Gradient nonlinearity correction to improve apparent diffusion coefficient accuracy and standardization in the american college of radiology imaging network 6698 breast cancer trial. Issue 4 (11th March 2015)

• Record Type: Journal Article
• Authors: Newitt, David C.; Tan, Ek T.; Wilmes, Lisa J.; Chenevert, Thomas L.; Kornak, John; Marinelli, Luca; Hylton, Nola
• Is Part Of: Journal of magnetic resonance imaging. Volume 42:Issue 4(2015)
• Page Start: 908
• Page End: 919
• Publication Date: 2015-03-11
• Subjects: Magnetic resonance imaging -- Periodicals; 616
• DOI: 10.1002/jmri.24883
• Languages: English
• ISSNs: 1053-1807

Abstract

<abstract abstract-type="main"> <title> <x xml:space="preserve">Abstract</x> </title> <sec id="jmri24883-sec-0001" sec-type="section"> <title>Purpose</title> <p>To evaluate a gradient nonlinearity correction (GNC) program for quantitative apparent diffusion coefficient (ADC) measurements on phantom and human subject diffusion‐weighted (DW) magnetic resonance imaging (MRI) scans in a multicenter breast cancer treatment response study</p> </sec> <sec id="jmri24883-sec-0002" sec-type="section"> <title>Materials and Methods</title> <p>A GNC program using fifth‐order spherical harmonics for gradient modeling was applied retrospectively to qualification phantom and human subject scans. Ice‐water phantoms of known diffusion coefficient were scanned at five different study centers with different scanners and receiver coils. Human in vivo data consisted of baseline and early‐treatment exams on 54 patients from four sites. ADC maps were generated with and without GNC. Regions of interest were defined to quantify absolute errors and changes with GNC over breast imaging positions.</p> </sec> <sec id="jmri24883-sec-0003" sec-type="section"> <title>Results</title> <p>Phantom ADC errors varied with region of interest (ROI) position and scanner configuration; the mean error by configuration ranged from 1.4% to 19.9%. GNC significantly reduced the overall mean error for all sites from 9.9% to 0.6% (<italic>P</italic> = 0.016). Spatial dependence of GNC was highest in the right‐left (RL) and anterior‐posterior (AP) directions. Human subject mean tumor ADC was reduced 0.2 to 12% by GNC at different sites. By regression, every 1‐cm change in tumor ROI position between baseline and follow‐up visits resulted in an estimated change of 2.4% in the ADC early‐treatment response measurement.</p> </sec> <sec id="jmri24883-sec-0004" sec-type="section"> <title>Conclusion</title> <p>GNC is effective for removing large, system‐dependent errors in quantitative breast DWI. GNC may be important in ensuring reproducibility in multicenter studies and in reducing errors in longitudinal treatment response measures arising from spatial variations in tumor position between visits. J. Magn. Reson. Imaging 2015;42:908–919.</p> </sec> </abstract>