[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-80130":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":11,"openIssues":13,"contributorsCount":13,"subscribersCount":13,"size":13,"stars1d":13,"stars7d":13,"stars30d":13,"stars90d":13,"forks30d":13,"starsTrendScore":13,"compositeScore":14,"rankGlobal":9,"rankLanguage":9,"license":9,"archived":15,"fork":15,"defaultBranch":16,"hasWiki":17,"hasPages":15,"topics":18,"createdAt":9,"pushedAt":9,"updatedAt":19,"readmeContent":20,"aiSummary":21,"trendingCount":13,"starSnapshotCount":13,"syncStatus":22,"lastSyncTime":23,"discoverSource":24},80130,"MultiClass-LungDisease-Detection-Using-XAI","Stevia-S\u002FMultiClass-LungDisease-Detection-Using-XAI","Stevia-S","Explainable deep learning framework for multi-class lung disease detection from CT scan images using ResNet50, VGG16 feature fusion, and Grad-CAM visualization.",null,"Python",54,1,0,37.9,false,"main",true,[],"2026-06-12 04:01:26","\n# 🫁 Explainable Deep Learning-Based Lung Disease Detection using ResNet50–VGG16 Fusion\n\n🧠 Python • TensorFlow • Transfer Learning • ResNet50-VGG16 Fusion • Grad-CAM • Medical AI\n\nAn end-to-end deep learning system that detects lung diseases from CT scan images using a CNN-based architecture and provides **explainable AI (XAI) visualizations** to improve interpretability in medical diagnosis.\n\n---\n\n## 📊 Project Overview\n\nMedical image diagnosis requires high accuracy and interpretability. Traditional deep learning models often achieve strong performance but lack transparency in decision-making.\n\nThis project presents an **Explainable Deep Learning-Based Lung Disease Detection System** using a **ResNet50–VGG16 Fusion Architecture** with **Transfer Learning** and **Grad-CAM Explainability**.\n\nThe system is designed to:\n\n- Classify CT scan images\n- Detect multiple lung diseases\n- Visualize affected regions using Grad-CAM heatmaps\n- Improve interpretability in AI-assisted medical diagnosis\n\nIt is designed to simulate real-world **AI-assisted radiology systems** for medical image analysis.\n\n---\n\n### 🔍 Key Features\n\n- ResNet50–VGG16 Fusion Model\n- Transfer Learning-Based Feature Extraction\n- Grad-CAM Explainable AI Visualization\n- Streamlit-Based Web Application\n- Multi-Class CT Scan Classification\n\n---\n\n### 🩺 Diseases Detected\n\n- COVID\n- Normal\n- Pneumonia\n\n---\n\n# 🎯 Problem Statement\n\nManual analysis of CT scans is:\n\n* Time-consuming\n* Dependent on expert availability\n* Prone to human error in complex cases\n\nThis project addresses these challenges by building a deep learning model that automatically:\n\n* Detects lung disease patterns\n* Provides interpretable visual explanations\n* Assists medical decision-making\n\n---\n\n# 🏗️ Pipeline Architecture\n\n```\nCT Scan Image Input\n        │\n        ▼\n┌──────────────────────┐\n│ Image Preprocessing  │  ← Resize, normalization\n└────────┬─────────────┘\n         │\n         ▼\n┌──────────────────────┐\n│   ResNet50 + VGG16   │  ← Feature extraction\n└────────┬─────────────┘\n         │\n         ▼\n┌──────────────────────┐\n│ Classification Head  │  ← Softmax output\n└────────┬─────────────┘\n         │\n         ▼\n┌──────────────────────┐\n│ Explainability (XAI) │  ← Grad-CAM heatmaps\n└────────┬─────────────┘\n         │\n         ▼\n Prediction + Visualization Output\n```\n\n---\n\n# 🧩 System Components\n\n## 🔬 Image Processing\n\n* CT scan resizing and normalization\n* Noise reduction for better feature learning\n\n---\n\n## 🧠 Deep Learning Model\n\n* CNN-based architecture\n* Automatic feature extraction from CT images\n* Multi-class classification\n\n---\n\n## 🔥 Explainable AI (XAI)\n\n* Grad-CAM heatmaps\n* Highlights infected lung regions\n* Improves model transparency\n\n---\n\n## 🎯 Output System\n\n* Disease prediction\n* Confidence score\n* Visual explanation map\n\n---\n\n# 🧠 Model Architecture\n\n| Layer        | Description          |\n| ------------ | -------------------- |\n| Conv2D       | Feature extraction   |\n| MaxPooling   | Spatial reduction    |\n| Dropout      | Prevent overfitting  |\n| Flatten      | Vector conversion    |\n| Dense Layers | Classification       |\n| Softmax      | Output probabilities |\n\n---\n\n# 📊 Model Performance\n\n| Metric           | Score                |\n| ---------------- | -------------------- |\n| Accuracy         | ~90% – 97%           |\n| Precision        | High                 |\n| Recall           | High                 |\n| Interpretability | Enabled via Grad-CAM |\n\n> Performance varies based on dataset quality and training configuration.\n\n---\n\n# 🔍 Explainable AI (XAI)\n\nGrad-CAM is used to:\n\n* Highlight infected lung regions\n* Show decision-making areas of CNN\n* Improve trust in predictions\n* Assist radiologists in validation\n\n---\n\n# ⚠️ Key Challenges Addressed\n\n* Similar visual patterns between lung diseases\n* Reducing false negatives in medical diagnosis\n* Improving interpretability of deep learning models\n* Handling limited labeled medical datasets\n\n---\n\n# 🚀 Getting Started\n\n## 📦 Installation\n\n```bash id=\"inst2\"\npip install tensorflow keras numpy matplotlib opencv-python scikit-learn\n```\n\n---\n\n## ▶️ Run Project\n\n```bash id=\"run3\"\npython train.py\npython predict.py\n```\n\n---\n\n# 📁 Project Structure\n\n```\nCT-Lung-Disease-XAI\u002F\n│\n├── dataset\u002F\n├── models\u002F\n├── utils\u002F\n│   ├── gradcam.py\n│\n├── outputs\u002F\n│   ├── confusion_matrix.png\n│   ├── gradcam_result.png\n│\n├── train.py\n├── predict.py\n├── requirements.txt\n└── README.md\n```\n\n---\n\n# 📈 Results\n\n* CNN achieves strong classification performance\n* Grad-CAM consistently highlights infected regions\n* Reliable multi-class lung disease prediction\n* Suitable for AI-assisted diagnosis systems\n\n---\n\n# 🔮 Future Improvements\n\n* Web-based CT scan upload system\n* Real-time hospital integration\n* Mobile deployment\n* Larger dataset training\n* Advanced XAI (Grad-CAM++)\n\n---\n\n# 🧠 Key Concepts Demonstrated\n\n* Convolutional Neural Networks (CNN)\n* Medical image preprocessing\n* Multi-class classification\n* Explainable AI (Grad-CAM)\n* Deep learning model evaluation\n\n---\n\n# 📜 License\n\nThis project is for academic and research purposes only.\n\n---\n\n\n","该项目是一个基于深度学习的多类别肺部疾病检测系统，能够从CT扫描图像中识别多种肺部疾病，并提供可解释的人工智能（XAI）可视化以提高医学诊断的可解释性。核心功能包括使用ResNet50与VGG16融合架构进行特征提取、通过迁移学习增强模型性能以及采用Grad-CAM技术生成热图来可视化受影响区域。此项目适用于需要高准确性和透明度的医学影像分析场景，特别是辅助放射科医生快速准确地识别如COVID-19、普通肺炎等肺部疾病的场合。该系统不仅提高了诊断效率，还通过直观的可视化结果增强了医生对AI决策的信任。",2,"2026-06-11 03:59:21","CREATED_QUERY"]