Mammary duct detection using self-supervised encoders
In person: 22 February 2022 • 2:00 PM - 2:20 PM PST
Duct detection in Hematoxylin and Eosin stained whole-slide images (WSIs) along with downstream analysis is necessary for the diagnosis and treatment planning of Ductal Carcinoma in-Situ (DCIS). This process can be facilitated using deep learning methods. We used novel self-supervised learning methods to produce feature encodings and compared their performance with ImageNet features and random initialisation on the downstream task of duct detection.
The Netherlands Cancer Institute (Netherlands)
Shannon Doyle is a PhD student at the AI for oncology group lead by Jonas Teuwen at the Netherlands Cancer Institute in Amsterdam. She is also supervised by Clarisa Sanchez who is a Full Professor for AI in Health at the UvA. Shannon's PhD focuses on the risk and outcome prediction of breast cancer featuring DCIS. Specifically, she is working on determining which higher-risk (DCIS) lesions are at low risk of progressing into ipsilateral invasive breast cancer based on histopathology data. With this goal in mind, she developed duct detectors using self-supervised learning and and object detection methods. This is the topic of her conference paper. Later in her PhD, Shannon will also develop deep learning methods to identify calcified lesions with a low harm risk during breast cancer screenings. Shannon's interests are spread over biomedicine, bioinformatics, entrepreneurship, and AI. As such,