HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Kim, S. H. Articles by Choi, B. I. Search for Related Content PubMed PubMed Citation Articles by Kim, S. H. Articles by Choi, B. I.

Appropriateness of a Donor Liver with Respect to Macrosteatosis: Application of Artificial Neural Networks to US Images—Initial Experience¹

¹ From the Department of Radiology (S.H.K., J.M.L., J.H.K., J.K.H., K.H.L., S.H.P., S.K.A., Y.J.K., K.R.S., B.I.C.), Institute of Radiation Medicine (J.M.L., J.H.K., J.K.H., B.I.C.), Department of Medical Engineering (J.H.K., K.G.K.), Department of Surgery (N.J.Y., K.S.S.), and Department of Pathology (H.S.L.), Seoul National University College of Medicine, 28 Yongon-dong, Chongno-gu, Seoul 110-744, Korea; and Department of Radiology, Seoul National University Bundang Hospital, Seoul, Korea (K.H.L.). Received January 25, 2004; revision requested April 1; revision received May 12; accepted June 15. Address correspondence to J.M.L. (e-mail: leejm@radcom.snu.ac.kr).

View larger version (118K): [in a new window]   Figure 1a. Set of two US images in right upper abdomen was obtained from each patient to evaluate hepatic steatosis. (a) Coronal US image obtained to allow comparison between liver and right kidney. (b) Transverse US image selected to quantify conspicuity of portal vein border. Portal vein branch of segment VIII (arrow) is located between right (R) and middle (M) hepatic vein and is visualized as an anechoic structure with round configuration.

View larger version (118K): [in a new window]   Figure 1b. Set of two US images in right upper abdomen was obtained from each patient to evaluate hepatic steatosis. (a) Coronal US image obtained to allow comparison between liver and right kidney. (b) Transverse US image selected to quantify conspicuity of portal vein border. Portal vein branch of segment VIII (arrow) is located between right (R) and middle (M) hepatic vein and is visualized as an anechoic structure with round configuration.

View larger version (128K): [in a new window]   Figure 2a. Application of ROIs. (a) On coronal US image, two ROIs of 1 x 1 cm are located on hepatic parenchyma (left ROI) and renal cortex (right ROI) so as not to contain large vessels or renal sinus within the ROI. ROIs were selected along the focusing area, ensuring similar distances from the probe, and near the center line of the image to avoid distorting effects in ultrasonic wave patterns. (b) On transverse image, ROI of 2 x 2 cm was applied to fully include portal vein branch of segment VIII and adjacent structures.

View larger version (124K): [in a new window]   Figure 2b. Application of ROIs. (a) On coronal US image, two ROIs of 1 x 1 cm are located on hepatic parenchyma (left ROI) and renal cortex (right ROI) so as not to contain large vessels or renal sinus within the ROI. ROIs were selected along the focusing area, ensuring similar distances from the probe, and near the center line of the image to avoid distorting effects in ultrasonic wave patterns. (b) On transverse image, ROI of 2 x 2 cm was applied to fully include portal vein branch of segment VIII and adjacent structures.

View larger version (20K): [in a new window]   Figure 3a. Echo histograms of ROI applied to portal vein of segment VIII and adjacent structures in patients with normal and hepatic macrosteatosis of moderate degree proved at histopathologic examination. (a) In a normal case, right-sided area (tail area, deviant creases) of predetermined point (20% of value of highest point, dashed lines) is relatively larger than that in (b) a steatotic case. For example, in a and b, ratios for tail area of portal vein echogenicity were 0.138 and 0.022 for normal and steatotic cases, respectively. In a normal case, distance ([d] = 56) from predetermined point to highest echoic point was also greater than in a steatotic case (d = 12). h = Height, w = width.

View larger version (17K): [in a new window]   Figure 3b. Echo histograms of ROI applied to portal vein of segment VIII and adjacent structures in patients with normal and hepatic macrosteatosis of moderate degree proved at histopathologic examination. (a) In a normal case, right-sided area (tail area, deviant creases) of predetermined point (20% of value of highest point, dashed lines) is relatively larger than that in (b) a steatotic case. For example, in a and b, ratios for tail area of portal vein echogenicity were 0.138 and 0.022 for normal and steatotic cases, respectively. In a normal case, distance ([d] = 56) from predetermined point to highest echoic point was also greater than in a steatotic case (d = 12). h = Height, w = width.

View larger version (30K): [in a new window]   Figure 4. Diagram of ANNs shows input (X), hidden, and output (Y) layers. Seven significant laboratory and US parameters were used as input variables. Hidden layer consists of 15 neurons connected to input layer and is connected to an output layer of a single neuron, which encodes whether US and laboratory findings are classified as appropriate (0) or inappropriate (1) as a liver donor. F = transfer function, i = number of input variables, j = number of neurons in hidden layer, u = activation, w = weight.

View larger version (27K): [in a new window]   Figure 5a. Regression plots show good correlation between pathologic grade of macrosteatosis and US grades. Solid line is regression line, lines of small dots are 95% CIs, and lines of long dashes are 95% prediction interval. (a) Correlation plot between pathologist and faculty radiologist (r = 0.729); Y = 0.625 + 0.546X (r2 = 0.54, P < .001). (b) Correlation plot between pathologist and fellow (r = 0.725); Y = 0.617 + 0.547X (r2 = 0.52, P < .001). (c) Correlation plot between pathologist and resident (r = 0.668); Y = 0.779 + 0.501X (r2 = 0.41, P < .001).

View larger version (29K): [in a new window]   Figure 5b. Regression plots show good correlation between pathologic grade of macrosteatosis and US grades. Solid line is regression line, lines of small dots are 95% CIs, and lines of long dashes are 95% prediction interval. (a) Correlation plot between pathologist and faculty radiologist (r = 0.729); Y = 0.625 + 0.546X (r2 = 0.54, P < .001). (b) Correlation plot between pathologist and fellow (r = 0.725); Y = 0.617 + 0.547X (r2 = 0.52, P < .001). (c) Correlation plot between pathologist and resident (r = 0.668); Y = 0.779 + 0.501X (r2 = 0.41, P < .001).

View larger version (27K): [in a new window]   Figure 5c. Regression plots show good correlation between pathologic grade of macrosteatosis and US grades. Solid line is regression line, lines of small dots are 95% CIs, and lines of long dashes are 95% prediction interval. (a) Correlation plot between pathologist and faculty radiologist (r = 0.729); Y = 0.625 + 0.546X (r2 = 0.54, P < .001). (b) Correlation plot between pathologist and fellow (r = 0.725); Y = 0.617 + 0.547X (r2 = 0.52, P < .001). (c) Correlation plot between pathologist and resident (r = 0.668); Y = 0.779 + 0.501X (r2 = 0.41, P < .001).

View larger version (30K): [in a new window]   Figure 6. Graph shows ROC curves of ANNs and three radiologists. ANNs showed highest Az value, followed by faculty radiologist, fellow, and resident in descending order. Differences between Az values of ANNs and each radiologist were statistically significant (P < .05).