https://github.com/batuhanaktan/respiratoryillnessdetectionusingml
In the current state of healthcare many hospitals are working over their capacity, and our group wanted to help fix this issue, we used machine learning to diagnose a healthy lung X-ray from a sick one.
https://github.com/batuhanaktan/respiratoryillnessdetectionusingml
healthcare machine-learning sequential-models tensorflow
Last synced: about 2 months ago
JSON representation
In the current state of healthcare many hospitals are working over their capacity, and our group wanted to help fix this issue, we used machine learning to diagnose a healthy lung X-ray from a sick one.
- Host: GitHub
- URL: https://github.com/batuhanaktan/respiratoryillnessdetectionusingml
- Owner: BatuhanAktan
- Created: 2020-12-19T16:36:38.000Z (over 5 years ago)
- Default Branch: main
- Last Pushed: 2020-12-20T13:38:41.000Z (over 5 years ago)
- Last Synced: 2025-02-25T00:37:33.734Z (over 1 year ago)
- Topics: healthcare, machine-learning, sequential-models, tensorflow
- Language: Python
- Homepage:
- Size: 85.3 MB
- Stars: 1
- Watchers: 2
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: ReadMe.md
Awesome Lists containing this project
README
# Diagnosis of respiratory illnesses.
In the current state of healthcare many hospitals are working over their capacity, and our group wanted to help fix this issue, we used machine learning to diagnose a healthy lung X-ray from a sick one.
### Getting Started
We used the public data provided with more than 4000 chest cts to develop and form a sequential model with tensorflow. We used 64 cts as our sample 32 infected and 32 healthy, with this model after 10 Epochs we were able to hit 100% accuracy.
### Prerequisites
All the libraries used are in requirements.txt, and they can be installed using pip command.
Example.
```
pip install numpy
```
### Installing
The machine learning model can be installed in the following google drive
```
https://drive.google.com/file/d/1r0XxZCt8s_TWgDvd6Kg_ObjuLs6C2iTJ/view
```
## Running the tests
To run these tests we used the model mentioned above and with to test various X-rays and we achieved a total of 100% accuracy in the prediction of 10 X-rays.
### Results
```
57/57 [==============================] - 40s 698ms/sample - loss: 1.1120e-05 - accuracy: 1.0000 - val_loss: 0.0429 - val_accuracy: 1.0000
```
After training our machine learning model with 64 X-ray images we were able to get 100% accuracy in the determination of respiratory illnesses and we expect this result to be close to 90% with X-rays that are out of our training set.
|||
|:---:|:---:|
|Healthy Lung: IM-0001-0001.jpeg|Sick Lung: person15_virus_46.jpeg|
```
Image being processed: IM-0001-0001.jpeg
Expected Result: Healthy
Result: Healthy
Image being processed: person15_virus_46.jpeg
Expected Result: Sick
Result: Sick
```
(All the images are in the TestImages folder)
These results can be replicated using our model and using Prediction.py.
### Conclusion
To conclude, our model can identify resipratory illness presence in a chest X-ray at around 90% accuracy, which in return could provide lifesaving support to patients and reduce the stress on healthcare professionals.
## Authors
* **Batuhan Aktan** - [Github](https://github.com/BatuhanAktan)
* **Jody Zhou** - [Github](https://github.com/JodyZ0203)
* **Sam Zhang** - [Github](https://github.com/Dam-Sam)
## License
This project is licensed under the MIT License.