Angular Super-Resolution in Diffusion MRI with a 3D Recurrent Convolutional Autoencoder
This work addresses a domain-specific problem for clinical MRI applications by providing an incremental improvement in angular super-resolution techniques.
The paper tackles the problem of limited angular resolution in diffusion MRI due to scanning time constraints by developing a 3D recurrent convolutional autoencoder for angular super-resolution, showing that it achieves the lowest error rates compared to baselines, with the greatest relative performance in very low angular resolution domains.
High resolution diffusion MRI (dMRI) data is often constrained by limited scanning time in clinical settings, thus restricting the use of downstream analysis techniques that would otherwise be available. In this work we develop a 3D recurrent convolutional neural network (RCNN) capable of super-resolving dMRI volumes in the angular (q-space) domain. Our approach formulates the task of angular super-resolution as a patch-wise regression using a 3D autoencoder conditioned on target b-vectors. Within the network we use a convolutional long short term memory (ConvLSTM) cell to model the relationship between q-space samples. We compare model performance against a baseline spherical harmonic interpolation and a 1D variant of the model architecture. We show that the 3D model has the lowest error rates across different subsampling schemes and b-values. The relative performance of the 3D RCNN is greatest in the very low angular resolution domain. Code for this project is available at https://github.com/m-lyon/dMRI-RCNN.