Lung Cancer: Short-Term Reproducibility of Intravoxel Incoherent Motion Parameters and Apparent Diffusion Coefficient at 3T.

Abstract

To prospectively evaluate the short-term reproducibility of intravoxel incoherent motion (IVIM) parameters and apparent diffusion coefficient (ADC) in lung cancer patients. In all, 50 patients (50 lesions) underwent free-breathing diffusion-weighted imaging (DWI) (b = 0, 300, 800 s/mm2 ) and IVIM (10 b-values, 0-1000 s/mm2 ) scans twice (0.5-1-hour interval) at 3T. Regions of interests (ROIs) were drawn on ADC maps and IVIM images to derive the mean ADC value and IVIM parameters D, D*, and f. Intra- and interobserver, test-retest reproducibility were assessed with intraclass correlation coefficients (ICCs), within coefficient-of-variations (WCVs), and Bland-Altman analysis. The effects of type, size, and location of lung lesions were compared with WCVs. D and ADC showed good intraobserver reproducibility and interobserver agreement, while D* and f showed relatively larger variability (WCV 20.89-34.97%). The test-retest reproducibility of D and ADC were good (ICC 0.763-0.837; WCV 11.12-12.55%), while those of D* and f were relatively poor (ICC 0.604-0.842; WCV 36.54-72.62%). D and ADC had decreased reproducibility for lesions <2 cm (WCV 14.20%, 16.34%, respectively) and for lesions in the lower lung zones (WCV 16.52%, 14.78%, respectively). f had decreased reproducibility in central lung cancers (WCV 50.11%) and lesions >2 cm (WCV 42.64%). D* had even worse reproducibility in peripheral lung cancers (WCV 84.11%) and lesions in the lower lung zones (WCV 80.84%). If the change in ADC, D, D*, and f values is less than ∼31%, 34%, 170%, and 130%, respectively, it may be caused by measurement error. The type, size, and location of lung lesions have an effect on measurement errors. 1 Technical Efficacy Stage: 1 J. Magn. Reson. Imaging 2018;47:1003-1012.