MICCAI 2021 Daily – Wednesday

Junyu’s picks of the day (Wednesday): Junyu Chen is currently a PhD candidate at Johns Hopkins University, where he is a member of the Radiological Physics Division in the Department of Radiology and Radiological Science. “ My research focuses on image analysis and deep learning applied to nuclear medicine imaging. I'm currently working on developing quantitative imaging methods for assessing the response of metastatic bone diseases to therapy. This is my first MICCAI, and it's unfortunate that we'll miss another chance to meet in person this year. Still, I'm looking forward to exploring all the innovative ideas and works that will be presented at this virtual conference. ” ORALS: (We-Oral-AM-A- 1794) Group Shift Pointwise Convolution for Volumetric Medical… (We-Oral-PM-B-1378) Detecting when pre-trained nnU- Net models fail silently for… POSTERS: (We-S2- 97) Medical Matting: A New Perspective on Medical Segmentation with… (We-S4-210) TransCT: Dual-path Transformer for Low Dose Computed Tomography (We-S4-1982) Task-Oriented Low-Dose CT Image Denoising (We-S4-192) UTNet: A Hybrid Transformer Architecture for Medical Image Segmentation For today, Wednesday 29 2 Junyu’s Picks DAILY MICCAI Wednesday Learn more about Junyu’s work! Junyu is presenting Targeted Gradient Descent: A Novel Method for Convolutional Neural Networks Fine-tuning and Online-learning on Wednesday at 11:00 (UTC). His group developed a novel way of fine-tuning a pre-trained network for PET denoising based on discovering that the network produces “redundant” or “useless” feature maps after training. Instead of blindly fine-tuning all the kernels in the specific layers for a new task, they proposed to specifically retrain the convolution kernels that generate these feature maps. “I am an amateur photographer. I love using cameras to capture the little moments of my life. ”