School of Engineering Department of Computer Science and Engineering 108 AI in Medical Imaging: Automatic Stroke Analysis on Brain CT Scans. Supervisor: CHEN, Hao / CSE Student: SON, Moo Hyun / DSCT Course: UROP1100, Summer This paper explores the application of deep learning models to create segmentation masks using lower temporal resolution Computed Tomography (CT) scan image with the aim of reducing radiation exposure. The study leverages advanced machine learning techniques to maintain image quality and accuracy despite the reduced resolution. Preliminary results indicate that the proposed method significantly minimizes radiation exposure without compromising the effectiveness of the segmentation process. This innovative approach could revolutionize CT scan procedures, enhancing patient safety while maintaining diagnostic integrity. Further research and validation are required to optimize the model and assess its potential for widespread clinical application. The implications of this study could extend to various medical imaging procedures, contributing to safer and more efficient diagnostic processes. Data-efficient, Domain Generalizable and Interpretable Deep Learning Supervisor: CHEN, Hao / CSE Student: TAN, Juin / SENG Course: UROP1000, Summer Cancer is a significant global cause of death, highlighting the importance of precise and prompt detection (WHO, 2020). Histopathology biopsy images are crucial for cancer diagnosis. However, manually identifying cancer cells is challenging due to subjectivity and interpretational variability. This report investigates machine learning techniques, specifically deep learning-based pipelines, for assisting in cancer diagnosis through the analysis of microscopy images. The Faster R-CNN model was trained on the ISBI dataset, with the parameters and results displayed and analyzed. The experiment's weaknesses are identified and suggestions for improvement are provided. In conclusion, further research can enhance cancer detection accuracy and implementation in real-world healthcare settings. Federated Learning in Healthcare with Privacy-Preserving Supervisor: CHEN, Hao / CSE Student: WONG, Hoi Tin / FINA Course: UROP1100, Summer This project aims to conduct comprehensive research on federated learning (FL) frameworks and related works in healthcare, especially in medical image recognition. The existing problem of domain shift is critical in medical image recognition in federated settings (Ying et al., 2023). Thus, this my research work will be focusing on domain generalization (DG) in FL. This progress report covers the first part of the project, research on background, including FL and DG, and various DG techniques has been conducted in this stage. Meanwhile, a method for privacy-preserving pairwise DG is proposed for the next stage of this project. Preliminary analysis of the proposed method will be discussed in the last part of this report.
RkJQdWJsaXNoZXIy NDk5Njg=