Dr. Emma Robinson is a Senior Lecturer (Assoc. Professor) at King’s College London. Her development of the Multimodal Surface Matching (MSM) software for cortical surface registration has been instrumental to the development of the Human Connectome Project’s multimodal parcellation of the human cortex. She is currently developing interpretable machine learning models to aid in the personalized prediction of disease progression. In this interview, Dr.Robinson describes the advantages of interpretable machine learning models, and the methodological challenges she faced during the development of this framework.

Her approach to identifying disease-related changes in individual brain scans attempts to circumvent two of the limitations of traditional approaches: (1) the over-reliance on population averages, and (2) the opacity of “black-box” machine learning algorithms such as deep neural networks. In addition, Dr. Robinson shared that, following her extensive experience working on the Human Connectome Project, she realized that traditional image registration methods may not be sufficient for individualized predictions.

Finally, Dr. Robinson shared how her relationship with her mentors shaped the trajectory of her current career. Her mentors not only guided her on the application of computational methods to neuroscience, but also encouraged her to develop her own methods.

At OHBM 2023, Dr. Robinson will present how her work contributes to improved personalized predictions of cortical features in patient populations and how interpretable machine learning approaches can enhance precision.