Chongwoo Chu1, Xiaopeng Yang2, Jaedo Yang3, Younggeun Choi2, Jeho Ryu1, Kwangho Yang1,Youngmok Park1, Hee Chul Yu3, Baik Hwan Cho3, and Heecheon You2

Approach• Preoperative GV measurement by Dr. Liver

(Humanopia, Inc., South Korea)1. GVw_vein measurement

2. GVw/o_vein measurement

Introduction Background

• Estimation of graft weight (GW) is important to both donor and recipient for safe and successfulliver surgery

• Existing regression models for GW estimation from preoperatively calculated graft volume with vein (GVw_vein)

• Limitations of the existing models1. Using GV with vein to estimate GW without

vein2. Small sample size (16) of Lemke et al.’s model3. Low adjusted R2 value (0.52) of Yoneyama et

al.’s model4. Not validated5. Not for Korean population

Materials & Methods

Discussion The proposed regression model by GVw/o_vein

showed superior to existing models in accuracy Surgeons need to spend 2~3 minutes more on GW

estimation using the proposed model than existing models due to veins extraction. Further validation using data from more medical

centers is needed.

Recipient DonorToo small graft ⇒

Posttransplantcomplicationssuch as small-for-size syndrome

Too much loss of liver tissue⇒

High risk of postoperative liver failure

Authors Regression Models

AdjustedR2

SampleNationality Size Age

Lemke et al. (2006)

GW = 143.704 + 0.678 × GVw_vein

0.76 Germany 16 45.4±14.1

Yoneyamaet al. (2011)

GW = 0.84 ×GVw_vein

0.52 Japan 39 N.A.

CT Data GW Measured in surgery

GVw/o_veinCalculated

pre-operatively

GW Estimation

Model

Regression analysis

LVw_vein

Veins

GW = 74.7 + 0.773 × GVw_vein (Adj. R2 = 0.87)

GW = 29.1 + 0.943 × GVw/o_vein (Adj. R2 = 0.94)

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

PAE > 10%: 15%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

SyngoVia by GVw_vein

Lemke et al. (2006)'s by GVw_vein

Yoneyama et al. (2011)'s by GVw_vein

Proposed by GVw_vein Proposed by GVw_vein

Proposed by GVw/o_vein

PAE > 10%: 31%

PAE > 10%: 10%

PAE > 10%: 10%

PAE > 10%: 0%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

PAE > 10%: 17%

Lemke et al. (2006)'s by GVw_vein

Yoneyama et al. (2011)'s by GVw_vein

PAE > 10%: 13%

Proposed by GVw/o_vein

PAE > 10%: 26%

PAE > 10%: 0%

Methods

PNUYH Data (n = 20) CBNUH Data (n =23)

Absolute Error (AE; g)

Percentage of AE (PAE)

Percen-tage of PAE > 10%

AE(g) PAE Percen-

tage of PAE > 10%Mean

(SD) SE Mean(SD) SE Mean

(SD) SE Mean(SD) SE

Lemke et al. (2006) by GVw_vein

36.6(22.5) 5.0 5.8%

(4.1%) 0.9% 15.0% 45.5(32.9) 6.9 6.1%

(4.1%) 0.9% 17.4%

Yoneyama et al. (2011) by GVw_vein

34.2(26.4) 5.9 5.0%

(3.5%) 0.8% 10.0% 42.0(36.8) 7.7 5.7%

(4.9%) 1.0% 13.0%

SyngoVia 52.5(49.7) 12.4 8.3%

(7.1%) 1.8% 31.3% - - - - -

Proposed by GVw_vein35.0

(25.1) 5.6 5.4%(3.9%) 0.9% 10.0% 41.8

(33.9) 7.1 5.8%(4.6%) 1.0% 26.1%

Proposed by GVw/o_vein16.3

(12.6) 2.8 2.6%(2.2%) 0.5% 0.0% 21.5

(16.5) 3.4 3.0%(2.3%) 0.5% 0.0%

Percentage of AE Percentage of PAE > 10%

Frequency distribution of PAE (left: PNUYH, right: CBNUH)

Results

Objectives of the Study1. Develop a regression model using graft volume

without vein (GVw/o_vein) for GW estimation2. Cross validate the proposed model

S1. Liver segmentation from CT and liver volume with vein (LVw_vein) measurement

S2. Liver surgery planning and GVw_vein calculation GVw_vein

LVw_vein

S1. Liver segmentation from CT and LVw_vein calculation

S2. Liver veins (portal and hepatic veins) segmentations from CT and their volumes calculation

S3. Exclude veins from the liver and calculate liver volume without vein (LVw/o_vein)

S4. Liver surgery planning and GVw/o_vein calculation

GVw/o_vein

LVw/o_vein

Data & Software for Regression• 40 cases (age: 29.7 ± 10.6 years) with CT images

and GW measured in surgery provided by Pusan National University Yangsan Hospital (PNUYH)

• Statistical software: Minitab version 14 (Minitab, Inc., USA)

Regression Models1. Proposed by GVw_vein

2. Proposed by GVw/o_vein

Cross Validation• Data

1. 20 cases (age: 24.1 ± 6.9 years) with CT images and GW measured in surgery provided by PNUYH

2. 23 cases (age: 29.6 ± 10.8 years) provided by Chonbuk National University Hospital (CBNUH)

• Comparison to Lemke et al.’s and Yoneyama et al.’s models and SyngoVia method (GW = 0.9 ×GVw_vein manually measured by SyngoVia:Siemens Co., Germany)

• Validation results

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

30.0%

35.0%

Lemke et al.'s(2006) byGVw_vein

Yoneyama etal.'s (2011) by

GVw_vein

Syngovia byGVw_vein

Proposed byGVw_vein

Proposed byGVw/o_vein

Perc

enta

ge o

f PA

EE >

10%

PNUYH

CBNUH

0%0.0%

2.0%

4.0%

6.0%

8.0%

10.0%

12.0%

Lemke et al.'s(2006) byGVw_vein

Yoneyama etal.'s (2011) by

GVw_vein

Syngovia byGVw_vein

Proposed byGVw_vein

Proposed byGVw/o_vein

S.E.

% o

f abs

olut

e es

timat

ion

erro

r

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

PAE > 10%: 15%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

SyngoVia by GVw_vein

Lemke et al. (2006)'s by GVw_vein

Yoneyama et al. (2011)'s by GVw_vein

Proposed by GVw_vein

Proposed by GVw/o_vein Proposed by GVw/o_vein

PAE > 10%: 31%

PAE > 10%: 10%

PAE > 10%: 10%

PAE > 10%: 0%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

PAE > 10%: 17%

Lemke et al. (2006)'s by GVw_vein

Yoneyama et al. (2011)'s by GVw_vein

PAE > 10%: 13%

Proposed by GVw_vein

PAE > 10%: 26%

PAE > 10%: 0%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

PAE > 10%: 15%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

SyngoVia by GVw_vein

Lemke et al. (2006)'s by GVw_vein

Yoneyama et al. (2011)'s by GVw_vein

Proposed by GVw_vein

Proposed by GVw/o_vein

PAE > 10%: 31%

PAE > 10%: 10%

PAE > 10%: 10%

PAE > 10%: 0%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

PAE > 10%: 17%

Lemke et al. (2006)'s by GVw_vein

Yoneyama et al. (2011)'s by GVw_vein

PAE > 10%: 13%

Proposed by GVw_vein

Proposed by GVw/o_vein

PAE > 10%: 26%

PAE > 10%: 0%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

PAE > 10%: 15%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

SyngoVia by GVw_vein

Lemke et al. (2006)'s by GVw_vein

Yoneyama et al. (2011)’s by GVw_vein Yoneyama et al. (2011)’s by GVw_vein

Proposed by GVw_vein

Proposed by GVw/o_vein

PAE > 10%: 31%

PAE > 10%: 10%

PAE > 10%: 10%

PAE > 10%: 0%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

PAE > 10%: 17%

Lemke et al. (2006)'s by GVw_vein

PAE > 10%: 13%

Proposed by GVw_vein

Proposed by GVw/o_vein

PAE > 10%: 26%

PAE > 10%: 0%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

PAE > 10%: 15%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

SyngoVia by GVw_vein

Lemke et al. (2006)’s by GVw_vein Lemke et al. (2006)’s by GVw_vein

Yoneyama et al. (2011)'s by GVw_vein

Proposed by GVw_vein

Proposed by GVw/o_vein

PAE > 10%: 31%

PAE > 10%: 10%

PAE > 10%: 10%

PAE > 10%: 0%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

0

2

4

6

8

10

12

14

2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 22% 24% 26%

PAE > 10%: 17%

Yoneyama et al. (2011)'s by GVw_vein

PAE > 10%: 13%

Proposed by GVw_vein

Proposed by GVw/o_vein

PAE > 10%: 26%

PAE > 10%: 0%

3Chonbuk National UniversityMedical School, Jeonju,561-712, South Korea

2Pohang University of Scienceand Technology, Pohang,790-784, South Korea

1Pusan National UniversityYangsan Hospital, Yangsan,626-770, South Korea

Development of a Graft Weight Prediction Model Using Preoperative CTVolumetric Measurement for Living Donor Liver Transplantation