Motivation: Accurate luminal water fraction (LWF) estimation in prostate MRI is essential for enhancing diagnostic accuracy, yet traditional NNLS-based model struggle with noise in clinical settings. Goal(s): To evaluate and compare the accuracy, precision, and contrast of various LWF estimation models, focusing on multi-component (MC) models and NNLS-based models. Approach: Simulation studies were conducted across a range of signal-to-noise ratios (SNRs) and prior knowledge conditions, and further validated using in vivo MRI data from prostate cancer patients. Results: MC-G-4 and MC-D-2 models showed superior accuracy and precision, providing reliable LWF estimates and effectively differentiating tissue types. Impact: The MC-G-4 and MC-D-2 models significantly enhance luminal water fraction estimation in prostate cancer MRI, offering robust, accurate, and noise-resilient tools that improve lesion detection and characterization, thereby potentially leading to better patient outcomes in clinical practice.
Zhang et al. (Tue,) studied this question.