More about HKUST
Automated cancer detection on histopathology images using deep convolutional neural networks
PhD Qualifying Examination Title: "Automated cancer detection on histopathology images using deep convolutional neural networks" by Mr. Yongxiang HUANG Abstract: Histopathology imaging allow pathologists to access the microscopic structure and elements of biopsy specimens, which is an essential diagnostic method to finalize the potential diseases such as cancer. Unlike natural images, pathology images require highly skilled pathologists to review, which is fairly time-consuming and error-prone. With the increasing ability to rapidly digitalize whole slide images with slide scanners, research communities, medical intuitions and companies raise interests in developing computer-assisted diagnostic algorithms for automatic detection of disease extent from pathology images. Deep learning of digitalized pathology slides offers the potential to reduce misdiagnosis and improve the speed of screening. This survey aims to give a comprehensive overview of automated cancer detection on histopathology images using deep convolution neural networks. We first introduce the workflow of digital histopathology analysis, in which state-of-the-art methods for histology images classification in glimpse-level and side-level are reviewed. Then we review methods for cancer metastasis detection and localization on giga-pixel histopathology images of lymph node section. Finally, we discuss the potential directions for further research on detecting lesions on histopathology images. Date: Wednesday, 6 March 2019 Time: 10:00am - 12:00noon Venue: Room 3494 Lifts 25/26 Committee Members: Prof. Albert Chung (Supervisor) Prof. Chiew-Lan Tai (Chairperson) Prof. Long Quan Dr. Pedro Sander **** ALL are Welcome ****