HIVE-Net: Centerline-Aware HIerarchical View-Ensemble Convolutional Network for Mitochondria Segmentation in EM Images
This work provides a more accurate and efficient method for mitochondria segmentation in EM images, which is crucial for researchers needing reliable morphological statistics, especially when training data is limited.
This paper addresses the problem of mitochondria segmentation in electron microscopy images, which often suffer from discontinuities and false positives. The authors propose HIVE-Net, a method that leverages centerline as a shape cue and a hierarchical view-ensemble convolution (HVEC) to learn 3D spatial contexts efficiently. The method achieves state-of-the-art accuracy and visual quality on two benchmarks with a greatly reduced model size and improved generalization.
Semantic segmentation of electron microscopy (EM) is an essential step to efficiently obtain reliable morphological statistics. Despite the great success achieved using deep convolutional neural networks (CNNs), they still produce coarse segmentations with lots of discontinuities and false positives for mitochondria segmentation. In this study, we introduce a centerline-aware multitask network by utilizing centerline as an intrinsic shape cue of mitochondria to regularize the segmentation. Since the application of 3D CNNs on large medical volumes is usually hindered by their substantial computational cost and storage overhead, we introduce a novel hierarchical view-ensemble convolution (HVEC), a simple alternative of 3D convolution to learn 3D spatial contexts using more efficient 2D convolutions. The HVEC enables both decomposing and sharing multi-view information, leading to increased learning capacity. Extensive validation results on two challenging benchmarks show that, the proposed method performs favorably against the state-of-the-art methods in accuracy and visual quality but with a greatly reduced model size. Moreover, the proposed model also shows significantly improved generalization ability, especially when training with quite limited amount of training data.