Enhanced dual-stage correlated diffusion imaging

Prostate cancer is the most common form of cancer and third leading cause of cancer death in Canadian men. Multi-parametric magnetic resonance imaging (mpMRI) has become a powerful non-invasive diagnostic tool for the detection of prostate cancer. Among mpMRI imaging modalities, diffusion-weighed imaging has shown the most promising results in accurate detection of prostate cancer. Introduced recently, correlated diffusion imaging (CDI) is a new form of diffusion imaging which accounts for the joint correlation of diffusion signal attenuation across multiple gradient pulse strengths and timings to improve the separability of cancerous and healthy tissues. Dual-stage CDI (D-CDI) is a newer generation of CDI where in contrast to CDI that does not capture anatomical information, an additional signal mixing stage between the correlated diffusion signal from the first signal mixing stage (CDI) and an auxiliary diffusion signal is performed to incorporate anatomical context. The core of D-CDI is a signal mixing algorithm that combines diffusion images at different b values to construct a single image. In this paper, we enhance the signal mixing algorithm to optimize the contribution of each single b-value image to maximize the separability of cancerous and healthy tissues. We evaluated the enhanced D-CDI (eD-CDI) using area under the ROC curve for datasets of 17 patient cases with confirmed prostate cancer and the results show that eD-CDI outperforms the original D-CDI as well as T2 weighted images and diffusion-weighed images used in the form of apparent diffusion coefficient maps.