[1H3-J-13] AI application: medicine and healthcare

[1H3-J-13-02] Diagnostic Classification of Chest X-Rays Pictures with Deep Learning Using Eye Gaze Data

〇Taiki Inoue¹, Nisei Kimura², Nakayama Kotaro^2,3, Kenya Sakka⁴, Abdul Ghani Abdul Rahman², Ai Nakajima⁵, Radkohl Patrick², Satoshi Iwai⁶, Yoshimasa Kawazoe⁶, Kazuhiko Ohe⁶ (1. Graduate School of Pharmaceutical Sciences, The University of Tokyo, 2. Graduate School of Engineering, The University of Tokyo, 3. NABLAS Inc., 4. Graduate School of Frontier Sciences, The University of Tokyo, 5. Aalto University, 6. Graduate School of Medicine, The University of Tokyo)

Keywords:medical images, deep learning

Automatic diagnosis of chest X-ray pictures with deep learning has been extensively studied in recent years. In order to improve the accuracy, it is important how to input small localized areas which are disease specific while at the same time using the information that can be obtained by the whole picture. We considered that human eye-gaze fixations can be a biomarker that indicates areas specific to disease. In this study, we propose a deep learning model utilizing eye-gaze data. We demonstrate that the classification shows the better accuracy on using eye-gaze data of experienced doctors than eye-gaze data of novice or non-use of eye-gaze information.