Deep-MI Lab

← Research

p3CNN

Cortical surface parcellation using projection-based 3D convolutional neural networks on spherical meshes

Spherical coordinate systems have become a standard for analyzing human cortical neuroimaging data. Surface-based signals such as curvature, folding patterns, functional activations, or estimates of myelination define relevant cortical regions. Surface-based deep learning approaches such as spherical CNNs, however, primarily focus on classification and cannot yet achieve satisfactory accuracy in segmentation tasks.

To perform surface-based segmentation of the human cortex, we introduce and evaluate a 2D parameter space approach with view aggregation (p3CNN). We show that it outperforms the spherical CNN by a margin, increasing the average Dice similarity score for cortical segmentation to above 0.9.

Spherical Segmentation
Comparison of a spherical CNN (ugscnn, middle) and our proposed view-aggregation on 2D spherical parameter spaces (p3CNN, right) for cortical segmentation (bottom row).

News

May 2020 — Congratulations to Leonie Henschel for 1st place: Best Scientific Submission at the BVM Workshop 2020.