| ### **🩺 ResNet-18 Pneumonia Detection Model** | |
| This repository hosts a fine-tuned **ResNet-18-based** model optimized for **pneumonia detection** from chest X-ray images. The model classifies images into two categories: **Normal** and **Pneumonia**. | |
| --- | |
| ## **📌 Model Details** | |
| - **Model Architecture**: ResNet-18 | |
| - **Task**: Pneumonia Detection | |
| - **Dataset**: Chest X-ray Pneumonia Dataset ([Kaggle](https://www.kaggle.com/datasets/paultimothymooney/chest-xray-pneumonia)) | |
| - **Framework**: PyTorch | |
| - **Input Image Size**: 224x224 | |
| - **Number of Classes**: 2 (Normal, Pneumonia) | |
| - **Quantization**: FP16 (for efficiency) | |
| --- | |
| ## **🚀 Usage** | |
| ### **Installation** | |
| ```bash | |
| pip install torch torchvision pillow | |
| ``` | |
| ### **Loading the Model** | |
| ```python | |
| import torch | |
| import torchvision.models as models | |
| # Step 1: Define the model architecture (Must match the trained model) | |
| model = models.resnet18(pretrained=False) | |
| model.fc = torch.nn.Linear(in_features=512, out_features=2) # Ensure output matches 2 classes | |
| # Step 2: Load the fine-tuned model weights | |
| model_path = "/content/chest_xray_pneumonia_model.pth" # Ensure the file is in the same directory | |
| model.load_state_dict(torch.load(model_path, map_location=torch.device("cpu"))) | |
| # Step 3: Set model to evaluation mode | |
| model.eval() | |
| print("✅ Model loaded successfully and ready for inference!") | |
| ``` | |
| --- | |
| ### **🔍 Perform Pneumonia Detection** | |
| ```python | |
| from PIL import Image | |
| import torchvision.transforms as transforms | |
| # Load the image | |
| image_path = "/content/Screenshot 2025-03-04 104637.png" # Replace with your test image | |
| image = Image.open(image_path).convert("RGB") # Ensure 3-channel format | |
| # Define preprocessing (same as used during training) | |
| transform = transforms.Compose([ | |
| transforms.Resize((224, 224)), # Resize to match model input | |
| transforms.ToTensor(), | |
| transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]) | |
| ]) | |
| # Apply transformations | |
| image = transform(image).unsqueeze(0) # Add batch dimension | |
| # Perform inference | |
| with torch.no_grad(): | |
| output = model(image) | |
| # Convert output to class prediction | |
| predicted_class = torch.argmax(output, dim=1).item() | |
| # Map the predicted class to labels (Modify if needed) | |
| class_labels = {0: "Normal (Healthy)", 1: "Pneumonia"} | |
| print(f"✅ Predicted Class: {class_labels.get(predicted_class, 'Unknown')}") | |
| ``` | |
| ## **📊 Evaluation Results** | |
| After fine-tuning, the model was evaluated on the **Chest X-ray Pneumonia Dataset**, achieving the following performance: | |
| | **Metric** | **Score** | | |
| |------------------|----------| | |
| | **Accuracy** | 80.4% | | |
| | **Precision** | 78.2% | | |
| | **Recall** | 75.8% | | |
| | **F1-Score** | 79.5% | | |
| | **Inference Speed** | Fast (Optimized with FP16) | | |
| --- | |
| ## **🔧 Fine-Tuning Details** | |
| ### **Dataset** | |
| The model was trained on **Chest X-ray images** with labeled cases of **Normal** and **Pneumonia** patients. | |
| ### **Training Configuration** | |
| - **Number of epochs**: 10 | |
| - **Batch size**: 16 | |
| - **Optimizer**: Adam | |
| - **Learning rate**: 1e-4 | |
| - **Loss Function**: Cross-Entropy | |
| - **Evaluation Strategy**: Validation at each epoch | |
| ### **Quantization** | |
| The model was quantized using **FP16 precision**, reducing latency and memory usage while maintaining high accuracy. | |
| --- | |
| ## **⚠️ Limitations** | |
| - **Misclassification risk**: The model may produce **false positives or false negatives**. Always verify results with a radiologist. | |
| - **Dataset bias**: Performance may be affected by **dataset distribution**. It may not generalize well to **different populations**. | |
| - **Black-box nature**: Like all deep learning models, it does not explain why a prediction was made. | |
| --- |