Abstract 17448: A Novel Technique for Respiratory Motion Correction in Rapid Left Ventricular Myocardial T1 Mapping and Quantitative Analysis of Myocardial Fibrosis
Introduction: Quantitative myocardial T1 mapping is increasingly being used to measure myocardial fibrosis, but this approach requires effective breath-holds during MRI. Respiratory artifacts from poor breath-holds result in motion-corrupted pixels and measurement error. We developed and tested the feasibility of an approach that applies motion correction (MC) followed by semi-automated segmentation to obtain motion-free T1 maps of the LV. Methods: Modified Look-Locker Inversion Recovery (MOLLI) data was acquired on 1.5T MRI scanner, where the endo- and epicardial borders were semi-automatically detected using noise characteristics of myocardial tissue [1,2] and followed by fully automated partitioning into AHA-defined segments [1]. Affine motion correction was then applied to each segment to generate MC-T1 maps of the heart. This approach was tested on 24 slices (12 before contrast injection [PRE]; 12 post [POST], 96x2 ROI segments) from 4 swine with no LV abnormality. The same segmented ROIs on T1 maps without MC were also assessed for comparison. Results: The standard deviation of T1 within each ROI became significantly lower after MC: [MC vs non-MC: 94 ± 37 vs 114 ± 51 ms (PRE, p<0.00005); 66 ± 51 vs 89 ± 67 ms (POST, p<0.0005)], suggesting less motion blurring and possibly less error in T1 measurements within each generated ROI (fig). Significant changes were observed in POST T1 values (446 ± 66 vs 435 ± 89ms; p < 0.0005), yielding an average increase of 2.6 ± 1.6% per segment. The inferior (+3.9%) and inferiolateral segments (+4.5%) yielded the most change, corresponding to regions with most motion across MOLLI images as assessed visually. PRE T1 changes were also significant (998 ± 94 vs 1008 ± 114 ms; p < 0.05). Conclusions: Our new semi-automated and motion-corrected T1 map assessment shows promise to improve the accuracy of T1 measurements but needs further validation in a larger dataset. This technique may become useful for objective evaluation of myocardial fibrosis.