Authors: Ajay Sonawane, Pranav Babrekar, Aditya Pandagale, Himanshu Saindlya
Abstract: Brain tumours are life-threatening conditions that demand early and precise diagnosis to improve patient outcomes. While deep learning has significantly advanced automated medical imaging, conventional convolutional neural network (CNN) models often require large annotated datasets and intensive computation, limiting their applicability in clinical settings. In experiments, the quantum-augmented models achieved notable performance gains. The hybrid MobileNetV2 model achieved the highest validation accuracy of 95.79%, outperforming traditional CNN baselines while offering faster inference and reduced computational overhead. These results suggest that integrating quantum layers enhances feature representation and model robustness.
DOI: https://doi.org/10.5281/zenodo.19279172
