SNR-Weighted Regularization of ADC Estimates from Double‐Echo in Steady‐State (DESS)
Osteoarthritis is a difficult condition that has painful and debilitating effects on many people, leading to high medical costs. As the US population gets older, this problem will only get worse. Measuring the diffusion of hydrogen atoms in cartilage could possibly help detect osteoarthritis at an early stage. MRI methods to measure diffusion often produce distorted images with low spatial resolution, making them challenging to use in cartilage. The DESS sequence can produce anatomical images and quantitative estimates, including high-resolution, undistorted diffusion measurements. However, these diffusion estimates can be noisy. In this work, we developed an algorithm that smoothes out diffusion estimate variations where they are likely due to noise, but leaves them intact where they are likely to be true.
Sveinsson B, Gold GE, Hargreaves BA, Yoon D. SNR-weighted regularization of ADC estimates from double-echo in steady-state (DESS). Magn Reson Med. 2019 Jan;81(1):711-8.
The effects of regularization in computer simulations (left) and in a knee scan (right). A higher λ gives stronger regularization. Regions with high signal-to-noise ratio (SNR) are left unregularized.
Bragi Sveinsson is an alumnus of the BMR group