3D brain slice classification and feature extraction using Deformable Hierarchical Heuristic Model. (October 2022)

• Record Type: Journal Article
• Authors: Sekaran, Ramesh; Munnangi, Ashok Kumar; Ramachandran, Manikandan; Gandomi, Amir H.
• Is Part Of: Computers in biology and medicine. Volume 149(2022)
• Publication Date: 2022-10
• Subjects: Brain tumors -- MRI -- 3D tumor -- VOI -- DHHM-DDRN; Medicine -- Data processing -- Periodicals; Biology -- Data processing -- Periodicals; 610.285
• DOI: 10.1016/j.compbiomed.2022.105990
• Languages: English
• ISSNs: 0010-4825

Abstract

Abstract: Brain tumors are the most frequently occurring and severe type of cancer, with a life expectancy of only a few months in most advanced stages. As a result, planning the best course of therapy is critical to improve a patient's ability to fight cancer and their quality of life. Various imaging modalities, such as computed tomography (CT), magnetic resonance imaging (MRI) and ultrasound imaging, are commonly employed to assess a brain tumor. This research proposes a novel technique for extracting and classifying tumor features in 3D brain slice images. After input images are processed for noise removal, resizing, and smoothening, features of brain tumor are extracted using Volume of Interest (VOI). The extracted features are then classified using the Deformable Hierarchical Heuristic Model-Deep Deconvolutional Residual Network (DHHM-DDRN) based on surfaces, curves, and geometric patterns. Experimental results show that proposed approach obtained an accuracy of 95%, DSC of 83%, precision of 80%, recall of 85%, and F1 score of 55% for classifying brain cancer features. Highlights: For more accurate prediction, additional class identifications were added to predict brain tumors in 3D images. The features in the region of the tumor processed brain image were extracted using Volume of Interest. Images were classified using the Deformable Hierarchical Heuristic Model based Deep Deconvolutional Residual Network. A parametric analysis was conducted to evaluate and compare the proposed technique with existing methods.