Improving Cytoarchitectonic Segmentation of Human Brain Areas with Self-supervised Siamese Networks

Abstract

Cytoarchitectonic parcellations of the human brain serve as anatomical references in multimodal atlas frameworks. They are based on analysis of cell-body stained histological sections and the identification of borders between brain areas. The de-facto standard involves a semi-automatic, reproducible border detection, but does not scale with high-throughput imaging in large series of sections at microscopical resolution. Automatic parcellation, however, is extremely challenging due to high variation in the data, and the need for a large field of view at microscopic resolution. The performance of a recently proposed Convolutional Neural Network model that addresses this problem especially suffers from the naturally limited amount of expert annotations for training. To circumvent this limitation, we propose to pre-train neural networks on a self-supervised auxiliary task, predicting the 3D distance between two patches sampled from the same brain. Compared to a random initialization, fine-tuning from these networks results in significantly better segmentations. We show that the self-supervised model has implicitly learned to distinguish several cortical brain areas -a strong indicator that the proposed auxiliary task is appropriate for cytoarchitectonic mapping.

Cytoarchitectonic Characterization and Functional Decoding of Four New Areas in the Human Lateral Orbitofrontal Cortex

“…Tangential sectioning of the cytoarchitecture represented a limitation to any type of analysis in 2D. However, alternative methods for cortical parcellation are under development, which use, for example, deep convolutional networks (Spitzer et al, 2018). In combination with high-resolution 3D models such as the Big Brain (Amunts et al, 2013), they open a new perspective to map the brain, independently on the angle of physical sectioning.…”

Efficient and Effective Training of COVID-19 Classification Networks With Self-Supervised Dual-Track Learning to Rank

“…To deal with the deficiency of annotated data, researchers attempted to exploit useful information from the unlabeled data with unsupervised approaches [28], [29]. More recently, the self-supervised learning, as a new paradigm of unsupervised learning, attracts increasing attentions from the community.…”

“…Zhang et al [29] defined a proxy task that sorted the 2D slices extracted from the conventional 3D CT and magnetic resonance imaging (MRI) volumes, to pre-train the neural networks for the fine-grained body part recognition (the target task). Spitzer et al [28] proposed to pre-train neural networks on a selfsupervised learning task, i.e., predicting the 3D distance between two patches sampled from the same brain, for the better segmentation of brain areas (the target task). Zhuang et al [12] proposed to pre-train 3D networks by playing a Rubik's cube game, which can be seen as an extension of 2D jigsaw puzzles [35].…”

Probabilistic Atlases to Enforce Topological Constraints

“…The affine transforms are used to warp the atlases and feed the features of the warped atlases to the decoder. This differs significantly from earlier approaches [14,15] that rely on pre-registered atlases. We validate our approach on segmentation of synaptic junctions in Electron Microscopy (EM) images and optic nerve head segmentation in retinal fundus images.…”