BrainAid: Efficient Brain Tumor Detection Using Lightweight CNNs and Explainable AI for Resource-Constrained Environments

High accuracy with practical deployment: BrainAid achieved 98% accuracy, precision, and F1 score in classifying four types of brain tumors (glioma, meningioma, pituitary tumor, and no-tumor), and was successfully deployed on a Raspberry Pi, demonstrating that highly accurate AI diagnostic tools can work in resource-limited settings without requiring expensive computing infrastructure.

Addresses real-world diagnostic inconsistency: The paper identifies that manual expert evaluation of MRI scans suffers from inconsistency due to different professionals using different criteria, which can delay diagnoses—BrainAid offers a standardized, objective alternative that could reduce these variations in clinical practice.

Explainability built into the system: By integrating Explainable AI (XAI) alongside the neural networks, BrainAid makes its diagnostic decisions interpretable to medical professionals, addressing a key concern in responsible AI deployment where doctors need to understand why the system made a particular classification rather than treating it as a "black box."

Efficiency enables accessibility: The use of lightweight CNNs and Vision Transformers allows the system to run on edge devices like Raspberry Pi, meaning diagnostic AI tools could potentially be deployed in under-resourced hospitals, clinics, and regions that cannot afford high-end computing infrastructure, improving healthcare equity.

Real-time diagnostic support potential: The system enables quick, on-site tumor classification that can provide immediate diagnostic support to medical professionals, potentially reducing wait times for initial assessments and enabling faster treatment decisions.