vascular anomalies. Am J Vet Res 2007;68:970974.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 16. Nabavi DG, Cenic A, Craen RA, et al. CT assessment of cerebral perfusion: experimental validation and initial clinical experience. Radiology 1999;213:141149.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 17. Longnecker DS. Anatomy and histology of the pancreas. Pancreapedia: exocrine pancreas knowledge base. Miami, Fla: American Pancreatic Association. Available at: pancreapedia.org/?q=node/8098. Accessed Nov 5, 2019.

    • Search Google Scholar
    • Export Citation
  • 18. Larvin M, Chalmers AG, McMahon MJ. Dynamic contrast enhanced computed tomography: a precise technique for identifying and localising pancreatic necrosis. BMJ 1990;300:14251428.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 19. Tsushima Y, Kusano S. Age-dependent decline in parenchymal perfusion in the normal human pancreas: measurement by dynamic computed tomography. Pancreas 1998;17:148152.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 20. Evans HE, de Lahunta A. The digestive apparatus and abdomen. In: Evans HE, de Lahunta A, eds. Miller's anatomy of the dog. 4th ed. St Louis: Elsevier Saunders, 2013;281337.

    • Search Google Scholar
    • Export Citation
  • 21. Schwarz T. Artifacts in CT. In: Schwarz T, Saunders J, eds. Veterinary computed tomography. Ames, Iowa: John Wiley & Sons, 2011;3556.

    • Search Google Scholar
    • Export Citation
  • 22. Konstas AA, Goldmakher GV, Lee TY, et al. Theoretic basis and technical implementations of CT perfusion in acute ischemic stroke, part 1: theoretic basis. AJNR Am J Neuroradiol 2009;30:662668.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 23. Lee S-K, Kim C, Yoon S, et al. Effect of contrast medium injection rate on computed tomography-derived renal perfusion estimates obtained with the maximum slope method in healthy Beagles. Am J Vet Res 2019;80:168173.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 24. Zussman BM, Boghosian G, Gorniak RJ, et al. The relative effect of vendor variability in CT perfusion results: a method comparison study. AJR Am J Roentgenol 2011;197:468473.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 25. Zussman B, Jabbour P, Talekar K, et al. Sources of variability in computed tomography perfusion: implications for acute stroke management. Neurosurg Focus 2011;30:E8.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 26. Ng CS, Wei W, Ghosh P, et al. Observer variability in CT perfusion parameters in primary and metastatic tumors in the lung. Technol Cancer Res Treat 2018;17:1533034618769767.

    • Search Google Scholar
    • Export Citation
  • 27. Ng QS, Goh V, Fichte H, et al. Lung cancer perfusion at multi-detector row CT: reproducibility of whole tumor quantitative measurements. Radiology 2006;239:547553.

    • Crossref
    • Search Google Scholar
    • Export Citation

Contributor Notes

Address correspondence to Dr. Choi (imsono@chonnam.ac.kr).