Advanced Deep Learning Framework for Brain Tumor Segmentation and Classification
Machines and Algorithms · 16.03.2016 · 5 (01) · 13-21
DOI: 10.66108/mna.v5i01.95 ↗Kurz gefasst
Fachartikel in der Zeitschrift Machines and Algorithms, online erschienen am 16.03.2016. Verfasst von Usama Aslam und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Magnetic Resonance Imaging (MRI) of the brain determines which brain tumors are necessary to be accurately segmented and classified to be effectively diagnosed and treated. Analysis by hand is tedious and subject to inter-observer error. This paper will present a multi-stage deep learning architecture, which utilizes the Convolutional Neural Networks (CNNs) model and implements denoising, image enhancement, skull stripping, and feature optimization to increase robustness and generalization. It tested itself on publicly available MRI datasets by 10-fold cross-validation. The experimental findings indicate that, the CNN attained an accuracy of 84.5, true positive percentage of 0.845 and ROC of 0.897 which is higher than that of the LSTM model at 83.71%. In addition, a hybrid CNNLSTM model (CABGD) based on a combination of spatial and sequential features achieved an 85 percent accuracy and illustrated improved classification performance. Reliability to the proposed approach was demonstrated based on misclassification analysis. These results represent the possibilities of deep learning, especially hybrid systems, as a clinical decision support system to detect and classify brain tumors automatically.
Zitieren
APA
Aslam, U., Muntaha, & Zahra, S. (2016). Advanced Deep Learning Framework for Brain Tumor Segmentation and Classification. Machines and Algorithms, 5(01), 13-21. https://doi.org/10.66108/mna.v5i01.95
Vancouver
Aslam U, Muntaha, Zahra S. Advanced Deep Learning Framework for Brain Tumor Segmentation and Classification. Machines and Algorithms. 2016;5(01):13-21. doi:10.66108/mna.v5i01.95
BibTeX
@article{Aslam2016,
author = {Aslam, Usama and Muntaha and Zahra, Shan},
title = {Advanced Deep Learning Framework for Brain Tumor Segmentation and Classification},
journal = {Machines and Algorithms},
year = {2016},
volume = {5},
number = {01},
pages = {13--21},
doi = {10.66108/mna.v5i01.95}
}
EUVIMED