Sep 2026· 2026 11th International Conference on Computer Science and Engineering (UBMK)· pp. 384-389· 0 citations· 22 references
TL;DR
An explainable framework using the Gazi Brains 2020 dataset, comprising 50 patients with high-grade glioma and 50 normal controls, is presented, which combines leakage-aware evaluation, quantitative explanation validation, shortcut testing, and external generalization analysis.
Abstract
Deep learning can support the detection of high-grade glioma in brain magnetic resonance images, but clinically useful systems also require reliable explanations. We present an explainable framework using the Gazi Brains 2020 dataset, comprising 50 patients with high-grade glioma and 50 normal controls. Skull-stripped fluid-attenuated inversion recovery, T1-weighted, and T2-weighted sequences were normalized, fused as three-channel images, and classified with an ImageNet-pretrained DenseNet121. Patient-level ten-fold cross-validation achieved 93.1% accuracy, a 92.8% F1 score, and a 97.3% area under the receiver operating characteristic curve. Two class-activation mapping methods were evaluated against expert tumor annotations; the gradient-based method achieved an insertion area under the curve of 0.962 and more specific localization. Brain and background masking indicated reliance on anatomical content, and external testing on the 2020 Brain Tumor Segmentation benchmark showed robust tumor sensitivity and localization under domain shift. The framework combines leakage-aware evaluation, quantitative explanation validation, shortcut testing, and external generalization analysis.
A deep learning (DL)-based solution to automate brain tumor detection using a customized convolutional neural network (CNN) architecture tailored for multi-class classification of glioma, meningioma, pituitary tumors, and non-tumor cases is proposed.
Ghaida Alsharef, R. Asiri, Lama Alsultan et al.· Bulletin of Electrical Engin...· 0 citations
Accurate brain tumor classification from magnetic resonance imaging (MRI) remains difficult because glioma, meningioma, and pituitary tumors may exhibit overlapping enhancement, morphology, and slice-dependent appearance. This study presents a controlled comparison of Vision Transformer, Data-efficient Image Transforme...
Tumor detection from MRI images is a crucial step that needs precision and fast diagnosis in order to yield positive results for the patient. Tumor classification via MRI scans manually by radiologists takes a lot of time and is vulnerable to human mistakes. In order to solve such problems, this research project sugges...
Hemavati, G. S, Arpan Pal et al.· 2026 Forth International Con...· 0 citations
Brain tumors require timely and reliable identification because their location and type can strongly influence clinical management. Owing to its ability to capture detailed anatomical structures without exposing patients to ionizing radiation, magnetic resonance imaging (MRI) has become a commonly preferred technique f...
R. Raj, N. Krishnamoorthy· 2026 7th International Confe...· 0 citations
Brain tumors constitute a significant global health challenge, and accurate, timely diagnosis is critical for effective treatment planning. Manual interpretation of magnetic resonance imaging (MRI) remains time-consuming, error-prone and subject to inter-observer variability. The objective of this study was to develop...
Adukwu Nasiru Ukwuteno, G. Obunadike, B. T. Mashi· Journal of Basics and Applie...· 0 citations
Accurate detection and segmentation of brain tumors from magnetic resonance imaging (MRI) are essential for early diagnosis, treatment planning, and outcome monitoring in neuro-oncology. Manual assessment is labor-intensive, subjective, and prone to inter- and intra-observer variability, motivating the development of a...
M. Alom, Udoy Roy Turza, Sayed Waliullah Oli et al.· International Conference on...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.