AmpliVision

AmpliVision is an open source program designed to Automate AMPLI rapid test workflows. Mainly, It is able to load and process AMPLI rapid test images from a specified folder, find the grid and Ampli blocks in the image, read the diagnostic result for each Ampli block, generate a synthetic dataset of images, train a CNN model to classify AMPLI tests with generated images, then run outlier detection.

Table of Contents

• Installation
• Usage
• Pre-Defined Use Cases
  • Scan AMPLI test images and extract results
  • Train a CNN LENET model to classify AMPLI test images
  • Detect Outliers With Trained Model And PyOD
  • Test a trained CNN model in Predicting Never Seen AMPLI tests
  • Visualize Generated Data in Jupyter Notebook
  • Display History After Trainings
• File Structure
• Contributing
• Implementing Your AMPLI Tests
• More Resources
  • Documentation
  • Presentation
• License
• Contact

File Structure

Installation

Clone the repository:

git clone "https://github.com/mattynb/AmpliVision.git"

Ensure you have python3.11.9 and the compatible GPU-Enabled TensorFlow setup in your environment. Then, from the AmpliVision root directory run:

pip install -r AmpliVision/requirements.txt

Usage

AmpliVision package takes the following arguments:

Python AmpliVision USE_CASE DATASET TAG*

1. Supported USE_CASE parameters:

   CORE

   • SCAN
   • LENET
   • PyOD

   ---

   EXTRA

   • test
   • check_data
   • history

2. Supported DATASET parameters:

   Determines what dataset will be used.

   • MARKER - The Marker dataset used for proof of concept runs
   • YOUR_TARGET
   • _ - For use cases that do not need datasets. Such as history.

3. TAG*

   optional ID string. Defaults to dataset if None

Pre-Defined Use Cases

CORE Use Cases

Scan AMPLI test images and extract results

Example command to scan and extract MARKER data. The scanned images will be in data/scanned_MARKER and extracted results in data/results/DATE

Python AmpliVision scan MARKER My_Identifiable_TAG

Train a CNN LENET model to clasify Ampli test images

python3 AmpliVision LENET scanned_MARKER MARKER

Detect Outliers With Trained Model And PyOD

python3 AmpliVision PyOD scanned_MARKER MARKER

---

EXTRA Use Cases

Test a trained CNN model in Predicting Never Seen Ampli tests

python3 AmpliVision test scanned_MARKER MARKER

Visualize Generated Data in Jupyter Notebook

python3 AmpliVision check_data scanned_MARKER MARKER

Display History After Trainings

python3 AmpliVision history scanned_MARKER MARKER

Contributing

If you'd like to contribute to this project, please follow these guidelines:

1. Fork the repository.
2. Create a new branch for your feature or bug fix.
3. Make your changes.
4. Test your changes thoroughly.
5. Create a pull request.

Implementing Your AMPLI Tests

1. Add the folder with your AMPLI test images in data. Make sure each file's name start with the result. For example, a test that shows positive result for DENV4 should be named DENV4_...

2. Define the targets of your AMPLI rapid test in __main__.py. "YOUR_TARGET" is the placehoder implementation.

   def manage_targets(dataset):
       """ assigns targets user wants the CNN to predict in specific datasets """
       
       dataset = dataset.upper()
   
       if "MARKER" in dataset:    
           TAG = 'MARKER'
           TARGETS = ['lung', 'thyroid', 'ovarian', 'prostate', 'skin', 'control', 'breast']
       
       elif "YOUR_TARGET" in dataset:
           # Here is an example to show where you can assign your own targets to your dataset     
           print(" YOUR_TARGET not implemented yet in manage_targets() [__main__.py file], exiting...")
           exit()

More Resources

• Documentation
• Presentation Slides

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contact

If you have any questions or need assistance, feel free to contact us:

• Matheus Berbet
• Josselyn Mata
• Daniel Haehn
• Kimberly Hamad

We hope you find this repository useful for your automated Ampli needs!