Proposed methods to improve the following deep learning applications for better robustness: Unet-based model (nnUnet) for Heart, Hippocampus and Prostate MRI images segmentation. Multi-task Unet-based model for cephalometric landmark detection. YOLO V5 for blood cell detection. Transformer-based model (TransUnet) for abdominal organ segmentation. The experiments were conducted via Python, Pytorch, Pandas, Scikit-learn, etc., on Tesla V100 GPU. The improved models have better accuracy under testing noises than the baseline models.

This is a Heading

This is a paragraph.

•