How mohtadaayesha/pneumonia-detection works
This repository is a notebook-based chest-X-ray classification experiment. Its principal workflow loads labelled images, explores and preprocesses them, trains a custom CNN and a VGG16 transfer-learning model, compares validation performance, then selects a threshold and evaluates VGG16 on the test set. The README documents these stages and evaluation capabilities; no notebook source excerpts were available, so the graph reflects documented relationships only.
Main entry points: Pneumonia_Detection.ipynb VGG16 transfer learning · images Analysis figures
Components 15
Data preparation
- Load and encode
Pneumonia_Detection.ipynb - Exploratory analysis
Pneumonia_Detection.ipynb - Resize and preprocess
Pneumonia_Detection.ipynb - Training augmentation
Pneumonia_Detection.ipynb
Model training
- Custom CNN
Pneumonia_Detection.ipynb - VGG16 transfer learning
Pneumonia_Detection.ipynb
Evaluation
- Validation comparison
Pneumonia_Detection.ipynb - Threshold selection
Pneumonia_Detection.ipynb - Test evaluation
Pneumonia_Detection.ipynb - Error analysis
Pneumonia_Detection.ipynb - Classification metrics
Pneumonia_Detection.ipynb - Analysis figures
images
Other components
- Researcher
- Google Drive
- Chest X-ray splits
Connections 21
- Researcher → Load and encode: runs notebook
- Load and encode → Google Drive: reads dataset
- Google Drive → Chest X-ray splits: provides splits
- Load and encode → Exploratory analysis: inspects data
- Load and encode → Resize and preprocess: loads and encodes
- Resize and preprocess → Training augmentation: augments training
- Resize and preprocess → Custom CNN: feeds images
- Training augmentation → Custom CNN: feeds training
- Resize and preprocess → VGG16 transfer learning: feeds images
- Training augmentation → VGG16 transfer learning: feeds training
- Custom CNN → Validation comparison: validation predictions
- VGG16 transfer learning → Validation comparison: validation predictions
- Validation comparison → Threshold selection: compares models
- VGG16 transfer learning → Threshold selection: provides probabilities
- Threshold selection → Test evaluation: selects threshold
- VGG16 transfer learning → Test evaluation: provides predictions
- Test evaluation → Classification metrics: computes scores
- Test evaluation → Error analysis: supports review
- Exploratory analysis → Analysis figures: produces plots
- Validation comparison → Analysis figures: produces plots
- Test evaluation → Analysis figures: produces plots