Mikkel Schöttner

The paper evaluates how well population-level connectome harmonics represent individual brains across different dimensions and finds they can be useful despite not perfectly matching individual data.

This paper shows that individualized connectome harmonics reveal a hierarchical organization linking brain structure, function, and cognition across 875 participants.

This paper identifies optimal graph signal processing features for subject identification through large-scale comparative analysis of parameter choices and feature types.

This study applies deep learning to resting-state fMRI time series to predict 58 human traits, finding comparable performance to kernel-based methods with some advantages in robustness to training set size reductions.