Brain Tumor Segmentation and Radiomics Survival Prediction: Contribution to the BRATS 2017 Challenge

TL;DR

This paper presents a deep learning-based method for brain tumor segmentation that outperforms previous approaches on the BraTS datasets and also explores survival prediction using shape features with ensemble models.

Contribution

The authors develop a modified U-Net architecture with dice loss and data augmentation for improved brain tumor segmentation and integrate shape-based features for survival prediction.

Findings

Achieved state-of-the-art segmentation scores on BraTS datasets

Dice scores of 0.858 (whole tumor), 0.775 (core), 0.647 (enhancing tumor)

Survival prediction accuracy of 52.6% with ensemble models

Abstract

Quantitative analysis of brain tumors is critical for clinical decision making. While manual segmentation is tedious, time consuming and subjective, this task is at the same time very challenging to solve for automatic segmentation methods. In this paper we present our most recent effort on developing a robust segmentation algorithm in the form of a convolutional neural network. Our network architecture was inspired by the popular U-Net and has been carefully modified to maximize brain tumor segmentation performance. We use a dice loss function to cope with class imbalances and use extensive data augmentation to successfully prevent overfitting. Our method beats the current state of the art on BraTS 2015, is one of the leading methods on the BraTS 2017 validation set (dice scores of 0.896, 0.797 and 0.732 for whole tumor, tumor core and enhancing tumor, respectively) and achieves very good Dice scores on the test set (0.858 for whole, 0.775 for core and 0.647 for enhancing tumor). We furthermore take part in the survival prediction subchallenge by training an ensemble of a random forest regressor and multilayer perceptrons on shape features describing the tumor subregions. Our approach achieves 52.6% accuracy, a Spearman correlation coefficient of 0.496 and a mean square error of 209607 on the test set.