Improved characterization of diffusion in normal and cancerous prostate tissue through optimization of the restriction spectrum imaging signal model

BackgroundOptimizing a restriction spectrum imaging (RSI) model for the prostate could lead to improved characterization of diffusion in the prostate and better discrimination of tumors. PurposeTo determine optimal apparent diffusion coefficients (ADCs) for prostate RSI models and evaluate the number of tissue compartments required to best describe diffusion in prostate tissue. Study TypeRetrospective. Population/SubjectsForty-six patients who underwent an extended MRI examination for suspected prostate cancer; 23 had prostate tumors and 23 had no detectable cancer. Field strength/Sequence3T multi-shell diffusion weighted sequence. AssessmentRSI models with 2-5 tissue compartments were fit to multi-shell DWI data from the prostate to determine optimal compartmental ADCs. Signal contributions from the different tissue compartments were computed using these ADCs and compared between normal tissues (peripheral zone, transition zone, seminal vesicles) and tumors. Statistical TestsThe Bayesian Information Criterion (BIC) was used to evaluate the optimality of different RSI models. Model-fitting residual (as percent variance) was recorded to assess the models goodness-of-fit and whether it varied between anatomical regions of the prostate. Two-sample t-tests (=0.05) were used to determine the statistical significance of any differences observed in compartmental signal-fraction between normal prostate tissue and tumors. ResultsThe lowest BIC was observed from the 4-compartment model. Optimal ADCs for the 4 compartments were 5.2e-4, 1.9e-3, 3.0e-3, and >>3.0e-3 mm2/s. Tumor tissue showed the largest reduction in fitting residual by increasing model order. Prostate tumors had a significantly (P<<0.05) greater proportion of signal from compartments 1 and 2 than normal tissue. Tumor conspicuity in compartment 1 increased substantially with model order. Data ConclusionAmong the examined RSI models, the 4-compartment model best described the diffusion-signal characteristics of the prostate. Compartmental signal fractions revealed by such a model may improve discrimination between cancerous and benign prostate tissue.