Optimal Estimation of Diffusion in DW-MRI by High-Order MRF-Based Joint Deformable Registration and Diffusion Modeling

Evgenios N. Kornaropoulos, Evangelia I. Zacharaki, Pierre Zerbib, Chieh Lin, Alain Rahmouni, Nikos Paragios; Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2016, pp. 178-185

Over the last years, the apparent diffusion coefficient (ADC), computed from diffusion-weighted magnetic resonance (DW-MR) images, has become an important imaging biomarker for evaluating and managing patients with neoplastic or cerebrovascular disease. Standard methods for the calculation of ADC ignore the presence of noise and motion between successive (in time) DW-MR images acquired by changing the b-value. In order to accurately quantify the diffusion process during image acquisition, we introduce a method based on a high-order Markov Random Field (MRF) formulation that jointly registers the DW-MR images and models the spatiotemporal diffusion. Spatial smoothness on the ADC map, as well as spatiotemporal deformation smoothness, is imposed towards producing anatomically meaningful representations. The high-order dependencies in our MRF model are handled through Dual Decomposition. Performance of registration is compared to a state-of-the art registration approach in terms of obtained fitting error of the diffusion model in the core of the tumor. We also assess the clinical significance of the proposed method by examining the ability of the mean ADC value inside the tumor to act as imaging surrogate for lymphoma staging. Preliminary results reveal a marginally better performance of our method when compared against the standard ADC map used in clinical practice, which indicates its potential as a means for extracting imaging biomarkers.