Medical Physics Seminar – Monday, March 27, 2023
Reproducibility of Liver ADC Measurements Using Motion Robust Diffusion Weighted Imaging
Timothy Allen
Graduate Research Assistant
Diffusion weighted imaging (DWI) is a magnetic resonance (MR) technique that probes tissue microstructure by sensitizing the image contrast to the mobility of water molecules. In abdominal applications, DWI plays an important clinical role for lesion detection. Additionally, quantitative metrics of molecular mobility, such as the Apparent Diffusion Coefficient (ADC), have tantalizing potential for lesion characterization, assessment of treatment response, and evaluation of diffuse disease such as liver fibrosis. However, DWI is highly sensitive to multiple technical challenges, particularly physiological motion that introduces bias and variability in ADC. Abdominal organs such as liver experience high levels of motion from cardiac and respiratory sources. The left lobe, which is directly below the heart, is particularly susceptible to motion artifacts and prone to unreliable ADC measurements. This technical challenge limits the current utility of DWI of the abdomen for clinical and research applications.
M1-optimized diffusion imaging (MODI) is a technique developed by our group that allows for motion robust DWI. Despite its promising early results, MODI has only been implemented on a high-performance 3.0T MR system. The feasibility of MODI on conventional 3.0T MR systems and 1.5T MR systems remains unknown. Also unknown is the reproducibility of ADC measurements using MODI across different MR scanners. Therefore, the purpose of this work is to demonstrate the feasibility of MODI using conventional 3.0T MR systems and 1.5T MR systems and to evaluate the reproducibility of liver ADC measurements across these systems. Upon successful completion, this work will advance the widespread dissemination of motion-robust DWI for clinical and research applications.
Automated MR Image Prescription of the Liver Using Deep Learning: Development, Evaluation, and Prospective Implementation
Ruiqi Geng
Graduate Research Assistant
This work developed a novel automated AI-based method for liver image prescription from a localizer and evaluated it in a large retrospective patient cohort (1,039 patients for training/testing), across pathologies, field strengths, and against radiologists' inter-reader reproducibility performance. AI-based 3D axial prescription achieved a S/I shift of <2.3 cm compared to manual prescription for 99.5% of test dataset. The AI method performed well across all sub-cohorts and better in 3D axial prescription than radiologists' inter-reader reproducibility performance. We successfully implemented the AI method on a clinical MR system, which demonstrated robust performance across localizer sequences.