Generation and Study of the Synthetic Brain Electron Microscopy Dataset for Segmentation Purpose

Abstract

Advanced microscopy technologies such as electron microscopy have opened up a new field of vision for biomedical researchers. The use of artificial intelligence methods for processing EM data is largely difficult due to the small amount of annotated data at the training stage. Therefore, we add synthetic images to an annotated real EM dataset or use a fully synthetic training dataset. In this work, we present an algorithm for the synthesis of 6 types of organelles. Based on the EPFL dataset, a training set of 860 real fragments 256x256 (ORG) and 6000 synthetic ones (SYN), as well as their combination (MIX), were generated. An experiment of training models for segmentation into 5 and 6 classes showed that, despite the imperfection of synthetic data, for an axon poorly represented in the training data set, the use of a synthetic data set improves the Dice metric from 0.3 on the original dataset to 0.8 on the mixed and synthetic datasets. The synthetic data strategy gives annotations for free, but shifts the effort to producing sufficiently realistic images.