Acceleration of skeletal age MR examination using compressed sensing

Purpose: To examine the feasibility of accelerating magnetic resonance (MR) image acquisition for children using compressed sensing (CS). Skeletal age assessment using MRI sometimes suffers from motion artifacts because of the long scan time in children. Reducing image acquisition time may provide benefits by reducing motion artifacts, increasing efficiency of examination, and creating a stress-free environment. Materials and Methods: Undersampling patterns for CS were optimized and CS-based examination with the acceleration factors of 3 (CS3, 55 seconds per scan) and 4 (CS4, 41 seconds per scan) was performed for 59 subjects (35 boys and 24 girls; mean age, 9.1 years; age range, 4.4–15.3 years) using a 0.3T scanner. The skeletal age was assessed by two raters (A and B). Results: The interrater and intrarater reproducibility in skeletal age assessment was high (Pearson's r = 0.966 [CS3(A1) vs. CS3(A2)], 0.962 [CS4(A1) vs. CS4(A2)], 0.935 [CS3(A1) vs. CS3(B)], and 0.964 [CS4(A1) vs. CS4(B)]; P < 0.001). The errors in skeletal age assessed on the basis of CS-reconstructed images were similar to those assessed on the basis of fully Nyquist-sampled images. Conclusion: These results demonstrate the validity and reliability of skeletal age examination accelerated by CS-MRI. We conclude that the acceleration factor of 3 was optimal. J. Magn. Reson. Imaging 2016;44:204–211.