Promises and pitfalls of imaging-based biomarkers in motor neuron diseases

Although neuroimaging in motor neuron diseases (MNDs) continues to generate important novel academic insights, the translation of novel radiological protocols into viable biomarkers remains challenging.

A multitude of technological advances contribute to the success of academic imaging in MND such as the availability of high-field MRI platforms, novel imaging techniques, quantitative spinal cord protocols to whole-brain spectroscopy. International collaborations, protocol harmonization efforts, open-source image analysis suites also fuel developments in the field. Despite the success of academic neuroimaging in MND, the meaningful interpretation of radiological data from single patients and accurate classification into relevant diagnostic, phenotypic and prognostic categories remain challenging. Appraising accruing disease burden over the short follow-up intervals typically used in pharmacological trials is also notoriously difficult.

Although we acknowledge the academic achievements of large descriptive studies, an unmet priority of neuroimaging in MND is the development of robust diagnostic, prognostic and monitoring applications to meet the practical demands of clinical decision-making and pharmacological trials. A paradigm shift from group-level analyses to individual-level data interpretation, accurate single-subject classification and disease-burden tracking is therefore urgently needed to distil raw spatially coded imaging data into practical biomarkers.