Enhancing Cell Detection via FC-HarDNet and Tissue Segmentation: OCELOT 2023 Challenge Approach

Accurate cell detection is essential for numerous applications in biomedical research, including cancer diagnosis, drug development, and understanding cellular mechanisms. It involves identifying and locating cells within images acquired from various microscopy techniques. In order to understand cell behavior and tissue structure, using computer-aided system is a efficient and promising way. In this paper, we present our approach for the OCELOT 2023 Cell Detection from Cell-Tissue Interaction challenge. Our proposed method utilizes the FC-HarDNet architecture for cell detection and tissue segmentation, which has shown promising results in various computer vision tasks. Additionally, we incorporate tissue segmentation results to aid in the classification of detected cells, leveraging the valuable information encoded in the spatial relationships between cells and their surrounding tissue. Our method achieved 0.6992 mean F1-score and ranked fifth in the OCELOT 2023 Challenge, demonstrating the potential of integrating cell-tissue interactions for improved cell detection in biomedical image analysis.