Malaria Parasite Detection in Thin Blood Smear Images by Retraining Pretrained Convolutional Neural Networks (VGG19)
Domain : Computer Vision, Machine Learning
Sub-Domain : Deep Learning, Image Recognition
Techniques : Deep Convolutional Neural Network, Transfer Learning, VGG19
Application : Image Classification, Medical Imaging, Bio-Medical Imaging
- Detection of malarial parasites from thin Blood Smear images. Images were collected from Malaria screening research activity by National Institutes of Health (NIH).
- Employed VGG19 Deep Learning (Convolutional Neural Network) and fine-tuned the model weights in the entire network to distinguish infected from uninfected images. Used Tensorflow 2.0 for model training. Incrementally unfroze and tuned all layers in the network.
- Image augmentation and resizing of images were done on the fly during the training process.
- Attained a loss (categorical crossentropy) 0.159 and an accuracy 95.7% on the test data.
Dataset Name : Malaria Cell Images Dataset Original Dataset : Malaria Datasets - National Institutes of Health (NIH) Number of Classes : 2
Languages : Python
Tools/IDE : Jupyter. Notebook
Libraries : TensorFlow 2.0, VGG19
| Dataset | Training | Validation |
|---|---|---|
| Accuracy | 0.9206 | 0.9503 |
| Loss | 0.14285 | 0.1762 |
| Precision | 0.96 | |
| Recall | 0.96 | |
| ROC-AUC |
| Parameter | Value |
|---|---|
| Base Model | VGG19 |
| Optimizer | Stochastic Gradient Descent |
| Loss Function | Categorical Crossentropy |
| Learning Rate | 0.0001 |
| Batch Size | 32 |
| Number of Epochs | Round #1 & #2: 10 epochs, Round#3: 35 epochs |