Invasive fungal infection (IFI) has emerged as an important cause of life-threatening opportunistic respiratory complication in severely immunocompromised patients. Early recognition of IFI indicating clinical outcome may be important for decisions concerning the strategy and duration of antifungal treatment, surgical intervention, and monitoring fungal manifestations¹. However, critical gaps in knowledge remain regarding diagnostic tools for assessing response to facilitate proper management of these infections. The purpose of this study was to determine whether intravoxel incoherent motion (IVIM) –derived parameters and apparent diffusion coefficient (ADC) could act as imaging biomarkers for predicting antifungal treatment response.This study was approved by the local ethics committee and informed consent was obtained from all participants. Forty-six consecutive patients (mean age, 33.9 ± 13.0 y) with newly diagnosed IFI in the lung according to ORTC/MSG criteria were prospectively enrolled. All patients underwent diffusion-weighted magnetic resonance (MR) imaging at 3.0 T MR scanner (Achieva TX, Philips Healthcare, Best, Netherlands) using 11 b values (0, 20, 50, 80, 110, 150, 200, 400, 600, 800, 1000 sec/mm²). Scanning parameters were as follows: TR/TE= 2100/55 ms, FOV= 375 mm × 300 mm; NEX= 2; slice thickness= 5 mm. A bi-exponential signal decay can be obtained and the two-step method was used for voxel wise fitting². Pseudodiffusion coffiecient D*, perfusion fraction f, and the diffusion coefficient D were calculated using the IVIM model. ADC was calculated using a mono-exponential model. A freehand region of interest (ROI) was manually outlined the largest pulmonary IFI lesion on ADC and IVIM parametric maps. Patients were stratified into favorable (n=32) and unfavorable response (n=14) groups based on follow-up for determination of the predictive powers of IVIM parameters using Student t test and receiver operating characteristic (ROC) curve analyses.DWI signal decay curves in favorable response group were found to be biexponential fit in comparison with the monoexponential fit for the unfavorable response group (Fig.1). D was consistently lower than ADC in all tissues. f values were significantly lower in the unfavorable response group (12.6%±4.4%) than in the favorable response group (30.2%±8.6%) (Z=4.989, P<0.001). However, the ADCtotal, D, and D* were not significantly different between the two groups (P>0.05, Fig.2). Receiver operating characteristic curve analyses showed f to be a significant predictor for differentiation(AUC=0.967), with a sensitivity of 93.8% and a specificity of 92.9%.In the present study, we attempted to validate the IVIM-derived perfusion and diffusion parameters to predict antifungal treatment response. Our study clarified the different perfusion characteristics of IFI lesions with favorable response and unfavorable response on the basis of the IVIM biexponential model. We found that the obtained f value was significantly lower in patients with unfavorable response than in those with favorable response, suggesting that f may be a biomarker for prediction of antifungal treatment. In this work, the decrease of f in patients with unfavorable response may well be caused by invasion of large pulmonary arteries and thrombosis formation in the IFI area as well as by concordant hypoperfusion³.