Image registration and subtraction in dynamic magnetic resonance lymphangiography (MRL) of the legs

Dynamic contrast-enhanced magnetic resonance lymphangiography (DCE-MRL) is regularly reported as unable to depict lymphatic vessels in healthy limbs. In this study, we aim to improve lymph vessel conspicuity with appropriate registration and subtraction of a reference baseline image. Five unaffected individuals and a single unilateral primary lymphoedema patient were recruited to undergo fat suppressed 3D T1 weighted spoiled gradient echo imaging of the lower limbs at 3.0 T. Images were quality assessed by two physicians and a medical physicist following registration via one of six registration pipelines, and subtraction of the first post-contrast dynamic image (PC1). Wilcoxon non-parametric testing was performed to compare image quality ranking vs the unregistered images and inter--rater reliability estimated using intraclass correlation coefficient. Signal enhancement curves were also computed in lymphatic vessels for two participants. Subtraction images were considered to improve lymphatic visibility, and three registration pipelines significantly (p < 0.05) outranked those without registration. Those registered to PC1 with an affine and elastic approach were rated best quality (p = 0.006). Moderate inter-rater reliability was observed (intraclass correlation coefficient = 0.71) and signal enhancement behaviour appears affected by registration when motion is evident across the DCE-MRL series. We therefore conclude that lymphatic vessel visibility in DCE-MRL images can be improved with registration and baseline subtraction.