Anatomical labeling of brain CT scan anomalies using multi-context nearest neighbor relation networks

TL;DR

This paper introduces a deep learning method combining local and global context with relation networks and a novel nearest neighbors training strategy to accurately label brain CT scan regions anatomically.

Contribution

It proposes a new training approach for relation networks using nearest neighbors, improving accuracy in anatomical labeling of brain CT scans.

Findings

Enhanced accuracy over baseline models

Effective use of local and global context

Improved training efficiency

Abstract

This work is an endeavor to develop a deep learning methodology for automated anatomical labeling of a given region of interest (ROI) in brain computed tomography (CT) scans. We combine both local and global context to obtain a representation of the ROI. We then use Relation Networks (RNs) to predict the corresponding anatomy of the ROI based on its relationship score for each class. Further, we propose a novel strategy employing nearest neighbors approach for training RNs. We train RNs to learn the relationship of the target ROI with the joint representation of its nearest neighbors in each class instead of all data-points in each class. The proposed strategy leads to better training of RNs along with increased performance as compared to training baseline RN network.