Interpretable Gallbladder Ultrasound Diagnosis: A Lightweight Web-Mobile Software Platform with Real-Time XAI

The system achieves 99.85% accuracy while remaining lightweight — with only 2.24M parameters, the AI model is small enough to run on mobile devices and web browsers, making advanced diagnostic tools accessible in resource-limited settings where gallbladder ultrasound expertise may be scarce.

Explainable AI visualizations support clinical decision-making — the software doesn't just provide a diagnosis; it shows doctors why it reached that conclusion through XAI visualizations, enabling clinicians to verify the AI's reasoning and maintain their role as the final decision-maker rather than blindly trusting the system.

Real-time diagnosis at the point of care improves accessibility — by deploying the tool as both web and mobile applications, the system can be used directly during ultrasound examinations, potentially reducing delays in diagnosis and making specialized gallbladder disease detection available in clinics without on-site ultrasound specialists.

The system distinguishes between nine disease types plus normal cases — rather than simply flagging "abnormal," the AI can classify specific gallbladder conditions, which could help guide appropriate treatment decisions and reduce unnecessary referrals or procedures.

Transparency and interpretability are built into the design — the emphasis on XAI and real-time explanations reflects responsible AI principles by ensuring the system augments rather than replaces clinical judgment, allowing healthcare providers to understand and question the AI's recommendations.