Fast Mitochondria Detection for Connectomics

TL;DR

This paper introduces a fast, accurate, fully automatic mitochondria detection method for connectomics data using a modified U-Net, enabling real-time analysis suitable for large datasets and surpassing previous methods in speed and accuracy.

Contribution

A novel modified U-Net architecture for mitochondria detection that achieves high accuracy and real-time processing speeds on large connectomics datasets.

Findings

Achieved a Jaccard index of up to 0.90.

Inference times lower than 16ms per 512x512px image tile.

Ranks first for real-time detection among existing methods.

Abstract

High-resolution connectomics data allows for the identification of dysfunctional mitochondria which are linked to a variety of diseases such as autism or bipolar. However, manual analysis is not feasible since datasets can be petabytes in size. We present a fully automatic mitochondria detector based on a modified U-Net architecture that yields high accuracy and fast processing times. We evaluate our method on multiple real-world connectomics datasets, including an improved version of the EPFL mitochondria benchmark. Our results show an Jaccard index of up to 0.90 with inference times lower than 16ms for a 512x512px image tile. This speed is faster than the acquisition speed of modern electron microscopes, enabling mitochondria detection in real-time. Our detector ranks first for real-time detection when compared to previous works and data, results, and code are openly available.