Automated segmentation of micro-CT images by deep learning and its application to comparative morphology

Abstract

Image segmentation is one of the most fascinating challenges of computer vision. A field of potential application is organismal biology, where researchers are increasingly using three-dimensional (3D) scanning which produces data-rich volumetric images for precise and comprehensive anatomical characterization. To date, the segmentation of anatomical structures remains a bottleneck to research, as it is commonly performed with highly tedious and time-consuming manual work. During recent years, however, machine learning methods are an emerging approach to overcoming this limitation, es- pecially with the use of deep learning techniques such as convolutional neural networks (CNNs), which proved to be very efficient and, as such, promising candidates for image segmentation. The main objective of this PhD project was to develop a pipeline for the fully-automated segmentation of anatomical structures in micro-computed tomog- raphy (micro-CT) images of insects using state-of-the-art deep learning methods. The restricted number of available high-resolution 3D labeled images necessitated the use of a CNN architecture that performs segmentation satisfactorily even with limited data; the U-Net architecture is such a CNN that has shown good performance in medical images using few annotated images. Ant brains were selected as the test case. Since no dataset of micro-CT images of ant brains existed for the current case study, a new extensive dataset was created across a wide variation of 94 ant species. Its existence can be of importance, as brain images of ants are similar to those of other insects; therefore, our dataset can act as a starting point for the development of a substantial library of micro-CT images of insects, and work as a pre-training dataset for future CNNs. Also, our network is generalizable for segmenting the whole neural system in full-body scans, and works in tests of distantly related and morphologically divergent insects (e.g., fruit flies). The latter suggests that algorithms such as our network can be applied generally across diverse taxa. The chosen species set was designed to be interesting for further evolutionary morphology analysis. Therefore, we used it to test the social brain hypothesis for ants, i.e., whether there is a connection between the brain investment and the sociality of each species. Volumetric statistical analysis was performed, also considering phylogenetic data; its results, however, did not validate the hypothesis.

Αυτόματη κατάτμηση εικόνων μικρο-υπολογιστικής τομογραφίας μέσω βαθείας μηχανικής εκμάθησης και η εφαρμογή της στη συγκριτική μορφολογία