DVNet: A Memory-Efficient Three-Dimensional CNN for Large-Scale Neurovascular Reconstruction

TL;DR

This paper introduces DVNet, a memory-efficient 3D CNN designed for large-scale neurovascular image segmentation, enabling detailed analysis of brain microarchitecture with improved performance and reduced memory usage.

Contribution

The paper presents a novel 3D CNN architecture with skip connections that reduces memory complexity and enhances segmentation accuracy for large neurovascular datasets.

Findings

Outperforms prior architectures on benchmark datasets

Enables efficient organ-scale neurovascular segmentation

Provides quantitative metrics for neurovascular analysis

Abstract

Maps of brain microarchitecture are important for understanding neurological function and behavior, including alterations caused by chronic conditions such as neurodegenerative disease. Techniques such as knife-edge scanning microscopy (KESM) provide the potential for whole organ imaging at sub-cellular resolution. However, multi-terabyte data sizes make manual annotation impractical and automatic segmentation challenging. Densely packed cells combined with interconnected microvascular networks are a challenge for current segmentation algorithms. The massive size of high-throughput microscopy data necessitates fast and largely unsupervised algorithms. In this paper, we investigate a fully-convolutional, deep, and densely-connected encoder-decoder for pixel-wise semantic segmentation. The excessive memory complexity often encountered with deep and dense networks is mitigated using skip connections, resulting in fewer parameters and enabling a significant performance increase over prior architectures. The proposed network provides superior performance for semantic segmentation problems applied to open-source benchmarks. We finally demonstrate our network for cellular and microvascular segmentation, enabling quantitative metrics for organ-scale neurovascular analysis.