MRI Breast Tumor Segmentation Using Different Encoder and Decoder CNN Architectures

Mahmoudi Mahmoudi,Benjelloun Benjelloun,Larhmam Larhmam,Adoui Adoui

doi:10.3390/computers8030052

Mahmoudi Mahmoudi, Benjelloun Benjelloun

Open Access

https://doi.org/10.3390/computers8030052

Copy DOI

Journal: Computers	Publication Date: Jun 29, 2019
Citations: 70	License type: CC BY 4.0

Affiliation: University of Mons

Abstract

Breast tumor segmentation in medical images is a decisive step for diagnosis and treatment follow-up. Automating this challenging task helps radiologists to reduce the high manual workload of breast cancer analysis. In this paper, we propose two deep learning approaches to automate the breast tumor segmentation in dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) by building two fully convolutional neural networks (CNN) based on SegNet and U-Net. The obtained models can handle both detection and segmentation on each single DCE-MRI slice. In this study, we used a dataset of 86 DCE-MRIs, acquired before and after two cycles of chemotherapy, of 43 patients with local advanced breast cancer, a total of 5452 slices were used to train and validate the proposed models. The data were annotated manually by an experienced radiologist. To reduce the training time, a high-performance architecture composed of graphic processing units was used. The model was trained and validated, respectively, on 85% and 15% of the data. A mean intersection over union (IoU) of 68.88 was achieved using SegNet and 76.14% using U-Net architecture.

Highlights

Breast cancer is one of the most common cancers in the world
We propose two deep learning approaches to automate the breast tumor segmentation in dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) by building two fully convolutional neural networks (CNN) based on SegNet and U-Net
We used a dataset of 86 DCE-MRIs, acquired before and after two cycles of chemotherapy, of 43 patients with local advanced breast cancer, a total of 5452 slices were used to train and validate the proposed models

Summary

Introduction

Breast cancer is one of the most common cancers in the world. A high number of breast medical examinations have been collected, which allowed for the development of several medical imaging techniques, such as tumor localization and segmentation [1]. These tasks aim to separate the tumor from the normal breast tissue, which can provide valuable information for further analysis. The most used medical imaging modalities for breast cancer detection and diagnosis are mammography and magnetic resonance imaging (MRI). Mammography is more dedicated to the early stage detection of breast tumors. Deep learning-based techniques have been successfully applied to the analysis of mammograms thanks to the availability of a relatively large dataset [2]

Methods

Results

Discussion

Conclusion