오승민 (아주대학교)

한빛사 논문

조회128

아주대학교

Int. J. Surg., Sep 15 2023, | https://doi.org/10.1097/JS9.0000000000000721 Deep learning using computed tomography to identify high-risk patients for acute small bowel obstruction: development and validation of a prediction model : A retrospective cohort study.

Oh, Seungmin BA^a; Ryu, Jongbin PhD^a,b; Shin, Ho-Jung MD, MS^c; Song, Jeong Ho MD, MS^c; Son, Sang-Yong MD, PhD^c; Hur, Hoon MD, PhD^c; Han, Sang-Uk MD, PhD^c

^aDepartment of Artificial Intelligence, Ajou University, South Korea

^bDepartment of Software and Computer Engineering, Ajou University, South Korea

^cDepartment of Surgery, Ajou University School of Medicine, Suwon, Korea

^*Corresponding authors: Ryu, Jongbin PhD, Shin, Ho-Jung MD, MS

Abstract

To build a novel classifier using an optimized 3D-convolutional neural network for predicting high-grade small bowel obstruction (SBO). Acute SBO (ASBO) is one of the most common acute abdominal diseases requiring urgent surgery. While AI and abdominal CT have been used to determine surgical treatment, differentiating normal cases, high-grade small bowel obstruction (HGSBO) requiring emergency surgery, and low-grade SBO or paralytic ileus (LGSBO) is difficult. A deep learning classifier was used to predict high-risk ASBO patients using CT images at a tertiary hospital. Images from three groups of subjects (normal, nonsurgical, and surgical) were extracted; the dataset used in the study included 578 cases from 250 normal subjects, with 209 HGSBO and 119 LGSBO patients; over 38,000 CT images were used. Data were analyzed from June 1, 2022, to February 5, 2023. The classification performance was assessed based on accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve. After 5-fold cross-validation, the WideResNet classifier using dual-branch architecture (DBA) with depth retention pooling (DRP) achieved an accuracy of 72.6%, an area under receiver operating characteristic of 0.90, a sensitivity of 72.6%, a specificity of 86.3%, a positive predictive value of 74.1%, and a negative predictive value of 86.6% on all the test sets. These results show the satisfactory performance of the deep learning classifier in predicting HGSBO compared to the previous machine learning model. The novel 3D classifier with DBA and DRP based on AI algorithms could be a reliable screening and decision-support tool for high-risk patients with SBO.

논문정보

형식Research article
게재일2023년 09월 (BRIC 등록일 2023-09-20)
연구진국내 연구진
분야

댓글 작성 로그인

CV 열람하기

연락처 보기