Malaria Parasite Species Classification Project

This project applies machine learning to classify malaria parasite species based on microscopic images, using a deep learning approach with the VGG16 architecture. Custom layers and image augmentation enhance the model’s performance and robustness.

Dataset

The dataset, available on Kaggle, is provided by Loddo et al. through the Malaria Parasite Image Database (MP-IDB). It contains labeled images for different malaria species. For more information:

• Kaggle Dataset: Malaria Parasite Species Dataset
• Full Dataset Description: MP-IDB Description

Citation

Loddo, A., Di Ruberto, C., Kocher, M., Prod’Hom, G. (2019). MP-IDB: The Malaria Parasite Image Database for Image Processing and Analysis. In: Lepore, N., Brieva, J., Romero, E., Racoceanu, D., Joskowicz, L. (eds) Processing and Analysis of Biomedical Information. SaMBa 2018. Lecture Notes in Computer Science(), vol 11379. Springer, Cham. https://doi.org/10.1007/978-3-030-13835-6_7

Project Structure

• ImageDataGenerator: Uses Keras for data augmentation, applying transformations like rotation, brightness adjustments, and random flips to increase image diversity during training.
• VGG16 Model: Implements transfer learning with the VGG16 architecture and custom fully connected layers with dropout regularization.
• Metrics: Model performance is measured using accuracy, AUC, precision, and recall.

Dependencies

• Python 3.7+
• TensorFlow (2.x)
• Keras
• Matplotlib
• Numpy
• JSON

Code Overview

Key Functions

• load_existing_results: Loads previously saved training results from a JSON file.
• save_results: Saves new metrics and hyperparameters to JSON for recording each training instance.
• get_data_generators: Creates data generators for training and validation using Keras' ImageDataGenerator with multiple transformations.
• build_vgg16_model: Builds and compiles a modified VGG16 model with custom layers and an Adam optimizer.
• train_and_evaluate_model: Trains the model, evaluates performance, saves the model, calculates metrics, and saves the model based on hyperparameters.

Main Code Block

The script initializes a set of hyperparameters, calls train_and_evaluate_model, prints metrics, and stores results.

Usage

1. Set the dataset directory path in dataset_dir.
2. Configure any desired hyperparameters in param_dict.
3. Run the script to train, evaluate, and save the model.
4. Results are saved in JSON format for easy reference.

Example

# Initialize hyperparameters
param_dict = {
    'learning_rate': 0.001,
    'batch_size': 32,
    'dropout_rate': 0.5,
    'activation_function': 'relu',
    'optimizer': 'adam'
}

# Train and evaluate the model
metrics = train_and_evaluate_model(**param_dict)
print(metrics)

# Save results
save_results(metrics)