Presentation information

Organized Session

Organized Session » OS-1

[2K5-OS-1a] 医療におけるAIの社会実装に向けて(1/2)

Wed. Jun 15, 2022 3:20 PM - 5:00 PM Room K (Room K)

オーガナイザ:小寺 聡(東京大学)[現地]、木村 仁星(東京大学)、小林 和馬(国立がん研究センター)、杉原 賢一(エムスリー)

4:40 PM - 5:00 PM

[2K5-OS-1a-05] Application of Automatically Generated Image Database in Pre-training Model for Cystoscopic Image Classification

〇Ryuunosuke Kounosu1, Hirokazu Nosato2, Yuu Nakajima1 (1. Toho University, 2. National Institute of Advanced Industrial Science and Technology)

Keywords:AI, transfer learning, cystoscopy, deep learning, automatically generated image database

When applying artificial intelligence to medical imaging, deep learning models pre-trained on ImageNet are commonly used. However, ImageNet cannot be used for commercial purposes, making it difficult to put to practical use even if excellent diagnostic support is achieved. Therefore, we propose a method to apply a deep learning model pre-trained on the FractalDB dataset, an automatically generated image dataset, to medical imaging. In this paper, we use cystoscopy images to validate the effectiveness against medical images of proposed pre-training method. As a result, the classification model using FractalDB-1k, which has 1000 classes among FractalDB, as a pre-training model outperformed the classification model trained only on cystoscopy images in terms of Accuracy, Sensitivity, Specificity, F1-Score, Precision, and AUC.

Authentication for paper PDF access

A password is required to view paper PDFs. If you are a registered participant, please log on the site from Participant Log In.
You could view the PDF with entering the PDF viewing password bellow.