2019/10/08 by Daniel Polak, Stephen Cauley, Polak, Daniel +11 · 2 citations
Medicine · #Advanced MRI Techniques and Applications #FOS: Electrical engineering #FOS: Physical sciences #Image and Video Processing (eess.IV) #MRI in cancer diagnosis #Medical Imaging Techniques and Applications #Medical Physics (physics.med-ph) #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.1910.03273
openalex publication_date 2019/10/08 · openalex created_date 2022/07/28 · openalex updated_date 2026/07/28
Purpose: To improve the image quality of highly accelerated multi-channel MRI\ndata by learning a joint variational network that reconstructs multiple\nclinical contrasts jointly.\n Methods: Data from our multi-contrast acquisition was embedded into the\nvariational network architecture where shared anatomical information is\nexchanged by mixing the input contrasts. Complementary k-space sampling across\nimaging contrasts and Bunch-Phase/Wave-Encoding were used for data acquisition\nto improve the reconstruction at high accelerations. At 3T, our joint\nvariational network approach across T1w, T2w and T2-FLAIR-weighted brain scans\nwas tested for retrospective under-sampling at R=6 (2D) and R=4x4 (3D)\nacceleration. Prospective acceleration was also performed for 3D data where the\ncombined acquisition time for whole brain coverage at 1 mm isotropic resolution\nacross three contrasts was less than three minutes.\n Results: Across all test datasets, our joint multi-contrast network better\npreserved fine anatomical details with reduced image-blurring when compared to\nthe corresponding single-contrast reconstructions. Improvement in image quality\nwas also obtained through complementary k-space sampling and\nBunch-Phase/Wave-Encoding where the synergistic combination yielded the overall\nbest performance as evidenced by exemplarily slices and quantitative error\nmetrics.\n Conclusion: By leveraging shared anatomical structures across the jointly\nreconstructed scans, our joint multi-contrast approach learnt more efficient\nregularizers which helped to retain natural image appearance and avoid\nover-smoothing. When synergistically combined with advanced encoding\ntechniques, the performance was further improved, enabling up to R=16-fold\nacceleration with good image quality. This should help pave the way to very\nrapid high-resolution brain exams.\n