Attention-based histological image analysis

• Main Author: Wang, Jerome Jie Rui
• Other Authors: Cai Yiyu
• Format: Student Research Poster
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Computer and Information Science; Medicine, Health and Life Sciences
• Online Access: https://hdl.handle.net/10356/180854
• Institution: Nanyang Technological University

With cancer being one of the leading causes of mortality worldwide, research into this field consistently features high up in priorities of the medical community. Hematoxylin & Eosin (H&E) stains represent the gold standard for medical diagnosis (NCI, n.d.). However, H&E-based cancer diagnosis is heavily dependent on the visual judgement of pathologists, which constitutes both a bottleneck and a potential risk – that expert judgement will vary across pathologists (NHE, 2016). Furthermore, manual assessments required as per current practice would result in high inter observer variability (Quaglia, 2018). Coupled with high cost and low throughput, manual annotation of H&E images presents a major challenge in scaling up oncological research (Qaiser, 2018). Thus, this project seeks to automate this process through a state-of-the-art deep learning model.