This is a simple deep learning-based project that classifies cell images as Parasitized or Uninfected. The entire pipeline—from preprocessing to model building, training, and evaluation—has been implemented in a single Jupyter Notebook.
Built this as part of exploring how deep learning can help in medical image analysis and early disease detection.
The notebook does the following:
- Loads and preprocesses the cell image dataset
- Performs data augmentation to improve generalization
- Builds and trains a Convolutional Neural Network (CNN) model
- Evaluates the model using accuracy and loss metrics
- Makes predictions on new/unseen cell images
- Python (3.x)
- Jupyter Notebook
Libraries:
- tensorflow, keras
- numpy
- matplotlib
- seaborn
1️⃣ Clone the repo or download the notebook.
2️⃣ Install the required Python libraries:
pip install tensorflow keras numpy matplotlib seaborn3️⃣ Prepare the dataset:
- Download a cell image dataset (e.g., NIH malaria dataset).
- Organize it in folders:
Parasitized/andUninfected/inside a main dataset folder.
4️⃣ Launch Jupyter Notebook:
jupyter notebook5️⃣ Open CNN_Cancer.ipynb and run the cells one by one.
| File | Description |
|---|---|
| CNN_Cancer.ipynb | The main notebook with full CNN pipeline |
| README.md | You’re reading it :) |
The CNN model achieves good accuracy in distinguishing between parasitized and uninfected cells.
You can further:
- Try different CNN architectures
- Experiment with hyperparameter tuning
- Test on completely new images to check generalization
Automated cell image classification can support faster and more accurate disease detection, potentially saving time for doctors and labs. This is a small step toward using deep learning to assist in healthcare and diagnostics 🩸🔬.