More about HKUST
Deep Learning for Computational Cytopathology: A Survey
PhD Qualifying Examination Title: "Deep Learning for Computational Cytopathology: A Survey" by Mr. Hao JIANG Abstract: Cytopathology enables early cancer detection through microscopic evaluation of cellular morphology to identify precancerous lesions and malignancies. With advancement of artificial intelligence (AI), computational cytopathology (Ccyto) increasingly leverages deep learning (DL)-based computer-aided techniques to enhance the traditional cytopathology workflow, potentially accelerating analysis tenfold and improving sensitivity. This survey reviews over 200 literatures on DL-based Ccyto, focusing on advanced DL methodologies and commercial applications. Specifically, we outline the Ccyto workflow, emphasizing key DL tasks: cytomorphological segmentation for delineating cellular structures, cellular identification for categorizing cancerous cell types, and whole slide classification for patient-level precancerous screening. We introduce DL paradigms, public datasets, and evaluation metrics. Finally, we discuss challenges and future directions for DL in computational cytopathology. Date: Tuesday, 13 May 2025 Time: 3:00pm - 5:00pm Venue: Room 3494 Lifts 25/26 Committee Members: Dr. Hao Chen (Supervisor) Prof. Pedro Sander (Chairperson) Dr. Terence Tsz Wai Wong (CBE)