Title：Medical image analysis and explainability
Time：10:00~11:00 on August 16
Venue：Room 446 in ICT
Summary：
Though deep learning has shown successful performance in classifying the label and severity stage of certain disease, most of them give few evidence on how to make prediction. Here, we propose to exploit the interpretability of deep learning application in medical diagnosis. Inspired by Koch’s Postulates, a well-known strategy in medical research to identify the property of pathogen, we define a pathological descriptor that can be extracted from the activated neurons of a diabetic retinopathy detector. To visualize the symptom and feature encoded in this descriptor, we propose a GAN based method to synthesize pathological retinal image given the descriptor and a binary vessel segmentation. Besides, with this descriptor, we can arbitrarily manipulate the position and quantity of lesions. As verified by a panel of 5 licensed ophthalmologists, our synthesized images carry the symptoms that are directly related to diabetic retinopathy diagnosis. The panel survey also shows that our generated images is both qualitatively and quantitatively superior to existing methods.
Brief introduction of the speaker：
Dr. Gu Linobtained his Ph.D. degree from the Australian National University in 2014 and worked as a postdoctoral researcher in A*STAR, Singapore from 2014 to 2016. In 2016, he joined the National Institute of Informatics of Japan as a special researcher and a part-time visiting researcher at Kyoto University. At the same time, Dr. Gu Lin participated in the light ultrasound medical equipment research and development project from the Japanese Cabinet Office, and is specifically responsible for image processing algorithms and clinical verification work. Dr. Gu Lin\'s research interests include general artificial intelligence technology and its applications in medical imaging and computational photography.
At present, more than ten papers have been published in high-level international journals and international conferences in these field. Many of the papers were published in the top journals of TIP, TMI and the top conferences of ICCV, CVPR,MICCAI. His works have been concerned and recognized by domestic and foreign counterparts. He has won the Best Student thesis Award of the International pattern recognition Society and the Australian pattern recognition Society respectively.
Download: