adammiltonbarker / all-detection-system-2020

Acute Lymphoblastic Leukemia Detection System 2020 uses Tensorflow 2 & Oculus Rift to provide a virtual diagnosis system. Project by Adam Milton-Barker.

Peter Moss Acute Myeloid & Lymphoblastic Leukemia AI Research Project

Acute Lymphoblastic Leukemia Detection System 2020

 

Table Of Contents

• Introduction
• DISCLAIMER
• ALL-IDB
  • ALL_IDB1
• Classifier
  • Results Overview
• Setup
• Contributing
  • Contributors
• Versioning
• License
• Bugs/Issues

 

Introduction

The Acute Lymphoblastic Leukemia Detection System 2020 uses Tensorflow 2 & Oculus Rift, and Raspberry Pi 4 to provide a virtual diagnosis system.

This project is made up of a number of components which work together as a network to make the system work. Follow the completed tutorials below in the provided order. The classifiers can be run indvidually but a full system setup requires the Tensorflow 2.0 CNN, Virtual Reality (Oculus Rift), an RPI 4 and Arduino IoT Alarm.

This project is compatible with the Acute Lymphoblastic Detection System feature of HIAS.

 

DISCLAIMER

This project should be used for research purposes only. The purpose of the project is to show the potential of Artificial Intelligence for medical support systems such as diagnosis systems. Although the program is fairly accurate and shows good results both on paper and in real world testing, it is not meant to be an alternative to professional medical diagnosis. I am a self taught developer with some experience in using Artificial Intelligence for detecting certain types of cancer. I am not a doctor, medical or cancer expert. Please use this system responsibly.

 

ALL-IDB

You need to be granted access to use the Acute Lymphoblastic Leukemia Image Database for Image Processing dataset. You can find the application form and information about getting access to the dataset on this page as well as information on how to contribute back to the project here. If you are not able to obtain a copy of the dataset please feel free to try this tutorial on your own dataset, we would be very happy to find additional AML & ALL datasets.

ALL_IDB1

In this project, ALL-IDB1 is used, one of the datsets from the Acute Lymphoblastic Leukemia Image Database for Image Processing dataset. We will use data augmentation to increase the amount of training and testing data we have.

"The ALL_IDB1 version 1.0 can be used both for testing segmentation capability of algorithms, as well as the classification systems and image preprocessing methods. This dataset is composed of 108 images collected during September, 2005. It contains about 39000 blood elements, where the lymphocytes has been labeled by expert oncologists. The images are taken with different magnifications of the microscope ranging from 300 to 500."

 

Classifier

Fig 1. Proposed architecture (Source)

In this project we will a network based on the proposed architecture in the Acute Leukemia Classification Using Convolution Neural Network In Clinical Decision Support System paper, built using Tensorflow 2.

Classifier Results

We have tested our model on a number of different hardwares, including Intel® CPUs & NVIDIA GPUs. Results seem to vary between CPU & GPU, and tests show further investigation into seeding and randomness introduced to our network via the GPU software. For reproducible results every time, it is suggested to train on a CPU, although this obviously requires more time.

One method to overcome reproducibility issues and get an good idea of how well our model is behaving on GPU would be to test the model multiple times and take an average. This is one way we will explore our model in the future.

Below are the results from individual training sessions.

OS	Hardware	Training	Validation	Test	Accuracy	Recall	Precision	AUC/ROC
Google Colab	Tesla K80 GPU	1180	404	20	0.9727723	0.9727723	0.9727723	0.9948964
Windows 10	NVIDIA GeoForce GTX 1060	1180	404	20	0.97066015	0.97066015	0.97066015	0.9908836
Ubuntu 18.04	NVIDIA GTX 1050 Ti Ti/PCIe/SSE2	1180	404	20	0.97772276	0.97772276	0.97772276	0.9989155
Ubuntu 18.04	Intel® Core™ i7-7700HQ CPU @ 2.80GHz × 8	1180	404	20	0.9752475	0.9752475	0.9752475	0.991492
Windows 10	Intel® Core™ i7-7700HQ CPU @ 2.80GHz × 8	1180	404	20	0.9851485	0.9851485	0.9851485	0.9985846
macOS Mojave 10.14.6	Intel® Core™ i5 CPU @ 2.4 GHz	1180	404	20	0.9589041	0.9589041	0.9589041	0.99483955

 

Setup

Follow the guides below to build your Virtual Reality Acute Lymphoblastic Leukemia Detection System. The classifiers can be used as standalone if required, but this project focuses on the system as a whole.

Project	Description	Status
Rift	A virtual room made with Unreal Engine 4. In the room the ALL data is displayed as blocks and the user can point at those blocks to classify the image.	Development
CNN	Applies filters to the original dataset and increases the amount of training / test data. Provides code for training a CNN for detecting ALL. Hosts a REST API endpoint that provides access to the model for inference.	COMPLETE
RPI 4	Applies filters to the original dataset and increases the amount of training / test data. Provides code for training a CNN for detecting ALL. Hosts a REST API endpoint that provides access to the model for inference.	COMPLETE
Arduino	An IoT connected Arduino ioT Alarm using an Arduino UNO, an ESP8266 and the HIAS iotJumpWay MQTT broker.	Development

 

Contributing

The Peter Moss Acute Myeloid & Lymphoblastic Leukemia AI Research project encourages and welcomes code contributions, bug fixes and enhancements from the Github.

Please read the CONTRIBUTING document for a full guide to forking our repositories and submitting your pull requests. You will also find information about our code of conduct on this page.

Contributors

• Adam Milton-Barker - Asociacion De Investigation En Inteligencia Artificial Para La Leucemia Peter Moss President/Lead Developer, Sabadell, Spain

• Javier Lopez Alonso - Asociacion De Investigation En Inteligencia Artificial Para La Leucemia Peter Moss Treasurer/Research & Development, Sabadell, Spain

• Rishabh Banga - Asociacion De Investigation En Inteligencia Artificial Para La Leucemia Peter Moss Research & Development, Delhi, India

 

Versioning

We use SemVer for versioning. For the versions available, see Releases.

 

License

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

 

Bugs/Issues

We use the repo issues to track bugs and general requests related to using this project. See CONTRIBUTING for more info on how to submit bugs, feature requests and proposals.