Assessing the Role of Random Forests in Medical Image Segmentation

TL;DR

This paper compares random forest methods to deep neural networks for medical image segmentation, showing that random forests can achieve comparable results and offer a GPU-free alternative.

Contribution

It introduces and evaluates two random forest approaches for medical image segmentation, demonstrating their competitive performance against deep neural networks.

Findings

Deep neural networks outperform random forests but the latter achieve similar high performance.

Random forests provide a GPU-free alternative for medical image segmentation.

Results are validated on PhC-C2DH-U373 and retinal imaging datasets.

Abstract

Neural networks represent a field of research that can quickly achieve very good results in the field of medical image segmentation using a GPU. A possible way to achieve good results without GPUs are random forests. For this purpose, two random forest approaches were compared with a state-of-the-art deep convolutional neural network. To make the comparison the PhC-C2DH-U373 and the retinal imaging datasets were used. The evaluation showed that the deep convolutional neutral network achieved the best results. However, one of the random forest approaches also achieved a similar high performance. Our results indicate that random forest approaches are a good alternative to deep convolutional neural networks and, thus, allow the usage of medical image segmentation without a GPU.