USM Graduate Student Presents AI-Driven Biomedical Imaging Research at ISVC 2025 and FMLDS 2025
Graduate student Sean Fletcher, working under the mentorship of Dr. Yuqi Song and Dr. Xin Zhang from the University of Southern Maine’s Computer Science Department, recently presented two research papers at the International Symposium on Visual Computing (ISVC 2025) and the IEEE International Conference on Future Machine Learning and Data Science (FMLDS 2025). These presentations mark an important milestone for USM’s growing research presence in artificial intelligence and computational biomedical imaging.
At ISVC 2025, Fletcher presented the team’s work titled “Interpretable Tile-Based Classification of
Paclitaxel Exposure.” This project introduces a tiling-based deep learning method designed to detect
subtle morphological changes in C6 glioma cell images treated with paclitaxel (Taxol). By dividing each microscopy image into smaller patches and aggregating predictions, the model achieved 97%
classification accuracy, surpassing previous baselines by 20 percentage points. The method enhances interpretability while improving sensitivity to small cellular variations–an important advancement for AI-assisted drug analysis.
Fletcher’s second presentation at FMLDS 2025 highlighted his first-author paper, “Deep Learning for Taxol Exposure Analysis: A New Cell Image Dataset and Attention-Based Baseline Model.” This research releases a new publicly available microscopy dataset of C6 glioma cells exposed to varying Taxol concentrations and proposes an innovative ResAttention-KNN architecture that combines ResNet-50, attention mechanisms, and k-NN classification to achieve robust performance. Together, these projects demonstrate USM’s commitment to integrating student research training with cutting-edge AI methodologies and contributing impactful tools to the scientific community.