Published online before print July 20, 2006, 10.1148/radiol.2403050850
(Radiology 2006;240:743-748.)
© RSNA, 2006
Automated Hepatic Volumetry for Living Related Liver Transplantation At Multisection CT1
Yoshiharu Nakayama, MD,
Qiang Li, PhD,
Shigehiko Katsuragawa, PhD,
Ryuji Ikeda, RT,
Yasuhiro Hiai, RT,
Kazuo Awai, MD,
Shinichiro Kusunoki, MD,
Yasuyuki Yamashita, MD,
Hideaki Okajima, MD,
Yukihiro Inomata, MD and
Kunio Doi, PhD
1 From the Departments of Diagnostic Radiology (Y.N., K.A., S. Kusunoki, Y.Y.) and Transplantation/Pediatric Surgery (H.O., Y.I.), Graduate School of Medical Sciences, Kumamoto University, 1-1-1 Honjo, Kumamoto City, Kumamoto, 860-8556, Japan; Department of Radiology, the Kurt Rossmann Laboratories for Radiologic Image Research, University of Chicago, Chicago, Ill (Q.L., K.D.); Department of Radiological Technology, School of Health Sciences, Kumamoto University, Kumamoto, Japan (S. Katsuragawa); Department of Radiology, Kumamoto University Hospital, Kumamoto, Japan (R.I., Y.H.). Received May 19, 2005; revision requested July 18; revision received September 4; accepted September 23; final version accepted, November 14. Supported in part by USPHS grants CA 62625 and CA 64370.
Address correspondence to Y.N. (e-mail: yosiharu156{at}lily.ocn.ne.jp).
 |
ABSTRACT
|
|---|
Purpose: To prospectively compare in vivo hepatic automated volumetry with manual volumetry and measured liver volume.
Materials and Methods: The study was conducted in accordance with the guidelines of the Institutional Review Board of Kumamoto University (Japan). Patient informed consent was obtained. Preoperative multisection computed tomography (CT) was performed in 35 consecutive patients (21 men, 14 women; mean age, 42.8 years; range, 28–72 years) with hepatic disease awaiting living related liver transplantation. The CT scans covered the entire liver at a section thickness of 2.5 mm. Liver volume was estimated by using both the automated and the manual methods. Actual liver weight was obtained for all patients and was converted to hepatic volume on the basis of a predetermined relationship between actual liver weight and volume. Processing time required for both methods was also recorded. Two-tailed paired t test, correlation coefficient, and Bland-Altman tests were used for statistical analyses.
Results: Mean liver weight was 881.7 g ± 249.8 (standard deviation), and mean measured liver volume was 956.00 cm3 ± 280.10. Volumetry performed with the automated and manual methods provided liver volumes of 982.99 cm3 ± 301.98 and 937.10 cm3 ± 301.31, respectively. There was good correlation between measured and estimated volumes obtained with the automated method (r = 0.792, P < .01). The manual and automated methods required 32.8 minutes ± 6.9 and 4.4 minutes ± 1.9, respectively.
Conclusion: The automated method reduced the time required for volumetry of the liver and provided acceptable measurements.
© RSNA, 2006
|
INTRODUCTION
|
|---|
The size of the liver is considered to be an important prognostic factor in patients with cirrhosis (1) or fulminant hepatic failure (2,3), and imaging techniques have been used for obtaining quantitative measurements of liver volume. In patients scheduled for liver surgery for primary hepatic tumor (4,5), metastatic lesions (6), and transplantation (7), the liver volume must be known preoperatively. The liver volume is one of the most important factors in the selection of appropriate donors, especially in a patient undergoing living related liver transplantation (LRLT). Volumetry of the hepatic graft and remnant is mandatory for LRLT and is usually performed with cross-sectional computed tomography (CT) or magnetic resonance (MR) imaging. These methods yield reliable organ volume measurements when appropriate scanning protocols are used (7–9). Volumetry of the liver on CT images is usually performed by manual tracing of the liver boundary and summation of the liver area on each section. However, manual methods require considerable user involvement in the segmentation of the liver on each section, which is a time-consuming process.
To our knowledge, the reliability of in vivo automated volumetric measurements of the liver has not been compared with the actual volume of resected livers. We have developed a fully automated hepatic volumetric method by using multisection CT. Thus, the purpose of our study was to prospectively compare the in vivo hepatic automated volumetry we developed with manual volumetry and measured liver volume.
|
MATERIALS AND METHODS
|
|---|
Patient Population
This study was conducted in accordance with the guidelines of the Institutional Review Board of Kumamoto University. Prior written informed consent was obtained from all patients.
Between November 1999 and September 2004, 35 consecutive patients (mean age, 42.8 years; range, 28–72 years) who underwent LRLT at our facility were included in this study. There were 14 women (mean age, 39.6 years; range, 28–56 years) and 21 men (mean age, 48.8 years; range, 28–72 years). The underlying liver diseases were liver cirrhosis (n = 21), familial amyloidotic polyneuropathy (n = 9), and fulminant hepatic failure (n = 5). All patients underwent a preoperative three-phase enhanced abdominal CT study within 3 months before LRLT. All recipients' livers excised at the time of liver transplantation and drained of all blood were weighed directly after the gallbladder, portal structures, attachment ligaments, and other extraneous tissues had been dissected free.
CT Protocol
CT was performed with a four-section scanner (LightSpeed QXi; GE Medical Systems, Milwaukee, Wis) in 20 patients and a 16-section scanner (IDT16, Philips Medical Systems, Cleveland, Ohio) in the remaining 15 patients. For patients with liver cirrhosis and fulminant hepatic failure, we used a multisection CT protocol to acquire three image sets of the liver (arterial, hepatic parenchymal, and equilibrium phases). Patients with familial amyloidotic polyneuropathy underwent scanning in the arterial, portal venous, and hepatic parenchymal phases to obtain three-dimensional (3D) images of the hepatic artery and the portal and hepatic veins. Because a patient with familial amyloidotic polyneuropathy is usually considered a domino (second) donor for a second recipient (10), the additional portal venous images were obtained to generate a 3D image of the portal vein. The scans for these phases were acquired at 25–35 seconds (arterial), 45–65 seconds (portal venous), 80–100 seconds (hepatic parenchymal), and at more than 200 seconds (equilibrium) after the administration of 100 mL of nonionic contrast material iopamidol (Iopamiron 370; Nihon Schering, Osaka, Japan) at a rate of 3.0 mL/sec through a 20-gauge catheter placed in an antecubital vein. Hepatic parenchymal phase data were used in this study.
With the four-section scanner, the scanning parameters for the portal venous phase were 120 kVp, 200–250 mAs, 0.8-second rotation time, four sections at 2.5-mm beam collimation, 15-mm table feed per rotation, 1.5 beam pitch, 36-cm field of view, and 512 x 512-pixel matrix. With the 16-section scanner, the scanning parameters were 120 kVp, 200–300 mAs, 0.75-second rotation time, 16 sections at 2.5-mm beam collimation, 21.1-mm table feed per rotation, 0.659 beam pitch, 36-cm field of view, and 512 x 512-pixel matrix. A section thickness of 5 mm (without overlaps) and 2.5 mm (with 1.25-mm overlaps) was used for manual and automated volumetry, respectively.
In Vitro Experiment
Prior to the in vitro study, a power analysis was performed to determine the study size. Because the sample size was calculated as 6.62 with the assumption of a significance level of .05, a power of .8, and a correlation coefficient of 0.9, we designed the sample size as 7.
To determine the relationship between liver weight and liver volume, we obtained the volume of the resected livers from another seven transplant recipients (three men and four women; age range, 32–61 years; mean, 45.3 years). The underlying liver diseases of those recipients were liver cirrhosis (n = 5) and hepatocellular carcinoma (n = 2). One radiologist (S. Kusunoki) with 5 years of abdominal CT experience directly measured the actual volumes of the excised livers by means of water displacement in a water bath filled with distilled water at 25°C. Immediately after volume measurement, the excised livers were placed on a platform scale, and the exact liver weights were measured. The excised livers were drained of all blood before water displacement and weight measurements. Finally, a regression line was obtained for the relationship between the weight and the volume of the liver by using the measured data of the seven livers.
Manual Method of Liver Volumetry
For imaging prior to LRLT, one radiologist (Y.N.) with 10 years of abdominal CT experience manually traced the contours of all liver sections on a Digital Imaging and Communications in Medicine viewer (spatial resolution of 1600 x 1200 pixels, RadiForce R22; Nanao, Ishikawa, Japan) by using an electronic cursor and recorded the time required for these measurements. The manufacturer's software automatically calculated the number of pixels included within the traced contours on each section and provided the cross-sectional area of the liver on a section-by-section basis. The circumscribed areas were then multiplied by the CT section thickness, yielding an approximate volume for each liver section, and the volumes of all sections were summed to give the total volume of the liver.
Automated Method of Liver Volumetry
CAD technologies developed in the Kurt Rossmann Laboratories, Department of Radiology, the University of Chicago have been licensed to companies including R2 Technology, Deus Technologies, Riverain Medical Group, Mitsubishi Space Software Median Technologies, GE, and Toshiba.
A radiologic technologist (R.I.) with 10 years of CT experience recorded the time used for the measurement of liver volume. The computerized method of liver volumetry used in this study was based on automated liver segmentation on multisection CT images. The overall scheme for automated volumetry is illustrated in Figure 1. There is no human input throughout the process. Our segmentation technique initially estimated the mean CT number of the liver by means of an analysis of CT numbers in a 3D volume of interest with a 32 x 32-pixel in-section matrix and an automatically determined height. The location of the in-section 32 x 32-pixel region was determined empirically at a location where the liver was likely to be included. The height of the 3D volume of interest was also automatically determined so that 32 x 32-pixel regions of the volume of interest were completely located inside liver. Next, the pixels in the range of the estimated mean CT number –30 and +30 HU were retained to form initial liver regions. A Sobel operator was then used for detecting the edge inside the initial liver regions to separate the liver from the other adjacent organs, such as the pancreas and spleen, that may have similar CT numbers as the liver. Many regions that included those organs were removed by means of feature analysis of CT numbers and locations. A 3D connected-component labeling technique was then used for selecting the 3D object that had the maximum volume among all objects, which further eliminated false liver regions. The selected 3D object was refined by using a restricted region-growing technique. Finally, the liver volume was automatically calculated by summation of the products of the section thickness and area of the segmented liver in each section (Fig 2).
View larger version (114K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2a: Automated segmentation of the liver on transverse CT images. (a) CT scan shows 3D volume of interest ( ) with 32 x 32-pixel in-section matrix and automatically determined height for estimation of the mean CT number. (b) The initial candidate liver regions. (c) Selection of 3D object with the maximum volume among all objects by using 3D connected-component labeling technique.
|
|
View larger version (16K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2b: Automated segmentation of the liver on transverse CT images. (a) CT scan shows 3D volume of interest () with 32 x 32-pixel in-section matrix and automatically determined height for estimation of the mean CT number. (b) The initial candidate liver regions. (c) Selection of 3D object with the maximum volume among all objects by using 3D connected-component labeling technique.
|
|
View larger version (8K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2c: Automated segmentation of the liver on transverse CT images. (a) CT scan shows 3D volume of interest () with 32 x 32-pixel in-section matrix and automatically determined height for estimation of the mean CT number. (b) The initial candidate liver regions. (c) Selection of 3D object with the maximum volume among all objects by using 3D connected-component labeling technique.
|
|
Determination of Measured Liver Volume
Determination of the measured liver volume was performed by a radiologist (Y.N.). The 35 patients underwent LRLT 2 days to 3 months (mean, 72.5 days; median, 75 days) after undergoing a preoperative CT examination. After complete resection, the entire liver was weighed and its volume was determined on the basis of the predetermined relationship between liver weight and volume. The entire liver volumes estimated by using the manual and automated methods were compared with the measured liver volume. The error (Er) between the estimated volume (Ve) and the measured volume (Vm) was defined as Er = (Ve – Vm)/Vm.
Statistical Analysis
A regression line between the estimated volume and the manual volume of the liver was generated and the correlation coefficient (r) was calculated. In addition, a two-tailed paired t test was used to compare the automated and manual methods in terms of the time required to estimate the liver volumes. The average errors for the two methods were determined. The 95% confidence interval (CI) of the error was calculated for both methods. Differences between the average errors of the manual and of the automated methods were evaluated with the two-tailed paired Student t test. In addition, Bland-Altman analysis was used to determine the agreement between these methods (11,12). All statistical analyses were performed with a software program (MedCalc Software; MedCalc, Mariakerke, Belgium). A P value of less than .05 was considered to indicate a significant difference.
|
RESULTS
|
|---|
In Vitro Experiment
The weight and volume of the resected livers were measured in vitro to predetermine the relationship between the weight and volume (Fig 3). The mean actual weight and volume were 660.0 g ± 145.6 (± standard deviation) (range, 500–900 g) and 670.0 cm3 ± 174.1 (range, 480–970 cm3), respectively. There was a statistically significant positive correlation (r = 0.957, P < .01) between actual liver volume and weight. The regression line was determined by a forced through the origin equation as y = 1.06x, where x and y indicate the measured weight and converted volume, respectively (Fig 3). This regression line was used for converting liver weight to liver volume. We used the converted volume as the reference standard. The power of analysis indicated greater than .99 in the sample size of seven, a significance level of .05, and a correlation coefficient of 0.957.
Surgical Results
After LRLT, the liver weight ranged from 450 to 1595 g (mean, 881.7 g ± 249.8); it was 985.0 g ± 166.2 (range, 680–1150 g) in nine patients with familial amyloidotic polyneuropathy, 919.5 g ± 242.8 (range, 650–1595 g) in 21 patients with liver cirrhosis, and 550.0 g ± 77.4 (range, 450–640 g) in five patients with fulminant hepatic failure. The measured volume converted from the actual weight ranged from 511.3 to 1690.7 cm3 (mean, 956.0 cm3 ± 280.1).
Manual Liver Volume Estimation
The mean liver volume estimated with the manual method was 937.10 cm3 ± 301.3 (range, 386.2–1709.5 cm3) (Fig 4). The relationship between the volume estimated with the manual method and the reference standard showed a strong correlation (y = 1.01x, r = 0.899, P < .001). The average error between the volume estimated with the manual method and the reference standard was –0.024 ± 0.157 (95% CI: –0.080, 0.032). The time required for calculating the liver volume with the manual method was 19.0–46.5 minutes (mean, 32.8 minutes ± 6.9).
View larger version (11K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4: Scatterplot shows measured liver volume and liver volume estimated with the manual method. There is close correlation between manual volumetry and measured liver volume (y = 1.01x, r = 0.899, P < .001).
|
|
Automated Liver Volume Estimation
The liver volume estimated with the automated method was 982.99 cm3 ± 301.98 (range, 390.4–1700.1 cm3). There was a good correlation between the volume estimated with the automated method and the measured liver volume (y = 1.04x, r = 0.782, P < .001) (Fig 5). The average error between the volume obtained with the automated method and the measured liver volume was 0.033 ± 0.220 (95% CI: –0.047, 0.114). The 95% CI of the automated method was slightly larger than that of the manual method, but there was no significant difference between the two methods (P = .407). The correlation between the two methods (Fig 6) was excellent (y = 1.03x, r = 0.883, P < .001). The mean time required for computerized data acquisition was 4.4 minutes ± 1.9 (range, 3.0–7.0 minutes). Results are summarized in the Table.
View larger version (12K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5: Scatterplot shows measured liver volume and liver volume estimated with the automated method. There is good correlation between computerized volumetry and measured liver volume (y = 1.04x, r = 0.792, P < .001).
|
|
View larger version (13K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6: Scatterplot shows the liver volume estimated with manual and automated methods. There is a close correlation between automated and manual volumetry (y = 1.03x, r = 0.883, P < .001).
|
|
View this table:
[in this window]
[in a new window]
|
Comparison of Manual and Automated Methods with Respect to Volume, Relative Error, and Time Required for Data Acquisition
|
|
Measurement Agreement
The mean difference between the manual method and the measured liver volume was 2.5 cm3 (95% CI: –42.8 cm3, 47.9 cm3). The limits of agreement were –256.3 and 261.3 cm3. The mean difference between the automated method and the measured liver volume was 51.3 cm3 (95% CI: –11.9 cm3, 114.5 cm3). The limits of agreement were –309.3 and 412.0 cm3. The mean difference between the automated and the manual methods was 48.8 cm3 (95% CI: –0.1 cm3, 97.7 cm3). The limits of agreement were –230.3 and 327.9 cm3.
|
DISCUSSION
|
|---|
To simplify the determination of liver volume in donors scheduled for LRLT, we developed a fully automated method. We found that there was a good correlation between the actual measurements and the liver volume estimated with the automated method. The average error of the automated method (0.033 ± 0.220) was slightly larger than that of the manual method (–0.024 ± 0.157). However, the difference between the average errors of the automated and manual methods was not statistically significant (P = .407). Although the best agreement was confirmed between the manual method and the measured liver volume, the limits of agreement (–256.3 and 261.3 cm3) were not small. In the automated and manual method agreement, the difference tended to give a higher reading by 48.8 cm3; however, the limits of agreement (–230.3 and 327.9 cm3) were within the acceptable range.
One benefit of computerized volumetry is the speed with which the liver volume can be estimated. The section thickness and the amount of data represent the limiting factors in manually performed volumetry; the effect of partial volume renders thinner sections preferable to thicker sections. Because multisection CT provides useful 3D images with thinner sections, multisection CT data sets are larger than those obtained with nonhelical or single-detector helical CT, which renders manual tracing cumbersome and time consuming. The large data sets produced with multisection CT call for automated segmentation methods. In our study, the automated method was 7.5 times faster than the manual method. Our automated CT volumetry method can be used in the clinical setting, for example, in the follow-up of patients who had a partial hepatectomy or fulminant hepatitis.
Volume measurements obtained manually are relatively accurate (7,13–16), and attempts have been made to determine liver volume by using semiautomated (17) and automated (18) methods. However, previous studies did not compare the liver volume estimated with these methods with the actually measured liver volume. Hermoye et al (9), who compared the measurements obtained by using semiautomated method on MR sections with the graft liver volume, assumed that the density of the liver is 1.0 g/cm3. In our preliminary study, we determined the actual liver volume based on water displacement in a water bath. We found an excellent correlation between liver volume and weight, although the density of the liver was not equal to that of water. Because the acquisition of exact volume is cumbersome in LRLT, we obtained the measured liver volume by converting the weight to the volume by using the regression line predetermined in the preliminary study.
Previously reported segmental volume measurements were performed in LRLT donor livers (7,19,20). These studies compared the segmental volume of the liver in vivo and its weight. However, because the actual cutting line of the graft is determined by temporary clumping of the hepatic vessels, it may differ from the preoperative estimation, and use of CT volumetric data from graft studies appears inappropriate. In our comparative study, we measured livers resected from transplant recipients; this made possible a direct comparison between values obtained on the actual entire liver in vitro and values obtained with CT volumetry in vivo, and good correlation was demonstrated.
Other investigations (9,21) have used MR images to obtain volume measurements. MR imaging studies of patients scheduled for LRLT reveal the arterial, portal venous, hepatic parenchymal, and biliary anatomy. However, because the spatial resolution is lower on MR images than it is on CT scans, CT—including CT angiography—appears necessary, especially for evaluation of donors. CT volumetry usually uses the same data set as CT angiography; however, if MR volumetry is to be used, the donors must undergo an additional preoperative MR study.
There were limitations to our study. First, although our automated method carefully extracted candidate areas on each CT section, some measurement errors did occur. Because the areas were extracted on the basis of their CT number, the errors may be attributable to a partial volume effect at the liver edge, to adhesion of the adjacent tissue with attenuation similar to that of the liver parenchyma, and/or to exclusion of intrahepatic regions that have CT numbers different from those of the surrounding parenchyma (eg, cysts or tumors). Second, for our liver volume measurements, we used livers of patients who had a long-standing hepatic disease. Therefore, their livers may have been severely damaged and deformed. These patients frequently had markedly developed collateral vessels and a large amount of ascites. We found that automated tracing of the contours of severely damaged livers was more difficult than tracing in the healthy donor liver, and in such patients relative errors tended to be larger. Finally, our technique can be applied to only a total liver volume determination. However, segmental liver volume, which is donated to a recipient, is not available. At the moment, automated segmental volumetry is difficult because the cutting-line determination is extremely complicated. We consider segmental volumetry with automated measurement as our next challenge.
In conclusion, our comparative study revealed that automated CT volumetric assessment of the liver volume in vivo yielded acceptable measurements when compared with data obtained from the resected liver. We found the automated method to be quicker than the manual method, and we suggest that automated hepatic CT volumetry may be useful for determination of total liver volume in the donor candidate scheduled to undergo LRLT and for monitoring of postoperative liver regeneration.
|
ADVANCES IN KNOWLEDGE
|
|---|
- Automated CT volumetry of the liver yields acceptable measurements when compared with data obtained from the resected liver.
- The automated method is quicker than the manual method.
|
ACKNOWLEDGMENTS
|
|---|
We are grateful to E. Lanzl, MA, for improving the manuscript.
|
FOOTNOTES
|
|---|
Abbreviations: CI = confidence interval • LRLT = living related liver transplantation • 3D = three-dimensional
S. Katsuragawa and K.D. are shareholders in R2 Technology, Los Altos, Calif.
Author contributions: Guarantors of integrity of entire study, Y.N., Y.Y., K.D.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, Y.N., Q.L., Y.H., S. Kusunoki; clinical studies, Y.N., R.I., S. Kusunoki, H.O., Y.I.; experimental studies, Y.N., Q.L., S. Katsuragawa, R.I.; statistical analysis, Y.N., S. Katsuragawa; and manuscript editing, Y.N., Q.L., S. Katsuragawa, K.A., Y.Y., K.D.
|
References
|
|---|
- Zoli M, Cordiani MR, Marchesini G, et al. Prognostic indicators in compensated cirrhosis. Am J Gastroenterol 1991;86:1508–1513.[Medline]
- Itai Y, Sekiyama K, Ahmadi T, Obuchi M, Yoshiba M. Fulminant hepatic failure: observation with serial CT. Radiology 1997;202:379–382.[Abstract/Free Full Text]
- Sekiyama K, Yoshiba M, Inoue K, Sugata F. Prognostic value of hepatic volumetry in fulminant hepatic failure. Dig Dis Sci 1994;39:240–244.[CrossRef][Medline]
- Okamoto E, Yamanaka N, Oriyama T, Tomoda F, Kyo A. Prediction of the safe limits of hepatectomy by combined volumetric and functional measurements in patients with impaired hepatic function. Cancer Treat Res 1994;69:293–299.[Medline]
- Rollo FD, DeLand FH. The determination of liver mass from radionuclide images. Radiology 1968;91:1191–1194.[Medline]
- Soyer P, Roche A, Elias D, Levesque M. Hepatic metastases from colorectal cancer: influence of hepatic volumetric analysis on surgical decision making. Radiology 1992;184:695–697.[Abstract/Free Full Text]
- Kawasaki S, Makuuchi M, Matsunami H, et al. Preoperative measurement of segmental liver volume of donors for living related liver transplantation. Hepatology 1993;18:1115–1120.[CrossRef][Medline]
- Chen YS, Cheng YF, De Villa VH, et al. Evaluation of living liver donors. Transplantation 2003;75(suppl 3):S16–S19.[CrossRef][Medline]
- Hermoye L, Laamari-Azjal I, Cao Z, et al. Liver segmentation in living liver transplant donors: comparison of semiautomatic and manual methods. Radiology 2005;234:171–178.[Abstract/Free Full Text]
- Nishizaki T, Kishikawa K, Yoshizumi T, et al. Domino liver transplantation from a living related donor. Transplantation 2000;70:1236–1239.[Medline]
- Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307–310.[CrossRef][Medline]
- Bland JM, Altman DG. Measuring agreement in method comparison studies. Stat Methods Med Res 1999;8:135–160.[Abstract/Free Full Text]
- Heymsfield SB, Fulenwider T, Nordlinger B, Barlow R, Sones P, Kutner M. Accurate measurement of liver, kidney, and spleen volume and mass by computerized axial tomography. Ann Intern Med 1979;90:185–187.[CrossRef][Medline]
- Henderson JM, Heymsfield SB, Horowitz J, Kutner MH. Measurement of liver and spleen volume by computed tomography: assessment of reproducibility and changes found following a selective distal splenorenal shunt. Radiology 1981;141:525–527.[Abstract/Free Full Text]
- Urata K, Kawasaki S, Matsunami H, et al. Calculation of child and adult standard liver volume for liver transplantation. Hepatology 1995;21:1317–1321.[CrossRef][Medline]
- Van Thiel DH, Hagler NG, Schade RR, et al. In vivo hepatic volume determination using sonography and computed tomography: validation and a comparison of the two techniques. Gastroenterology 1985;88:1812–1817.[Medline]
- Farjo LA, Williams DM, Bland PH, Francis IR, Meyer CR. Determination of liver volume from CT scans using histogram cluster analysis. J Comput Assist Tomogr 1992;16:674–683.[Medline]
- Gao L, Heath DG, Kuszyk BS, Fishman EK. Automatic liver segmentation technique for three-dimensional visualization of CT data. Radiology 1996;201:359–364.[Abstract/Free Full Text]
- Schroeder T, Nadalin S, Stattaus J, Debatin JF, Malago M, Ruehm SG. Potential living liver donors: evaluation with an all-in-one protocol with multi–detector row CT. Radiology 2002;224:586–591.[Abstract/Free Full Text]
- Kamel IR, Kruskal JB, Warmbrand G, Goldberg SN, Pomfret EA, Raptopoulos V. Accuracy of volumetric measurements after virtual right hepatectomy in potential donors undergoing living adult liver transplantation. AJR Am J Roentgenol 2001;176:483–487.[Abstract/Free Full Text]
- Mazonakis M, Damilakis J, Maris T, Prassopoulos P, Gourtsoyiannis N. Comparison of two volumetric techniques for estimating liver volume using magnetic resonance imaging. J Magn Reson Imaging 2002;15:557–563.[CrossRef][Medline]
TOP
ABSTRACT
INTRODUCTION
