How thanushamh/medical-image-classification-multicancer works
This notebook implements a deep-learning workflow for classifying medical images into eight cancer types. The documented stages include preprocessing and normalization, augmentation, training a CNN from scratch, and transfer learning with ResNet50 followed by fine-tuning. The fine-tuned model is evaluated with accuracy, precision, recall, F1-score, and a confusion matrix. The dataset is external and is not included in the repository. No notebook source excerpts were available, so the relationships shown here are based only on the README.
Main entry points: Multi_Cancer_Medical_Image_Classification.ipynb Preprocessing
Components 9
Image Preparation
- Preprocessing
Multi_Cancer_Medical_Image_Classification.ipynb - Data Augmentation
Multi_Cancer_Medical_Image_Classification.ipynb
Model Training
- Scratch CNN
Multi_Cancer_Medical_Image_Classification.ipynb - ResNet50 Transfer
Multi_Cancer_Medical_Image_Classification.ipynb - Fine-Tuning
Multi_Cancer_Medical_Image_Classification.ipynb
Evaluation
- Model Evaluation
Multi_Cancer_Medical_Image_Classification.ipynb - Classification Metrics
Multi_Cancer_Medical_Image_Classification.ipynb
Other components
- Analyst
- Kaggle Image Dataset
Connections 10
- Analyst → Preprocessing: classifies images
README - Kaggle Image Dataset → Preprocessing: provides images
README - Preprocessing → Data Augmentation: prepares images
README - Data Augmentation → Scratch CNN: augments training
README - Data Augmentation → ResNet50 Transfer: augments training
README - Preprocessing → Scratch CNN: normalizes inputs
README - Preprocessing → ResNet50 Transfer: normalizes inputs
README - ResNet50 Transfer → Fine-Tuning: fine-tunes layers
README - Fine-Tuning → Model Evaluation: evaluates model
README - Model Evaluation → Classification Metrics: reports scores
README