Three Deep Learning Techniques for 3D diffusion MRI Image Enhancement

In this talk, we discuss three deep learning techniques to improve the
image quality of 3D diffusion MRI Images. We first introduce a novel
low-memory method, which allows us to control the GPU memory usage
during training therefore allowing us to handle the processing of
3-dimensional, high-resolution, multi-channeled medical
images. Secondly we present the first multi-task learning approach in
data harmonization, where we integrate information from multiple
acquisitions to improve the predictive performance and learning
efficiency of the training procedure. Thirdly we present an extension
of the transposed convolution, where we learn both the offsets of
target locations and a blur to interpolate the fractional
positions. All three techniques can be applied in other image-related
paradigms.