Multi-Task Deep Supervision on Attention R2U-Net for Brain Tumor Segmentation.

Shiqiang Ma,Jijun Tang,Fei Guo

doi:10.3389/fonc.2021.704850

Abstract

Accurate automatic medical image segmentation technology plays an important role for the diagnosis and treatment of brain tumor. However, simple deep learning models are difficult to locate the tumor area and obtain accurate segmentation boundaries. In order to solve the problems above, we propose a 2D end-to-end model of attention R2U-Net with multi-task deep supervision (MTDS). MTDS can extract rich semantic information from images, obtain accurate segmentation boundaries, and prevent overfitting problems in deep learning. Furthermore, we propose the attention pre-activation residual module (APR), which is an attention mechanism based on multi-scale fusion methods. APR is suitable for a deep learning model to help the network locate the tumor area accurately. Finally, we evaluate our proposed model on the public BraTS 2020 validation dataset which consists of 125 cases, and got a competitive brain tumor segmentation result. Compared with the state-of-the-art brain tumor segmentation methods, our method has the characteristics of a small parameter and low computational cost.

Highlights

Brain tumors are the most common primary malignant tumors of the brain caused by the canceration of glial cells in the brain and spinal cord
With the development of convolutional neural networks, the brain tumor automatic segmentation technology based on deep learning had achieved a high segmentation accuracy
We introduce the evaluation criteria for the brain tumor segmentation task, and report the results consisting of the ablation experiment and comparison with the state-of-theart methods

Summary

Introduction

Brain tumors are the most common primary malignant tumors of the brain caused by the canceration of glial cells in the brain and spinal cord. Brain tumors have the characteristics of high morbidity and mortality. Automatic segmentation technology of brain tumor can assist professional doctors to diagnose brain lesions and provide imaging technical support for the diagnosis and treatment of brain tumor patients. With the development of convolutional neural networks, the brain tumor automatic segmentation technology based on deep learning had achieved a high segmentation accuracy. The location of brain tumor regions and accurate segmentation of tumor edges have always been the difficulties of deep learning methods. In order to obtain accurate segmentation results, deep learning methods usually require a numerous parameters and a long calculation time, which leads to extremely high demands on the hardware. It is of great significance to develop a simple and efficient network architecture

Methods

Results

Conclusion