Comparison of UNet, ENet, and BoxENet for Segmentation of Mast Cells in Scans of Histological Slices

TL;DR

This study compares UNet, ENet, and BoxENet architectures for segmenting mast cells in histological images, highlighting a trade-off between accuracy and computational speed.

Contribution

It introduces a modified ENet with box-convolutions and evaluates its performance against UNet and standard ENet for biomedical image segmentation.

Findings

ENet is 8-15 times faster than UNet.

ENet's accuracy is only 1-2% lower than UNet.

BoxENet's performance was analyzed but specific results are not detailed.

Abstract

Deep neural networks show high accuracy in theproblem of semantic and instance segmentation of biomedicaldata. However, this approach is computationally expensive. Thecomputational cost may be reduced with network simplificationafter training or choosing the proper architecture, which providessegmentation with less accuracy but does it much faster. In thepresent study, we analyzed the accuracy and performance ofUNet and ENet architectures for the problem of semantic imagesegmentation. In addition, we investigated the ENet architecture by replacing of some convolution layers with box-convolutionlayers. The analysis performed on the original dataset consisted of histology slices with mast cells. These cells provide a region forsegmentation with different types of borders, which vary fromclearly visible to ragged. ENet was less accurate than UNet byonly about 1-2%, but ENet performance was 8-15 times faster than UNet one.