Home / Current Issue / Paper 1717806
A Lightweight Convolutional Neural Network for Automated Ependymoma Detection from Brain MRI
Subject area: Science,Engineering and Technology · Area of research: Convolutional Neural Network
DOI: https://doi.org/10.64388/IREV9I11-1717806
Abstract
Ependymoma is a rare tumour of the central nervous system, which can be successfully diagnosed by Magnetic Resonance Imaging (MRI) in order to provide treatment in initial stages. Automated diagnostic assistance is important since manual examination of MRI scans is labor-intensive and plagued with observer variability. This research offers the suggestion of a lightweight Convolutional Neural Network (CNN) which aims to classify brain MRI scans as ependymoma or non-tumor. The framework is trained and tested on a publicly available dataset of MRIs (3,264 images) split (with the help of stratified 70/15/15 train-validation-test dataset) and 5-fold cross validation. There are approximately 0.9 million parameters in the proposed network architecture, which ensures computational efficiency. Experimental evaluation has shown that the performance is good, which reaches 94.1% accuracy, 94.9% accuracy, 94.3% recall and F1-score, and 0.968 AUC value when given the test data. Comparative analysis against transfer learning models such as Resnet-50, VGG16 and MobileNetV2 proves that the proposed method achieves competitive results with a significantly lower formality of parameters. A lightweight Web-based deployment system coupled with PostgreSQL is implemented for structured prediction logging.
Keywords
Ependymoma, Brain Magnetic Resonance Imaging, Convolutional Neural Network, Deep Learning, Medical Image Classification, Computer Aided Diagnosis.
How to cite this paper
@article{1717806,
author = {Sneha M, Navinkumar D, Surya Prakash M, Siva K},
title = {A Lightweight Convolutional Neural Network for Automated Ependymoma Detection from Brain MRI},
journal = {Iconic Research And Engineering Journals},
year = {2026},
volume = {9},
number = {11},
pages = {1876-1883},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1717806.pdf},
abstract = {Ependymoma is a rare tumour of the central nervous system, which can be successfully diagnosed by Magnetic Resonance Imaging (MRI) in order to provide treatment in initial stages. Automated diagnostic assistance is important since manual examination of MRI scans is labor-intensive and plagued with observer variability. This research offers the suggestion of a lightweight Convolutional Neural Network (CNN) which aims to classify brain MRI scans as ependymoma or non-tumor. The framework is trained and tested on a publicly available dataset of MRIs (3,264 images) split (with the help of stratified 70/15/15 train-validation-test dataset) and 5-fold cross validation. There are approximately 0.9 million parameters in the proposed network architecture, which ensures computational efficiency. Experimental evaluation has shown that the performance is good, which reaches 94.1% accuracy, 94.9% accuracy, 94.3% recall and F1-score, and 0.968 AUC value when given the test data. Comparative analysis against transfer learning models such as Resnet-50, VGG16 and MobileNetV2 proves that the proposed method achieves competitive results with a significantly lower formality of parameters. A lightweight Web-based deployment system coupled with PostgreSQL is implemented for structured prediction logging.},
keywords = {Ependymoma, Brain Magnetic Resonance Imaging, Convolutional Neural Network, Deep Learning, Medical Image Classification, Computer Aided Diagnosis. },
month = {May},
doi = {https://doi.org/10.64388/IREV9I11-1717806}
}