We are thrilled to announce that our Ph.D. students, Bingyu Xin and Meng Ye, earned first place in both tasks of the MICCAI CMRxRecon2024 Challenge, outshining over 200 teams from 32 countries. This remarkable accomplishment highlights the CS team's cutting-edge research and dedication to medical image reconstruction.
The CMRxRecon2024 Challenge was part of the 27th International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI 2024), held from October 6th to 10th, 2024, in Marrakesh, Morocco. The CMRxRecon2024 Challenge focused on accelerating cardiac MRI reconstruction from highly undersampled k-space data. This year’s competition included two task tracks: Multi-contrast CMR Reconstruction and Random-sampling CMR Reconstruction.
The CS team applied the method presented in their recent ECCV (European Conference on Computer vision) 2024 oral paper, collaborating with medical experts from New York University Grossman School of Medicine, titled "Rethinking Deep Unrolled Model for Accelerated MRI Reconstruction." This approach improves the performance and efficiency of deep unrolled models for accelerated MRI reconstruction by introducing adaptive unrolling and efficient sensitivity map estimation.
This scientific advancement provides superior MR image reconstruction quality, potentially enabling more precise MR image analysis and clinical diagnosis, and will allow for more aggressive k-space undersampling, leading to faster MRI scanning.
“Having won the challenge last year as well, we are excited to report that our novel MRI reconstruction method achieves even better image reconstruction quality while using 55% less memory and processing 1.6 times faster," said CS Distinguished Professor Dimitri Metaxas. “We are well on our way to making MRI scanning, processing, and analytics in under 10 minutes, which is what all top groups in medical image analytics are trying to do. And we are ahead. Most importantly, it will help patients tremendously in terms of getting more accurate diagnoses in many cases.”
Congratulations and another first place win to Professor Metaxas and his team. To learn more https://cmrxrecon.github.io/2024/Home.html
