📷 Upload Chest X-Ray
Click to upload or drag & drop
PNG • JPEG • Frontal PA view recommended
📊 AI Analysis Result
Risk Level–
Model Confidence–
Decision Threshold–
File–
🤖 AI Attention Map — Grad-CAM
Shows which lung regions drove the prediction
Red/warm = high AI attention • Blue/cool = low attention
Low attentionHigh attention
Original chest X-ray (CLAHE enhanced)
Original X-Ray
Grad-CAM Attention Map
🔍 What the AI Found & Why
🎯 Prediction Reasoning
📚 TB Radiological Features Guide
- Upper lobe infiltrates — most common TB location (right > left upper lobe)
- Cavitation — air-filled cavities, highly specific for active TB
- Consolidation — opacification of lung tissue, suggests active infection
- Miliary pattern — tiny nodules throughout both lungs (severe TB)
- Pleural effusion — fluid around lung, can be TB-associated
- Lymphadenopathy — enlarged lymph nodes visible on X-ray
⚠️ Model Confidence Interpretation
- Prob < 30% — Low risk. No significant TB radiological pattern detected.
- Prob 30–60% — Inconclusive. Ambiguous features present. Clinical judgement needed.
- Prob > 60% — High risk. Clear TB-consistent radiological features detected.
📋 Recommended Next Steps
🏥 Clinician Review — Human in the Loop
The AI result above is a decision support tool only. Please review the X-ray, Grad-CAM, and AI findings, then record your clinical decision below.
Patient Information
Clinician Information
Clinical Decision
SIGHTAI Decision Record
SIGHTAI Clinical Decision Record
🤖 AI Prediction
Diagnosis–
TB Probability–
Risk Level–
Threshold–
🏥 Clinician Decision
Clinician–
Facility–
AI Agreement
Final Diagnosis–
👥 Patient
Patient ID–
Age–
Sex–
File–
📌 Action Plan
Recommended–
Urgency–
Review ID–
CLINICAL NOTES
No notes recorded.
⚠️ Medical Disclaimer
SIGHTAI is an AI-powered decision-support tool intended to assist qualified healthcare professionals only.
It does not constitute a clinical diagnosis. All AI results must be reviewed and confirmed by a licensed
clinician before any treatment decision is made. The Grad-CAM heatmap indicates model attention regions
and does not replace formal radiological interpretation.