Artificial intelligence that determines the clinical significance of capsule endoscopy images can increase the efficiency of reading.

Junseok Park,Youngbae Hwang,Dong Jun Oh,Yun Jeong Lim,Su Hwan Kim,Ki Bae Kim,Ji Hyung Nam,Sun Hyung Kang,Min Kyu Jung,Hyun Joo Song,Sudipta Roy

doi:10.1371/journal.pone.0241474

Abstract

Artificial intelligence (AI), which has demonstrated outstanding achievements in image recognition, can be useful for the tedious capsule endoscopy (CE) reading. We aimed to develop a practical AI-based method that can identify various types of lesions and tried to evaluate the effectiveness of the method under clinical settings. A total of 203,244 CE images were collected from multiple centers selected considering the regional distribution. The AI based on the Inception-Resnet-V2 model was trained with images that were classified into two categories according to their clinical significance. The performance of AI was evaluated with a comparative test involving two groups of reviewers with different experiences. The AI summarized 67,008 (31.89%) images with a probability of more than 0.8 for containing lesions in 210,100 frames of 20 selected CE videos. Using the AI-assisted reading model, reviewers in both the groups exhibited increased lesion detection rates compared to those achieved using the conventional reading model (experts; 34.3%-73.0%; p = 0.029, trainees; 24.7%-53.1%; p = 0.029). The improved result for trainees was comparable to that for the experts (p = 0.057). Further, the AI-assisted reading model significantly shortened the reading time for trainees (1621.0-746.8 min; p = 0.029). Thus, we have developed an AI-assisted reading model that can detect various lesions and can successfully summarize CE images according to clinical significance. The assistance rendered by AI can increase the lesion detection rates of reviewers. Especially, trainees could improve their efficiency of reading as a result of reduced reading time using the AI-assisted model.

Highlights

Deep learning-based artificial intelligence (AI) has demonstrated outstanding levels of achievement in image recognition [1]
The calculated probabilities of significance and the results of class activation map of all the 40,000 test set images were reviewed by three professors who contributed to the classification of the AI training set
A total of 860 lesions in the 20 Capsule endoscopy (CE) videos were reported as references by the three professors who contributed to the classification of the AI training set

Summary

Introduction

Deep learning-based artificial intelligence (AI) has demonstrated outstanding levels of achievement in image recognition [1]. The diagnostic yield of medical imaging tests relies on lesion detection, for which the use of AI we can be take the advantageous of AI [2,3,4]. Capsule endoscopy (CE) creates a large number of images while examining the mucosal surface of small bowel [5]. The manufacturers of capsule endoscopes develop specially designed software to increase the diagnostic yield of CE. The reviewers detect lesions by looking at the context of the images reconstructed by the software according to the image-acquisition time. To alleviate the burden on reviewers, many studies have attempted to apply AI to CE reading and showed impressive results in detecting small bowel lesions [7,8,9]. The effectiveness of AI should be evaluated in situations similar to those of conventional reading methods

Objectives

Methods

Results

Conclusion