Motivation: Subpopulation universal pulses (UPs) have shown promise as a compromise between one-size-fits-all UPs and tailored solutions. However, their practical application was not straightforward. Goal(s): Design an automated online image-processing pipeline for head shape clustering and pulse selection, providing subpopulation 5kT-inversion UPs. Approach: Implementation of a Python pipeline integrated into the image reconstruction routine, using localizer scans to classify head shapes and positions. The pipeline selects one of five precomputed subpopulation UPs or defaults to the generic UP without extending scan time. Results: The pipeline successfully classified head shapes and selected appropriate pulses, resulting in improved contrast of MP2RAGE images. Impact: The novel head clustering and pulse selection pipeline enables subpopulation universal pulses in clinical practice. It enables pulse selection for each head shape and position without extending scan time, facilitating personalized and efficient ultra-high field MRI.
Tyshchenko et al. (Tue,) studied this question.