GCU Lahore AI Project

AI-Powered
Radiology.

Automated diagnostic system detecting 14 pathologies with clinical precision. Built with DenseNet121 & Explainable AI.

Interpretability Demo Portfolio Home
108K
NIH Dataset
14
Class Labels
0.9+
Max AUROC
Dense
Net121
Patient X-ray
Model Inference ID: 2841-X
Infiltration 89%
Pathological Insight

Bridging Global
Radiology Gaps.

In regions with limited specialist access, automated interpretation provides a vital second opinion. Our model prioritizes critical findings like Pneumothorax to ensure immediate clinical intervention.

Time Efficiency

Reduces diagnostic turnaround from hours to seconds.

Second Opinion

Serves as a reliable validation tool for medical staff.

Clinical Conditions

Targeting 14 distinct pathological findings from the NIH ChestX-ray8 dataset.

Atelectasis

Collapse of lung tissue affecting gas exchange.

Effusion

Fluid buildup between pleural layers.

Consolidation

Air sacs filled with fluid or inflammatory cells.

Pneumonia

Infection inflaming the air sacs in one or both lungs.

XAI: Seeing Through the AI.

GradCAM visualization identifies anatomical hotspots, ensuring the model isn't learning from artifacts or medical equipment.

Input
Input Image
Heatmap
GradCAM Heatmap