Parasite Egg Segmentation With a FLIM-BoFP Network

• Grant number: 24/23772-3
• Support Opportunities: Scholarships abroad - Research Internship - Scientific Initiation
• Start date: March 27, 2025
• End date: June 26, 2025
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Alexandre Xavier Falcão
• Grantee: João Deltregia Martinelli
• Supervisor: Jefersson A dos Santos
• Host Institution: Instituto de Computação (IC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Institution abroad: University of Sheffield, England
• Associated to the scholarship: 24/08332-7 - Kernel Estimation in FLIM-based CNNs for Object Detection, BP.IC

Abstract

Deep learning models have been successful in image segmentation at the price of considerable human effort in data annotation and network adaptation. Such drawback is relevant in medical applications due to the time constraints faced by professionals and the cost of their time. FLIM (Feature Learning from Image Markers) addresses this issue by estimating filters from markers drawn by an expert on discriminative regions of very few images (e.g., 5), with no need for backpropagation. By that, FLIM can also create lightweight networks for training and inference in hardware-constrained environments. The ongoing FAPESP research project compares two filter estimation methods based on the FLIM methodology: patch clustering block-by-block (FLIM-Cluster) and Bag of Feature Points (FLIM-BoFP). This FAPESP-BEPE proposal aims to refine this comparison by incorporating additional lightweight and few-shot learning networks and enhancing FLIM-BoFP with an adaptive decoder and delineator for segmenting parasite eggs in microscopy images. The research will be conducted at the University of Sheffield, supervised by Dr. Jefersson Alex dos Santos.