BT Diagnosis and Segmentations using Designing of Ontological Framework

• Published in: 2024 4th International Conference on Advance Computing and Innovative Technologies in Engineering (ICACITE) pp. 27 - 30
• Main Authors: Vijayaragavan, P., Krishnammal, P. Muthu, Saranya, A., Naik, Pramod Kumar, Dheidan, Ali Hasan, Hameed, Hind
• Format: Conference Proceeding
• Language: English
• Published: IEEE 14-05-2024
• Subjects: Brain modeling; Brain MR image classification; BraTS MRI dataset; Computational modeling; Convolutional neural networks; DeepTumor framework; Fuzzy C-means (FCM); Image segmentation; Magnetic resonance imaging; Ontologies; ontology; Surgery; tumor classification; tumor segmentation
• DOI: 10.1109/ICACITE60783.2024.10616804

As the brain is so sensitive, it is important for medical professionals and patients to accurately segment brain tumors in medical images. To effectively target specific brain areas during surgical interventions, precision is required. Precise segmentation also makes multi-class tumor classification easier. This work aims to advance three main areas of brain MR image processing: tumor region segmentation, tumor classification, and brain MR image classification. By providing a multi-stage method for classifying glioma tumors into four separate groups-Edema, Necrosis, Enhancing, and Non-enhancing-the suggested framework, DeepTumor, addresses these issues. Firstly, two deep Convolutional Neural Network (CNN) and ontology models are introduced for the binary brain MR image classification (Tumorous and Non-tumorous). A seven-layer model with fifteen thousand trainable parameters and an improved eight-layer model with seventy thousand trainable parameters are among them. In the next step, a more accurate and efficient method based on fuzzy C-means (FCM) is suggested for tumor segmentation in brain magnetic resonance imaging.