---
language: en
license: mit
tags:
- image-classification
- active-learning
- medical-imaging
datasets:
- medmnist
metrics:
- loss
- accuracy
- area under the curve
---

# MedMNIST Active Learning Model

## Overview

This model is designed for image classification tasks within the medical imaging domain, specifically targeting the MedMNIST dataset. It employs a ResNet-50 architecture tailored for 28x28 pixel images and incorporates active learning strategies to enhance performance with limited labeled data.

## Model Architecture

- **Base Model:** ResNet-50
- **Modifications:**
  - Adjusted initial convolution layer to accommodate 28x28 input images.
  - Removed max pooling layer to preserve spatial dimensions.
  - Customized fully connected layer to output predictions for 9 classes.

## Training Procedure

### Training Hyperparameters

| Hyperparameter         | Value                  |
|------------------------|------------------------|
| Batch Size             | 53                    |
| Initial Labeled Size   | 3559                  |
| Learning Rate          | 0.01332344940133225    |
| MC Dropout Passes      | 6                     |
| Samples to Label       | 4430                  |
| Weight Decay           | 0.00021921795989143406 |

### Optimizer Settings

The optimizer used during training was Stochastic Gradient Descent(SDG), with the following settings and a Learning Rate Scheduler of ReduceLROnPlateau:
- `learning_rate = 0.01332344940133225`
- `momentum = 0.9`
- `weight_decay = 0.00021921795989143406`

The model was trained with float32 precision.

### Dataset
[PathMNIST](https://medmnist.com/)

### Data Augmentation
  - Random resized cropping
  - Horizontal flipping
  - Random rotations
  - Color jittering
  - Gaussian blur
  - RandAugment

### Active Learning Strategy

The active learning process was based on a mixed sampling strategy:
- **Uncertainty Sampling**: Monte Carlo (MC) dropout was used to estimate uncertainty.
- **Diversity Sampling**: K-means clustering was employed to ensure diverse samples.

## Evaluation

The model was evaluated on the validation set of PathMNIST. Key performance metrics include:

- **Accuracy:** 94.72%
- **Loss:** 0.2397
- **AUC:** 99.73%

## Graphs

The following plots illustrates the validation loss, validation accuracy, and validation auc over batches(number of iterations over the dataset) during the active learning process.

- **Validation Loss**
![Validation Loss](images/test_loss.png)
- **Validation Accuracy**
![Validation Accuracy](images/test_accuracy.png)
- **Validation AUC**
![Validation AUC](images/test_auc.png)

## Usage
All code for this model can be accessed in the following GitHub Repository:
[Allen Cheung Determined_AI_Hackathon](https://github.com/AllenCheung0213/Determined_AI_Hackathon)

To utilize this model:

1. **Install Dependencies:**
   Ensure the following Python packages are installed:
   - `torch`
   - `torchvision`
   - `medmnist`
   - `scikit-learn`
   - `determined`

   Install them using pip:
   ```bash
   pip install torch torchvision medmnist scikit-learn determined
   ```

2. **Load the Model:**
    ```python
    import torch
    from model import ResNet50_28

    model = ResNet50_28(num_classes=9)
    model.load_state_dict(torch.load('pytorch_model.bin'))
    model.eval()
    ```
    
3. **Inference:**
    ```python
    from torchvision import transforms
    from PIL import Image

    transform = transforms.Compose([
        transforms.Resize((28, 28)),
        transforms.ToTensor(),
        transforms.Normalize(mean=[0.5], std=[0.5])
    ])

    image = Image.open('path_to_image.jpg')
    input_tensor = transform(image).unsqueeze(0)
    output = model(input_tensor)
    prediction = output.argmax(dim=1).item()
    print(f"Predicted class: {prediction}")
    ```

## License

This project is licensed under the MIT License.

## Acknowledgements

- [MedMNIST Dataset](https://medmnist.com/)
- [Determined AI](https://determined.ai/)
- **Survey on Deep Active Learning**: Wang, H., Jin, Q., Li, S., Liu, S., Wang, M., & Song, Z. (2024). A comprehensive survey on deep active learning in medical image analysis. *Medical Image Analysis*, 95, 103201. [https://doi.org/10.1016/j.media.2024.103201](https://doi.org/10.1016/j.media.2024.103201)