Research Projects

Diffusion weighted imaging (DWI) is a form of magnetic resonance imaging (MRI) that utilizes Brownian motion of water molecules within biological tissues. In oncological assessments, DWI provides the extent of cancer in whole body skeleton screening. Whole body DWI is performed from head to the mid thigh with images acquired using two to three diffusion weightings with several repetitions for better quality. Therefore, whole body scanning using DWI requires approximately one hour time. Accelerating Diffusion Weighted imaging has the potential to transform whole-body cancer screening by making it more practical and scalable in clinical settings. Recent developments in deep learning show its effectiveness in restoring under sampled imaging data. Therefore, deep learning has the potential to decrease the acquisition time of whole body DWI by reducing the number of repetitions in the scanning. This project aims to develop faster DWI models to enable more efficient whole-body cancer screening. The project will proceed through stages of model design, optimization, and testing. First, we will develop a model for accelerating whole-body scanning, which will restore the whole body scans acquired with less number of repetitions. Second, we will develop region-specific models which will enhance voxel level restoration. Then, we will enhance our models for low strength DWI scanners to reduce imaging artifacts. We will validate the models on a set of public datasets and one locally acquired dataset to ensure image fidelity and diagnostic accuracy. A clinical pilot study will follow, comparing the models against conventional DWI to confirm diagnostic sensitivity and specificity.The expected outcomes of the project include the validation of models for accelerated DWI, enhanced scanning efficiency, improved patient comfort, and substantial cost savings. Successfully adopting accelerated DWI has the potential to revolutionize whole-body cancer screening, making it more practical, accessible, and cost-effective for clinical use.