[#21-21] MICS在线学术讲座：洪义
paper:MDA-Net: Multi-Dimensional Attention-Based Neural Network for 3D Image Segmentation
报告摘要
In medical image analysis, segmentation and regression are two fundamental techniques for understanding an individual image or a population of images. However, conventional methods suffer from high computational costs, especially for high-resolution image volumes; and deep-learning-based approaches have the generalization issue due to insufficient data coverage and limited annotations at the image pixel/voxel level. In this talk, I will present our efficient computational models, which tackle these challenges by fully leveraging available data and decomposing complex models. Applications of our methods include extracting regions of interest or anomalies in MRI and CT scans and understanding brain degeneration and Alzheimer’s Disease.
数据集：
既往研究：网络模型：
训练参数设定:
训练结果:
总结:
