Evaluation of fitting models for prostate tissue characterization using extended-range b-factor diffusion-weighted imaging

摘要: Purpose To compare the fitting and tissue discrimination performance of biexponential, kurtosis, stretched exponential, gamma distribution models for high b-factor diffusion-weighted images in prostate cancer. Methods Diffusion-weighted with 15 b-factors ranging from b = 0 to 3500 s/mm2 were obtained 62 cancer patients. Pixel-wise signal decay fits each model evaluated Akaike Information Criterion (AIC). Parameter values determined within normal index lesion. Their potential differentiate cancerous was investigated through receiver operating characteristic analysis comparison Gleason score. Results The biexponential slow diffusion fraction fslow , apparent kurtosis coefficient ADCK excess factor K differ significantly among peripheral zone (PZ), transition (TZ), tumor PZ, TZ. Biexponential result lowest AIC, indicating a superior fit. Maximum areas under curve (AUCs) all ranged 0.93 0.96 PZ 0.95 0.97 Similar AUCs also (ADC) monoexponential fit sub-range up 1250 . For exponential models, single parameters yield highest AUCs, whereas linear combinations produce AUCs. Parameters AUC show trend differentiating low score, no such ability. Conclusion All including lower-b sub-range, achieve similar tissue. model, which is favored statistically, appears provide insight into disease-related microstructural changes. Magn Reson Med 79:2346-2358, 2018. © 2017 International Society Magnetic Resonance Medicine.