Optimization of Regularization Parameters in Compressed Sensing of Magnetic Resonance Angiography: Can Statistical Image Metrics Mimic Radiologists' Perception?

• DOI: 10.1371/journal.pone.0146548
• Publication Date: 2016-01-08T18:38:43Z
• Authors: Thai Akasaka; Koji Fujimoto; Takayuki Yamamoto; Tomohisa Okada; Yasutaka Fushumi; Akira Yamamoto; Toshiyuki Tanaka; Kaori Togashi
• Keywords: Regularization; Scale-invariant feature transform

In Compressed Sensing (CS) of MRI, optimization the regularization parameters is not a trivial task. We aimed to establish method that could determine optimal weights for in CS time-of-flight MR angiography (TOF-MRA) by comparing various image metrics with radiologists' visual evaluation. TOF-MRA healthy volunteer was scanned using 3T-MR system. Images were reconstructed from retrospectively under-sampled data varying L1 norm wavelet coefficients and total variation. The images evaluated both quantitatively statistical including structural similarity (SSIM), scale invariant feature transform (SIFT) contrast-to-noise ratio (CNR), qualitatively scoring. results quantitative qualitative scorings compared. SSIM SIFT conjunction brain masks CNR artery-to-parenchyma correlated very well By carefully selecting region measure, we have shown can reflect evaluation, thus enabling an appropriate CS.