Computer Vision News - August 2020

Runner-up for Best Paper Award 26 Best of MIDL 2020 Sriprabha Ramanarayanan is a project engineer in the Healthcare Technology Innovation Centre at the Indian Institute of Technology Madras (IITM) and a PhD scholar at IITM. Her paper on MRI reconstruction was a runner-up in the Best Paper award at MIDL 2020. She speaks to us about her work. MRI reconstruction is a key part of the MRI workflow and the quality of medical images is critical for clinical diagnosis. This work seeks to transform MRI reconstruction by exploiting the benefits of recent advances in deep learning . Existing convolutional neural network-based MRI reconstruction techniques provide fast and high- quality reconstructions. However, these networks lack flexibility and only operate and perform well for specific acquisition contexts. This limits their practical use when translated into a clinical environment . By incorporating flexibility to multiple contexts, this new method seeks to take existing techniques to the next level. It proposes a model that performs well on multiple tasks and in unseen acquisition contexts . MAC-ReconNet: A Multiple Acquisition Context based Convolutional Neural Network for MR Image Reconstruction using Dynamic Weight Prediction “This new network providesgeneralizability and flexibility in image reconstruction.”