Breakthrough in Brain Tumor Diagnosis: A Cutting-Edge Hybrid Depthwise-Direct Acyclic Graph Network for MRI Image Classification

Felix Joseph X (Professor, Department of Electrical and Electronics Engineering, Loyola Institute of Technology and Science, Thovalai, Tamilnadu, India.)
Maithili Vijayakumar (Department of Artificial Intelligence, R. M. K. College of Engineering and Technology, Tiruvallur, Chennai, India.)
Sujatha Therese P (Department of Electrical and Electronics Engineering, Noorul Islam Centre for Higher Education, Kumaracoil, Tamilnadu, India.)
Josephin Shermila P (Associate Professor, Department of Artificial Intelligence and Data Science, R. M. K. College of Engineering and Technology, Tiruvallur, Chennai, India.)
Eugine Prince M (Assistant Professor, Department of Physics, S.T.Hindu College, Nagercoil, Tamilnadu, India.)
Maris Murugan T (Associate Professor, Department of Electronics and Instrumentation Engineering, Erode Sengunthar Engineering College, Erode, Tamilnadu, India.)

Journal of Applied Engineering and Technological Science (JAETS)
Vol. 6 No. 1 (2024): Journal of Applied Engineering and Technological Science (JAETS)

Both adults and children are at risk of dying from brain tumors. On the other hand, prompt and precise detection can save lives. Early detection is necessary for a proper diagnosis of a brain tumor, and magnetic resonance imaging (MRI) is often used in this context. To assist in the early diagnosis of sickness, neuro-oncologists have used Computer-Aided Diagnosis (CAD) in a number of ways. In this study, proposed a hybrid Depthwise-Direct Acyclic Graph Network (D-DAGNET)-based deep learning was developed to distinguish between cancers and non-tumors. Three processes make up this method: pre-processing, classification, and feature extraction. Pre-processing methods used in this study included contrast enhancement and image shrinking. The MRI picture is processed to get the wavelet and texture properties and used to build a classifier. Using MRI scans, the proposed hybrid Depthwise-Direct Acyclic Graph Network (D-DAGNET) model classifies two types of brain tumors: tumor and non-tumor. Performance criteria such as accuracy (ACC), specificity (SP), and sensitivity (SE) are used to assess the suggested hybrid Depthwise-Direct Acyclic Graph Network (D-DAGNET) model. Based on 850 images, the studies yielded a 99.54% categorization accuracy demonstrate that the suggested hybrid Depthwise-Direct Acyclic Graph Network (D-DAGNET) model beats the most advanced methods.