Adversarial learning for computer-assisted cancer diagnosis

• Grant number: 19/06533-7
• Support Opportunities: Regular Research Grants
• Start date: October 01, 2019
• End date: December 31, 2022
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Agreement: DAAD
• Mobility Program: SPRINT - Projetos de pesquisa - Mobilidade
• Principal Investigator: João Paulo Papa
• Grantee: João Paulo Papa
• Principal researcher abroad: Christoph Palm
• Institution abroad: Ostbayerische Technische Hochschule Regensburg (OTH Regensburg), Germany
• Host Institution: Faculdade de Ciências (FC). Universidade Estadual Paulista (UNESP). Campus de Bauru. Bauru , SP, Brazil
• Associated researchers: Alexandre Xavier Falcão
• Associated research grant: 14/12236-1 - AnImaLS: Annotation of Images in Large Scale: what can machines and specialists learn from interaction?, AP.TEM

Abstract

Barrett's esophagus (BE) comprises a very severe and growing disease in the last decades, and since BE is often not identified properly at the early stages, it may evolve to a more aggressive version, and even to cancer. However, the early diagnosis of dysplastic tissue in BE-diagnosed patients may provide very high rates of remission after the treatment. In the past few years, deep learning techniques have been extensively employed for automatic feature extraction, thus achieving promising results in different computer vision and image analysis applications including automatic endoscopic image analysis. However, one of the main problems related to deep neural networks is the inherent bias and tendency to overfit due to imbalance of data. Generative Adversarial Networks have been used in a number of situations aiming at creating artificial images, mainly in the context of medicine-driven applications, but not for computer-assisted BE identification. Therefore, this proposal aims at fostering the research towards synthetic data generation in the context of endoscopic image analysis, and it involves Ph.D. students, a post-doctoral researcher, and professors from three distinct universities. (AU)