Evaluation of the size of the liver by means of diagnostic imaging is important in the identification of hepatic abnormalities. Liver size is a significant prognostic indicator of survival in humans with compensated cirrhosis and hepatic failure.1 Hepatomegaly indicates pathological conditions such as inflammation, neoplasia, congestion, edema, and fatty infiltration, whereas microhepatia suggests hepatic fibrosis, cirrhosis, congenital atrophy, emaciation, or PSS.2,3
Various imaging techniques have been used to determine liver size in dogs, including radiography, ultrasonography, scintigraphy, CT, and MRI.2,4-7 Radiographic evaluation of liver size involves both morphological and quantitative assessment. In dogs with generalized liver enlargement, the caudoventral liver margin appears round or blunt and extends beyond the costal arch, and caudal deviation of the gastric axis is seen.3,8-10 Radiographic signs of focal liver enlargement include bulges or alterations in the hepatic margin or localized displacement of the adjacent organs (ie, fundus, gastric body, pylorus, right kidney, cranial duodenal flexure, transverse colon, head of the spleen, or diaphragm). Microhepatia is diagnosed on the basis of a smaller than typical distance between the diaphragm and stomach, cranial displacement of the pylorus, a gastric axis that appears vertically or cranially angled, and cranial deviation of the cranial duodenal flexure, right kidney, and transverse colon.8 However, veterinarians considering these signs should take into account the dog's age, BW, thoracic conformation, and depth of respiration, and assessments relying solely on these radiographic signs are subjective and insensitive to subtle changes.8,11
For more objective, quantitative assessment of liver size in dogs, a single linear measurement of the liver shadow on a lateral radiographic image can yield liver length (length of the liver tip protruding from the costal arch or liver depth from the caudal vena cava to the liver tip) or liver volume.5,12,13 However, measurement of the length of the liver tip protruding from the costal arch is not widely used in clinical veterinary medicine because it can be affected by the patient's position (right or left recumbency or rotation of the trunk), respiratory phase, or x-ray beam center.12 Moreover, such single linear measurements poorly reflect actual liver size because the liver has a complex shape that may differ among even healthy dogs. These complexities in liver size measurement are exacerbated by the wide variety of breed conformations and BWs among dogs.4,14
Measurement of organ volume via CT is a rapid and fairly simple procedure that is reportedly accurate to within 5% in anesthetized dogs.15 Measurements of liver volume via CT are highly correlated with the actual weight of the liver, regardless of any underlying liver disease. Therefore, CT volumetry is considered the reference standard for estimation of liver volume.2,16,17 Liver volume estimation provides invaluable clinical information such as the correlation between functional reserve and liver tumors in cirrhotic patients and the progression of various hepatic failures, and it forms the basis for surgical decisions about hepatic resection to ensure that sufficient volume remains to provide adequate metabolic function.2,17 However, few reports have described CT assessment of normal liver volume in dogs coupled with prediction of liver volume by radiographic measurement. The purpose of the study reported here was to determine the feasibility of radiographic measurement of liver area by comparing the accuracy of RLA, RLL, and RLV measurements in dogs with radiographically and clinicopathologically normal livers, microhepatia, or hepatomegaly. Another objective was to determine the correlation between liver size measured via CT and via radiography.
Materials and Methods
Animals
Medical records for small-breed dogs (BW, ≤ 10 kg) evaluated at the Konkuk University Veterinary Medical Teaching Hospital between February 2013 and July 2018 were retrospectively reviewed. Only dogs that had both orthogonal thoracic and abdominal radiographs and CT scans available were eligible for inclusion in the study.
Data collection
Information was extracted from the medical records of each eligible dog regarding breed, age, sex, reproductive status, BW, clinical history, and CBC and serum biochemical values. Dogs were classified as having a normal liver size, microhepatia, or hepatomegaly by 2 experienced veterinary radiologists (HY and KE), who used the available radiographs. Dogs with a normal gastric axis on abdominal radiographs were categorized as normal,5 dogs with cranioventral deviation of the gastric axis were classified as having microhepatia,8 and dogs with rounding of the caudal hepatic borders and remarkable caudodorsal displacement of the gastric axis were classified as having hepatomegaly.8 The 2 radiologists were blinded to the dog's clinical information during these evaluations, and dogs were excluded from the study if the radiologists could not agree on the suitable classification.
Dogs were then allocated to normal, microhepatia, and hepatomegaly groups on the basis of the radiographic classification and clinicopathologic findings recorded in the medical records. Dogs for which the radiographic classification did not correspond to the clinicopathologic findings were excluded from the study. The normal group included dogs with no abnormalities identified on a CBC and serum biochemical analysis (glucose, total protein, albumin, and total bilirubin concentrations and alanine aminotransferase, alkaline phosphatase, and aspartate aminotransferase activities) and no historical or clinical signs of hepatic disease, congestive heart failure, peritoneal effusion, abdominal masses, or diaphragmatic hernias. The microhepatia group included dogs with a PSS confirmed by CT.8 The hepatomegaly group included dogs with diagnoses associated with an enlarged liver, such as hyperadrenocorticism, congestive heart failure, or infiltrative hepatopathy.8
Radiographic examination
All radiographic examinations had been performed with a digital radiography unit.a Orthogonal (right lateral and ventrodorsal) thoracic and abdominal radiographs had been obtained with dogs in standard positions and at a setting of 300 mA. Peak voltage had been selected on the basis of the thickness of the subject's thorax or abdomen.
CT examination
Abdominal CT scans had been performed for each dog by use of a 4-channel helical CT scannerb with settings of 120 kVp and 160 mA and a slice thickness of 1.25 to 2.5 mm. All scans were performed with the dogs anesthetized, and respiration was controlled to minimize motion artifacts. Contrast images were acquired 50 seconds after injection of nonionic iodinated contrast mediumc (900 mg of iodine/kg, IV) by hand into the cephalic vein.
Radiographic measurements
Digital radiographic images were analyzed by one of the investigators (SL), who used a DICOM viewerd and electronic calipers for measurements. Duplicate measurements were performed, with the second one obtained 2 weeks after the first measurement.
On right lateral radiographic images, the RLA and area of T11 were automatically calculated by free-hand outlining by use of the viewer software. In this process, the RLA was outlined to reflect the cranial border as delineated by the radiolucent lung-diaphragm border and the caudal border as delineated by the cranial margin of the stomach, proximal aspect of the duodenum, and cranial pole of the right kidney (Figure 1). An attempt was made to identify the liver boundaries as clearly as possible by adjusting the gray contrast in the viewer. The T11 area was defined by outlining the boundary of the vertebral body. To control for various BWs and sizes of the included dogs, the data were normalized by calculating the RLA:T11 area ratio and RLA:BW ratio.
Representative right lateral abdominal radiographic images used for measuring the RLA and T11 area in dogs. A—Image of a dog with a radiographically normal liver. B—Same image as in panel A, but with an outline indicating how the RLA (black line) was measured, as demarcated by the diaphragm, stomach (S), and right kidney (RK). The T11 area was measured by drawing a line around the vertebral body (white line).
Citation: American Journal of Veterinary Research 80, 12; 10.2460/ajvr.80.12.1122
To allow comparison of conventional measurement approaches for assessing liver size with those in the present study, the RLL, T11 length, and TD were also measured on right lateral radiographic views and the TW on ventrodorsal views as described elsewhere.5,8,13,18 The RLL:T11 length ratio was then computed, and RLV was calculated with the following formula19:
The RLV:BW and TD:TW ratios were then calculated.
CTLV measurements
Axial CT images had been saved as DICOM files and were loaded into imaging softwared for use in measuring liver volume for the study. All CT images were viewed with a window width of 350 HU and window level of 40 HU.
Liver volume was measured by one of the investigators (SL) using electronic segmentation of the liver parenchyma on sequential transverse post-contrast CT slices. Again, duplicate measurements were performed, with the second one obtained 2 weeks after the first measurement.
For these measurements, using segmentation tools, the investigator pointed at the liver parenchyma and roughly selected the regions occupied by the liver in each transverse image from the cranial area at the diaphragm to the most caudal part of the liver adjacent to the right kidney and spleen20 (Figure 2). The gallbladder and visible fissures of the liver were excluded from these ROIs. If the vessels (ie, hepatic arteries, veins, and portal veins) were clearly within the parenchyma, they were included in the ROIs, but if they appeared distinctly extrinsic to the liver, they were excluded. Volume-rendered images of the liver and total liver volume were created automatically from these ROIs. Liver volume and volume-rendered images were stored in a database.2 Finally, the CTLV:BW ratio was determined.
Representative transverse abdominal CT images used for measuring liver volume in a dog showing the cranial (A and B) and caudal (C and D) portions of the liver (selected as ROIs; green regions) and summation of selected ROIs (E). After the ROIs were selected, the liver area of each transverse image was automatically calculated by the imaging softwared and summed. Slice thickness, 2.5 mm; window width, 350 HU; and window level, 40 HU.
Citation: American Journal of Veterinary Research 80, 12; 10.2460/ajvr.80.12.1122
Statistical analysis
All statistical tests were performed with statistical software.e Normality of distribution of continuous data was visually confirmed by creation of probability plots against 95% confidence intervals. Data for liver size, dog age, and BW are consequently reported as mean ± SD.
An ANOVA was performed to evaluate differences in liver size among the normal, microhepatia, and hepatomegaly groups. Differences in liver size (normalized RLA, RLL, and CTLV) between sexes were assessed with the Student t test. Pearson correlation coefficients (r) were calculated to assess relationships among dog age, BW, and liver size. Values from 0.71 to 1.00, 0.31 to 0.70, and 0.10 to 0.30 were considered strong, moderate, and weak correlations, respectively. Intraclass correlation coefficients were calculated to assess intraobserver agreement in measurements. Multiple linear regression analysis was used to assess the relationship between CTLV and other measurements of liver size as well as dog BW and age. Intraobserver correlation coefficients for duplicate measurements of RLA and CTLV were estimated by means of reliability analysis. Values of P < 0.05 were considered significant.
Results
Animals
A total of 107 dogs were selected for inclusion in the study, and their clinical histories and laboratory test results were evaluated. Thirty-six (34%) dogs were classified as having a radiographically and clinicopathologically normal liver (9 Maltese, 6 mixed-breed dogs, 4 Shih Tzus, 3 Beagles, 3 Poodles, 2 Chihuahuas, 2 Pomeranians, and 1 each of 7 other breeds), 34 as having microhepatia (11 Maltese, 7 Shih Tzus, 3 Pomeranians, 3 Poodles, 2 Chihuahuas, 2 mixed-breed dogs, and 1 each of 6 other breeds), and 37 as having hepatomegaly (11 Maltese, 6 Cocker Spaniels, 4 Schnauzers, 4 Shih Tzus, 4 Yorkshire Terriers, 2 mixed-breed dogs, and 1 each of 6 other breeds). Other characteristics of the dogs were summarized (Table 1).
Characteristics of dogs with radiographically and clinically normal livers, microhepatia, or hepatomegaly.
| Characteristic | Normal (n = 36) | Microhepatia (n = 34) | Hepatomegaly (n = 37) |
|---|---|---|---|
| 6.7 ± 4.8a | 4.6 ± 4.8a | 11.1 ± 2.6b | |
| Reproductive status | |||
| Castrated male | 8 (22) | 11 (32) | 21 (57) |
| Sexually intact male | 4 (11) | 5 (15) | 1 (3) |
| Spayed female | 6 (17) | 9 (26) | 13 (35) |
| Sexually intact female | 18 (50) | 9 (26) | 2 (5) |
| ≥ BW (kg) | 5.42 ± 3.09a | 3.63 ± 1.93b | 6.57 ± 3.04a |
Data are reported as mean ± SD (age and BW) or number (%) of all dogs in the group with the indicated reproductive status.
Values with different superscript letters differ significantly (P < 0.05).
The mean age of dogs in the hepatomegaly group was significantly (P < 0.001) greater than the mean age of dogs in the other 2 groups (Table 1). On the other hand, the mean BW of dogs in the microhepatia group was significantly lower than the mean BW of dogs in the normal (P = 0.02) and hepatomegaly (P < 0.001) groups.
Radiographic and CT liver measurements
Nonnormalized and normalized RLA values were summarized by group (Table 2). All mean RLA values for dogs in the microhepatia group were smaller than those for dogs in the normal group, and all mean normalized RLA values for dogs in the hepatomegaly group were larger than those for dogs in the normal group.
Mean ± SD values of radiographic and CT liver measurements for the dogs of Table 1.
| Measurement | Normal (n = 36) | Microhepatia (n = 34) | Hepatomegaly (n = 37) |
|---|---|---|---|
| RLA (cm2) | 35.58 ± 17.07a | 18.21 ± 12.77b | 55.05 ± 18.96c |
| RLA:T11 area ratio | 32.7 ± 6.2a | 20.4 ± 7.4b | 51.0 ± 12.9c |
| RLA:BW ratio | 7.0 ± 1.4a | 4.9 ± 2.6b | 9.1 ± 2.4c |
| RLL:T11 length ratio | 5.9 ± 1.0a | 4.9 ± 1.0b | 8.2 ± 1.5c |
| RLV (cm3) | 159.69 ± 96.00a | 89.81 ± 68.42b | 219.57 ± 110.81c |
| RLV:BW ratio | 27.5 ± 5.6a | 22.8 ± 9.4a | 34.4 ± 9.4b |
| CTLV (cm3) | 167.33 ± 111.88a | 97.15 ± 88.48b | 270.56 ± 124.04 |
| CTLV:BW ratio | 30.3 ± 6.1a | 23.1 ± 14.5b | 43.9 ± 15.5c |
| TD:TW ratio | 0.78 ± 0.08a | 0.74 ± 0.07a | 0.74 ± 0.08a |
See Table 1 for key.
Mean RLL:T11 length ratio also differed significantly (P < 0.001) among the 3 groups. The mean RLV for dogs in the normal group differed significantly from that for dogs in the microhepatia (P = 0.03) and hepatomegaly (P = 0.004) groups, but no significant (P = 0.08) difference in mean normalized RL:BW ratio was identified between dogs in the normal and microhepatia groups. Mean TD:TW ratio did not differ significantly among groups.
Mean CTLV for dogs in the normal group was significantly different (P ≤ 0.02) from the mean value for dogs in the other groups (Table 2). Mean CTLV normalized to BW also differed among the groups.
Correlations between liver size and other variables
The RLA was correlated with BW (r = 0.91, 0.79, and 0.76 in the normal, microhepatia and hepatomegaly groups, respectively; P < 0.001) and CTLV (r = 0.89, 0.96, and 0.91, respectively; P < 0.001). For dogs in the normal group, CTLV was not correlated with age or sex but was strongly and positively correlated with BW (r = 0.93; P < 0.001). For all dogs, CTLV was strongly correlated with RLA (r = 0.94), RLL (r = 0.87), RLV (r = 0.91), BW (r = 0.82), and T11 area (r = 0.75; P < 0.001 for all). The correlation of CTLV with age was moderate (r = 0.46; P < 0.001). The CTLV:BW ratio was strongly correlated with the RLV:BW ratio (r = 0.83) and RLA:BW ratio (r = 0.83) and moderately correlated with the RLA:T11 area ratio (r = 0.68) and RLL:T11 length ratio (r = 0.58; P < 0.001 for all).
Results of multiple linear regression analysis indicated that RLA was strongly (r2 = 0.94; P < 0.001) correlated with CTLV in all dogs. The following equation was obtained from this analysis:
Intraobserver agreement in liver size measurements
Calculated ICC values (95% confidence intervals) indicated high intraobserver agreement in duplicate (2 weeks apart) RLA and CTLV measurements for dogs in the normal group (0.861 [0.765 to 0.918] and 0.948 [0.912 to 0.970], respectively), microhepatia group (0.986 [0.972 to 0.993] and 0.969 [0.937 to 0.984], respectively), and hepatomegaly group (0.933 [0.869 to 0.965] and 0.989 [0.980 to 0.995], respectively; P < 0.001 for all). The RLA measurements were generally of lower reproducibility than were the CTLV measurements, except for dogs in the microhepatia group. The ICCs for measurements involving dogs in the normal group were consistently lower than those for dogs in the microhepatia and hepatomegaly groups.
Discussion
The present study provided the first evidence of a relationship between CTLV and radiographic liver size in dogs. Liver area measurement on right lateral abdominal radiographic images was shown to be a feasible method for estimation of liver size in dogs. Significant differences were identified in radiographic measurements among dogs with normal livers, microhepatia, or hepatomegaly, and a high correlation was identified between RLA measurements and CTLV that exceeded the correlations with CTLV observed for other approaches to liver size measurement, such as RLL and RLV. Also, the high intraobserver agreement observed for duplicate radiographic measurements indicated excellent reproducibility of many measurements (ICCs, 0.861 to 0.986). In addition, the RLA:BW ratio and RLA:T11 area ratio were determined for dogs with normal livers.
The CTLV:BW ratio was more strongly correlated with the RLA:BW ratio (r = 0.83) than with the RLA:T11 area ratio (r = 0.68), suggesting that the RLA:BW ratio allowed more accurate evaluation of liver size. This difference in strengths of correlation was likely because CTLV was more strongly correlated with BW (r = 0.82) than with T11 area (r = 0.75), which supported the supposition that optimal normalization can be achieved using BW. We chose BW and T11 area for RLA normalization because a previous study5 of radiographic liver size assessment in dogs showed the usefulness of normalizing data to indices based on T11, which is easily identifiable on radiographic images, and because almost all thoracic and abdominal radiographs include T11 and BW is routinely measured in clinical practice.
In the present study, RLA, RLL, and RLV were strongly correlated with CTLV. Notably, RLA had the highest correlation with CTLV (r = 0.94), which may be explained by the RLA reflecting a larger portion of the liver silhouette than the other measurements. In elderly dogs, the liver silhouette on abdominal radiographs generally has sagging and caudal extension because of stretching or elongation of the triangular ligaments.9 However, no significant correlation was identified between liver size and age for dogs with normal livers in our study. Therefore, the liver silhouette is considered more important for quantitative assessment than for morphological assessment of liver size. In a dog with microhepatia, the lateral liver margin generally appears round or blunt owing to fibrotic or cirrhotic change,21 which may cause difficulty in pointing to the liver edge when measuring RLL with imaging software. Therefore, RLA measurement in such dogs would be an easier method than RLL measurement for assessing liver size on radiographic images.
Liver volume assessment by use of CT images is the most accurate in vivo technique for this purpose and is regarded as the reference standard in veterinary medicine.2,7 However, CT is typically performed with animals anesthetized, and volumetric analysis of CT data requires the use of specific software. To calculate CTLV, the ROI needs to be defined on each transverse CT image, which may be time-consuming and labor-intensive.22 In contrast, RLA measurements can be obtained within a few seconds by use of the DICOM viewer without additional image processing or the need for a 3-D workstation. Such measurements are also easily comparable between serially obtained images of the same patient. In a dog with a PSS, hepatic growth after shunt attenuation is used as an indicator of prognosis,7 so such repeated measurements would have considerable diagnostic value. The RLA is well suited for rapid, easy measurement of liver size in such dogs, without the anesthesia necessary for CT scanning.
In the study reported here, values for RLL:T11 length ratio for dogs in the normal group (mean ± SD, 5.9 ± 1.0) were consistent with those in a previous study18 (5.5 ± 0.8). However, values for CTLV:BW ratio in this group (mean ± SD, 30.3 ± 6.1 cm3/kg) differed from those previously reported for dogs with a normal liver (24.4 ± 5.6 cm3/kg).2 This difference between studies may have been attributable to the difference in sample size (36 vs 6 dogs, respectively).
The present study had several limitations. First, dogs were deemed to have normal livers in part on the basis of CBC and serum biochemical results, and no histologic examination of the liver was performed. Second, RLA measurements were made solely with dogs positioned in right lateral recumbency, and it has been suggested that in this position, the left liver lobe can move caudally, causing the liver to cast a larger shadow than in left lateral recumbency.3 Third, although intraobserver repeatability was assessed and deemed strong, interobserver variability was not assessed. Fourth, it was difficult to accurately delineate the border of the liver in dogs when the stomach was collapsed or filled with ingesta. However, repeated adjustment of the gray contrast when using the DICOM viewer helped clarify definition of the liver border. Lastly, we included only small-breed dogs and therefore additional research is needed involving deep-chested or dolichocephalic breeds to address the challenges of defining the liver margins that arise from differences in chest conformation.
In conclusion, RLA measurements in the present study correlated more strongly with CTLV than did other types of radiographic liver size measurements in dogs. Values for RLA:BW and RLA:T11 area ratios were also obtained for dogs with normal livers that might be useful as reference values. Use of the RLA approach described here could allow veterinarians to evaluate liver size more accurately and easily than with other measurement approaches, particularly in dogs with a PSS or cirrhosis, and could provide important prognostic information on alterations in liver size.
Acknowledgments
This paper was supported by Konkuk University in 2017. The authors declare that there were no conflicts of interest.
ABBREVIATIONS
| BW | Body weight |
| CTLV | CT-measured liver volume |
| ICC | Intraobserver correlation coefficient |
| PSS | Portosystemic shunt |
| RLA | Radiographic liver area |
| RLL | Radiographic liver length |
| RLV | Radiographic liver volume |
| ROI | Region of interest |
| TD | Thoracic depth |
| TW | Thoracic width |
Footnotes
Titan 2000V, Comed Medical System, Seoul, South Korea.
LightSpeed, GE Healthcare, Milwaukee, Wis.
Omnipaque 300, GE Healthcare, Shanghai, China.
OsiriX Lite, version 9.5.2, Pixmeo, Bernex, Switzerland.
SPSS, version 24, IBM Corp, Armonk, NY.
References
1. Sekiyama K, Yoshiba M, Inoue K, et al. Prognostic value of hepatic volumetry in fulminant hepatic failure. Dig Dis Sci 1994;39:240–244.
2. Stieger SM, Zwingenberger A, Pollard RE, et al. Hepatic volume estimation using quantitative computed tomography in dogs with portosystemic shunts. Vet Radiol Ultrasound 2007;48:409–413.
3. Suter PF. Radiographic diagnosis of liver disease in dogs and cats. Vet Clin North Am Small Anim Pract 1982;12:153–173.
4. Barr F. Ultrasonographic assessment of liver size in the dog. J Small Anim Pract 1992;33:359–364.
5. Choi J, Keh S, Kim H, et al. Radiographic liver size in Pekingese dogs versus other dog breeds. Vet Radiol Ultrasound 2013;54:103–106.
6. Godshalk CP, Twardock AR, Kneller SK. Nuclear scintigraphic assessment of liver size in clinically normal dogs. Am J Vet Res 1989;50:645–650.
7. Kummeling A, Vrakking DJE, Rothuizen J, et al. Hepatic volume measurements in dogs with extrahepatic congenital portosystemic shunts before and after surgical attenuation. J Vet Intern Med 2010;24:114–119.
8. Partington BP, Biller DS. Hepatic imaging with radiology and ultrasound. Vet Clin North Am Small Anim Pract 1995;25:305–335.
9. Larson MM. Liver and spleen. In: Thrall DE, ed. Textbook of veterinary diagnostic radiology. 7th ed. St Louis: Elsevier Saunders, 2018;792–822.
10. Wrigley RH. Radiographic and ultrasonographic diagnosis of liver diseases in dogs and cats. Vet Clin North Am Small Anim Pract 1985;15:21–38.
11. Cockett PA. Radiographic anatomy of the canine liver: simple measurements determined from the lateral radiograph. J Small Anim Pract 1986;27:577–589.
12. van Bree H, Sackx A. Evaluation of radiographic liver size in twenty-seven normal deep-chested dogs. J Small Anim Pract 1987;28:693–703.
13. Godshalk CP, Kneller SK, Badertscher RR II, et al. Quantitative noninvasive assessment of liver size in clinically normal dogs. Am J Vet Res 1990;51:1421–1426.
14. Sharma M, Singh A, Goel S, et al. Assessment of liver volume with spiral computerized tomography scanning: predicting liver volume by age and height. Int J Res Med Sci 2016;4:3020–3023.
15. Moss AA, Friedman MA, Brito AC. Determination of liver, kidney, and spleen volumes by computed tomography: an experimental study in dogs. J Comput Assist Tomogr 1981;5:12–14.
16. Geraghty EM, Boone JM, McGahan JP, et al. Normal organ volume assessment from abdominal CT. Abdom Imaging 2004;29:482–490.
17. Kayaalp C, Arda K, Oto A, et al. Liver volume measurement by spiral CT an in vitro study. Clin Imaging 2002;26:122–124.
18. Jepsen-Grant K, Pollard RE, Johnson LR. Vertebral heart scores in eight dog breeds. Vet Radiol Ultrasound 2013;54:3–8.
19. van Bree H, Jacobs V, Vandekerckhove P. Radiographic assessment of liver volume in dogs. Am J Vet Res 1989;50:1613–1615.
20. Gotra A, Sivakumaran L, Chartrand G, et al. Liver segmentation: indications, techniques and future directions. Insights Imaging 2017;8:377–392.
21. Willard MD, Twedt DC. Gastrointestinal, pancreatic, and hepatic disorders. In: Willard MD, Tvedten H, eds. Small animal clinical diagnosis by laboratory methods. 5th ed. St Louis: Elsevier Saunders, 2011;209.
22. Verma SK, McClure K, Mitchell DG, et al. Simple linear measurements of the normal liver: interobserver agreement and correlation with hepatic volume on MRI. Clin Radiol 2010;65:315–318.
