Rapid MRI Acquisition

Magnetic resonance imaging (MRI) is an excellent diagnostic imaging modality, however, more widespread use of MRI is limited by its long acquisition duration and costs. We develop new techniques to accelerate acquiring MRI scans without compromising diagnostic value and to convert MRI from a qualitative to quantitative technology.

New Sequence Development

The MIMI group and collaborators have developed and optimized the use of the quantitative double-echo in steady-state pulse sequence (qDESS) that enables acquiring 3D, distortion-free, high-resolution, and multi-contrast images, along with simultaneous quantitative T2 mapping. The qDESS technique enables comprehensive diagnostic knee MRI and acquisition of osteoarthritis biomarkers in only a five minute acquisition. qDESS further enables high accuracy segmentation of multiple tissues in the knee.

Select Publications:

  • Chaudhari A, Grissom M, Fang Z, Sveinsson B, Lee JH, Gold G, Hargreaves B, and Stevens K. Diagnostic Accuracy of Quantitative Multi-Contrast 5-Minute Knee MRI Using Prospective Artificial Intelligence Image Quality Enhancement. American Journal of Roentgenology (2021) 216:6, 1614-1625. doi:10.2214/AJR.20.24172
  • Eijgenraam S, Chaudhari A, Reijmam, Bierma-Zeinstra S, Hargreaves B, Runhaar J, Heijboer F, Gold G, and Oei E.H. Time-saving Opportunities in Knee Osteoarthritis: Structural Imaging and T2 mapping in the Knee using a Single 5-minute MRI Scan. (2020), European Radiology, 30(4), 2231-2240. doi: 10.1007/s00330-019-06542-9
  • Chaudhari A, Stevens K, Sveinsson B, Wood J, Beaulieu C, Oei E.H., Rosenberg J, Kogan F, Alley M, Gold G, and Hargreaves B. Combined 5-Minute Double-Echo in Steady-State with Separated Echoes and 2-Minute Proton-Density-Weighted 2D FSE Sequence for Comprehensive Whole-Joint Knee MRI Assessment. Journal of Magnetic Resonance Imaging. (2019), 49(7), e183-e194. doi: 10.1002/jmri.26582
  • Chaudhari A, Black M, Eijgenraam S, Wirth W, Maschek S, Sveinsson B, Eckstein F, Oei, E.H, Gold G, and Hargreaves B. 5-Minute Knee MRI for Simultaneous Morphometry and T2 Relaxometry of Cartilage and Meniscus and for Semi-Quantitative Radiological Assessment using Double-Echo in Steady-State at 3T. Journal of Magnetic Resonance Imaging (2018), 47(5), 1328-1341. doi: 10.1002/jmri.25883
  • Chaudhari A, Sveinsson B, Moran C, McWalter E, Johnson E, Zhang T, Gold G, and Hargreaves B. Imaging and T2 relaxometry of short‐T2 connective tissues in the knee using ultrashort echo‐time double‐echo steady‐state (UTEDESS). Magnetic Resonance in Medicine (2017), 78:2136–2148. doi: 10.1002/mrm.26577

Accelerated MRI

We develop new machine learning-based image reconstruction techniques to enable rapid and highly-undersampled MRI acquisition. Such inverse problems consist both of image-to-image networks as well as networks that transform raw k-space MRI data. We consider varying regimes with different availability of training data, and have developed supervised, semi- supervised, and unsupervised MRI reconstruction techniques. Beyond evaluating performance on in-distribution datasets, we characterize and account for adversarial perturbations and common distribution shifts that occurring in MRI acquisition. Our research consists of developing new algorithms for reconstruction along with new dataset curation and clinical deployment.

Select Publications:

  • Chaudhari A, Fang Z, Kogan F, Wood J, Stevens K, Gibbons E, Lee J.H, Gold G, and Hargreaves B. Super-Resolution Musculoskeletal MRI Using Deep learning. Magnetic Resonance in Medicine (2018), 80(5):2139-2154. doi: 10.1002/mrm.27178
  • Darestani MZ, Chaudhari AS, Heckel R. Measuring Robustness in Deep Learning Based Compressive Sensing. In: Proceedings of the 38th International Conference on Machine Learning. Vol 139. Proceedings of Machine Learning Research. PMLR; 2021:2433-2444.
  • Chaudhari A, Stevens K, Wood J, Chakraborty A, Gibbons E, Fang Z, Desai A, Lee JH, Gold G, Hargreaves B. Utility of Deep Learning Super-Resolution in the Context of Osteoarthritis MRI Biomarkers. Journal of Magnetic Resonance Imaging. (2019). 51 (3), 768-779. doi: 10.1002/jmri.26872
  • Chaudhari A, Grissom M, Fang Z, Sveinsson B, Lee JH, Gold G, Hargreaves B, and Stevens K. Diagnostic Accuracy of Quantitative Multi-Contrast 5-Minute Knee MRI Using Prospective Artificial Intelligence Image Quality Enhancement. American Journal of Roentgenology (2021) 216:6, 1614-1625. doi:10.2214/AJR.20.24172
  • Sveinsson B, Chaudhari A, Zhu B, Koonjoo N, Torriani M, Gold G, and Rosen M. Synthesizing Quantitative T2 Maps in Right Lateral Knee Femoral Condyles from Multi-Contrast Anatomical Data with a Conditional GAN. Radiology: Artificial Intelligence (2021). doi: 10.1148/ryai.2021200122