cover image: Risk Assessment & Control Coordination for AI services : - Case08 Cancer Diagnostic AI - Institute for Future Initiatives, The University of Tokyo

20.500.12592/7hzsfk

Risk Assessment & Control Coordination for AI services : - Case08 Cancer Diagnostic AI - Institute for Future Initiatives, The University of Tokyo

25 Apr 2022

It uses multimodal learning, incorporating patient medical history/heredity info (Data 1) and endoscopic images (Data 2) to perform real-time highlighting of places with high likelihood of cancer during endoscope medical examinations. [...] ③ During the endoscopic medical exam by a physician, Data 1 and Data 2 are fed into a deep learning model, and performs real-time highlighting of places on endoscopic images with high likelihood of cancer. [...] Uses a multimodal deep neural network (DNN), and is composed of a network that incorporates the following two types of data, and a network that outputs abnormal places in images in real time. [...] Data 1: Table data built from patient medical chart and medical exam Data 2: Endoscope camera image Case08 : Cancer Diagnostic AI - System Overview - AI System Company A) AI model development department Develop AI for real-time detection of places with signs of early cancer Real-time detection of early cancer sign locations during endoscope AI Service Provider Company A) Medical IT services depart. [...] Social Responsibilit Observance of Change of Cannot deal with standard or guideline changes at ministries, agencies or institutions 3 3-2 R011 y of Medical standards/rules standards/rules (institution evidence reflected in AI model learning data) Institution Exposure of AI External exposure of AI medical exam results regardless of the fact that physicians perform R012 medical exam Information ◼ Da.

Authors

LW member

Pages
33
Published in
Japan