Development and Evaluation of an Automated Pipeline for ASL Scan Planning and Labeling Plane Positioning

Motivation: Accurate labeling plane positioning is crucial for reliable ASL perfusion measurements. Manual positioning using angiography are complex and inconsistent. Traditional methods relying solely on structural distance often fail to achieve symmetric labeling efficiency. Goal(s): Develop an automated ASL planning pipeline incorporating angiography for reliable and reproducible perfusion measurements. Approach: The pipeline integrates deep learning for FOV planning, artery segmentation via VB-net, and gradient descent for plane positioning optimization. It was evaluated against traditional methods by experienced physicians on 217 subjects. Results: The pipeline successfully positioned labeling planes in 97% of cases, versus 47% for traditional methods, and demonstrated better performance on tortuous arteries. Impact: The proposed automated ASL planning pipeline achieves both standardized brain orientation and optimized labeling plane positioning, potentially reducing labeling inefficiencies on tortuous arteries and minimizing false-positive hypoperfusion cases.