Fully Automatic Brain Tumor Segmentation using End-To-End Incremental Deep Neural Networks in MRI images

Abstract

Background and Objective: Nowadays, getting an efficient Brain Tumor Segmentation in Multi-Sequence MR images as soon as possible, gives an early clinical diagnosis, treatment and follow-up. The aim of this study is to develop a new deep learning model for the segmentation of brain tumors. The proposed models are used to segment the brain tumors of Glioblastomas (with both high and low grade). Glioblastomas have four properties: different sizes, shapes, contrasts, in addition, Glioblastomas appear anywhere in the brain.Methods: In this paper, we propose three end-to-end Incremental Deep Convolutional Neural Networks models for fully automatic Brain Tumor Segmentation. Our proposed models are different from the other CNNs-based models that follow the technique of trial and error process which does not use any guided approach to get the suitable hyper-parameters. Moreover, we adopt the technique of Ensemble Learning to design a more efficient model. For solving the problem of training CNNs model, we propose a new training strategy which takes into account the most influencing hyper-parameters by bounding and setting a roof to these hyper-parameters to accelerate the training.Results: Our experiment results reported on BRATS-2017 dataset. The proposed deep learning models achieve the state-of-the-art performance without any post-processing operations. Indeed, our models achieve in average 0.88 Dice score over the complete region. Moreover, the efficient design with the advantage of GPU implementation, allows our three deep learning models to achieve brain segmentation results in average 20.87 seconds.Conclusions: The proposed deep learning models are effective for the segmentation of brain tumors and allow to obtain high accurate results. Moreover, the proposed models could help the physician experts to reduce the time of diagnostic.

Brain tumor detection and classification: A framework of marker‐based watershed algorithm and multilevel priority features selection

“…The BRATS 2013 and BRATS 2015 data sets are utilized for evaluation the presented method and provide improved performance. Bennaceur, Saouli, Akil, and Kachouri () introduced a deep learning model for brain tumor segmentation from MRI images. The introduced model is different from other CNN models in terms of error and trial procedures.…”

Deep Learning in Omics Data Analysis and Precision Medicine

“…This task is very challenging because MRI data consist of 3D images where tumors are very different between patients, and in addition, they are very heterogeneous images depending on the device and experimental procedures employed (69). Several researchers addressed this challenge using CNNs (70)(71)(72) or SAEs (73).…”

Deep Convolutional Neural Networks for Brain Tumor Segmentation: Boosting Performance Using Deep Transfer Learning: Preliminary Results

“…Where in CNNs [4], we find a feature extractor with a bank of convolution layers, then pooling layers to make the images less sensitive and invariant to small translations, then the last step in CNNs is a classifier (in general Softmax layer) that classifies each pixel into one of a set of classes. After the breakthrough in 2012 of AlexNet [5] model that outperformed the state-of-the-art methods in the field of object recognition, many methods obtained high results in many fields especially in medical field such as [6], [7], [8], [11], [9], [10]. In general, these methods are trained on 4 types of MRI images: Flair, T1, T1c, and T2.…”

“…In general, these methods are trained on 4 types of MRI images: Flair, T1, T1c, and T2. Our ongoing work is based on our previous work [9], [15]. In this paper, we are focusing on two major issues: (1) false positive regions -where the model predicts non-tumor regions as tumor regions but in fact they are not-, (2) false negative regions -where the model classifies some regions as non-tumor regions but in fact they are.…”

“…In this paper, we are focusing on two major issues: (1) false positive regions -where the model predicts non-tumor regions as tumor regions but in fact they are not-, (2) false negative regions -where the model classifies some regions as non-tumor regions but in fact they are. In [9], [15] we addressed the problem of false positive regions by two steps: we used a global threshold for each slice to remove small non-tumoral regions based on connected-components, then in second step, to enhance the post-processing step more we used a morphological opening operator. Despite the success of these two post-processing steps, further steps are required to improve the segmentation results.…”