Development of a User-Centered Virtual Liver Surgery System for Preoperative Liver Surgery Planning

Abstract

This study developed a user-centered 3D liver surgery planning system, called Dr. Liver, to support liver surgery planning in a clinical environment for safe and rational liver surgery. The currently available virtual surgery systems need to be customized to liver surgery and improved for better usability and time efficiency. This study established use scenarios of Dr. Liver through literature review, benchmarking, and interviews with surgeons. The use scenarios consist of high level tasks including liver extraction, vessel extraction, tumor extraction, liver segmentation, and liver surgery planning and low level tasks to accomplish the high level tasks. Based on the use scenarios, detailed user interfaces were designed and image processing algorithms were developed. For better usability, Dr. Liver provides various user-friendly features such as procedure status indication and color coding, 3D view indication box and resetting buttons for easier 3D object manipulation, and hotkey menus appearing on the screen to decrease users’ cognitive workload.This study developed a hybrid semi-automatic method to extract the liver from abdominal computerized tomography (CT) images. The proposed hybrid method consists of a customized fast-marching level-set method for detection of an optimal initial liver region from multiple seed points selected by the user and a threshold-based level-set method for extraction of the actual liver region based on the initial liver region. The performance of the hybrid method was compared with those of the 2D region growing method implemented in OsiriX using abdominal CT datasets of 15 patients. The hybrid method showed a significantly higher accuracy in liver extraction (similarity index, SI = 97.6% ± 0.5%

false positive error, FPE = 2.2% ± 0.7%

false negative error, FNE = 2.5% ± 0.8%

average symmetric surface distance, ASD = 1.4 ± 0.5 mm) than the 2D (SI = 94.0% ± 1.9%

FPE = 5.3% ± 1.1%

FNE = 6.5% ± 3.7%

ASD = 6.7 ± 3.8 mm) region growing method. The total liver extraction time per CT dataset of the hybrid method (77 ± 10 sec) is significantly less than the 2D region growing method (575 ± 136 sec). The interaction time per CT dataset between the user and a computer of the hybrid method (28 ± 4 sec) is significantly shorter than the 2D region growing method (484 ± 126 sec). The proposed hybrid method was found preferred for liver segmentation in preoperative virtual liver surgery planning.This study developed an interactive method for efficient liver vessel extraction from abdominal CT images. The proposed interactive liver extraction method consists of (1) pre-processing of CT images in which multiple phases of abdominal CT images are denoised, registered, and masked with the extracted liver region, (2) selection of multiple seed points, (3) identification of multiple threshold intervals based on the intensity values of the selected seed points, (4) vessel segmentation with identified threshold intervals using region growing method, (5) display of multiple segmentation results for the user to select an appropriate segmentation result, and (6) interactive editing of the extracted vessel trees if necessary. The performance of the interactive method was accessed by an expert radiologist using 15 abdominal CT datasets. No false positive errors were found in the extracted vessel branches. False negative errors were identified at some distal branches of the vessel tree due to small diameter and low contrast. No connections among the extracted portal vein, hepatic vein, and hepatic artery were found in the 15 segmented datasets. A 7-point Likert scale was used for assessment of suitability for liver surgery planning, ‘1’ for very poor and ‘7’ for very good. The average ( S.D.) score of suitability for liver surgery planning was 6.4 ( 0.7). The interaction time and total vessel extraction time were 33 ( 4) sec and 75 ( 8) sec respectively. The proposed interactive liver extraction method was found suitable for clinical application such as liver surgery planning.Ergonomic usability tests consisting of a preliminary test and a main test were conducted at different system development stages of Dr. Liver. While the preliminary usability test conducted at an early system development stage helped developers to identify potential usability problems of Dr. Liver and produce recommendations to resolve the problems, the main usability test verified the improvement of the usability of Dr. Liver. The usability of Dr. Liver was evaluated using a comprehensive set of performance (completion time, similarity index, false positive error, false negative error, number of mouse clicks, and number of keystrokes) and preference (usefulness, ease of use, learnability, informativeness, clarity, tolerance, and overall satisfaction) measures. Ten male medical doctors (aged from 30s to 60s

experienced in liver anatomy and liver surgery) from five different medical centers participated in the main usability test, consisting of five test modules. The system received a high score of satisfaction (mean = 6.2, S.D. = 0.7) as measured using a 7-point Likert scale throughout the five test modules. The present study will help practitioners evaluate the usability of a system and identify potential usability problems in a systematic manner.