Lightweight CNN achieves 99%+ accuracy in brain tumor MRI classification

A new paper from researchers Md Fahimul Kabir Chowdhury and Jannatul Ferdous introduces a computationally efficient Convolutional Neural Network (CNN) for multi-class brain tumor classification from MRI images. The lightweight model targets four categories: gliomas, meningiomas, pituitary tumors, and healthy (no tumor) instances. It was rigorously evaluated on two publicly accessible datasets from Figshare and Kaggle. The architecture leverages efficient feature extraction and optimized training strategies to deliver high performance while keeping the parameter count low.

The results are impressive: the CNN achieved classification accuracies of 99.03% and 99.28% on Dataset 1 and Dataset 2 respectively, with ROC area under the curve scores of 99.88% and 99.94%. Critically, the model utilizes significantly fewer parameters than cutting-edge pre-trained architectures like DenseNet201, MobileNetV2, VGG19, Xception, InceptionV3, and ResNet50, yet consistently outperforms them in accuracy with reduced computational overhead. This makes the proposed model a strong candidate for practical deployment in clinical settings where compute resources may be limited. The findings highlight the potential for AI-assisted brain tumor diagnosis that is both accurate and computationally affordable.