Skip to content

In this project, I implemented a framework to try and infer COVID diagnosis from chest X-rays. I first trained a specific U-net to generate masks to segment specific areas of chest x-rays, and CNNs to infer Covid-19 diagnosis on the segmented areas of the lungs.

Yskandar/Covid-Diagnosis-from-Chest-Xrays

Repository files navigation

This project is an attempt at establishing a COVID diagnosis using chest x-rays only. This project is based on the following github repository : https://github.com/ieee8023/covid-chestxray-dataset

Project Description

To infer covid diagnosis from chest x-rays, here is the approach that I took. I divided the task in two main parts: segmenting the chest x-rays in order to keep only the relevant parts, and infering covid diagnosis from the processed x-rays. If the first task is achieved satisfyingly, it is not the case of the second part. I have stated at the end of the second notebook the reasons I was able to identify that might explain the unsatisfying results obtained.

How do I run it ?

To run this notebook, one needs to launch Images Diagnosis first, then Images Segmentation. The first notebook processes hundreds of images, which might cause memory errors on certain computers.

Annotations

Lung Bounding Boxes and Chest X-ray Segmentation (license: CC BY 4.0) contributed by General Blockchain, Inc.

Pneumonia severity scores for 94 images (license: CC BY-SA) from the paper Predicting COVID-19 Pneumonia Severity on Chest X-ray with Deep Learning

Generated Lung Segmentations (license: CC BY-SA) from the paper Lung Segmentation from Chest X-rays using Variational Data Imputation

Brixia score for 192 images (license: CC BY-NC-SA) from the paper End-to-end learning for semiquantitative rating of COVID-19 severity on Chest X-rays

Lung and other segmentations for 517 images (license: CC BY) in COCO and raster formats by v7labs

References

  1. O.Ronneberger, P.Fischer and T.Brox : U-Net: Convolutional Networks for Biomedical Image Segmentation. In arXiv:1505.04597v1 [cs.CV] 18 May 2015.

  2. M.Sandler, A.Howard, M.Zhu, A.Zhmoginov, and L.Chen : MobileNetV2: Inverted Residuals and Linear Bottlenecks. In arXiv:1801.04381v4 [cs.CV] 21 Mar 2019.

  3. P.Isola, J.Zhu, T.Zhou and A.Efros : Image-to-Image Translation with Conditional Adversar- ial Networks. In arXiv:1611.07004v3 [cs.CV] 26 Nov 2018.

  4. Rasmita Lenkaa, Asimananda Khandualb : Application of CNN and gated recurrent network for visual improvement of dehazing. Materialstoday : proceeding. 16 January 2021.

  5. Priya Dwivedi : Understanding and Coding a ResNet in Keras. 4 June 2016 [online].

  6. Chengwei : How to do Transfer learning with Efficientnet. [online]

  7. Gaurav Singhal : Introduction to DenseNet with TensorFlow. 6 May 2020 [online].

  8. Detecting and Visualising the Infectious Regions of COVID-19 in X-ray Im-ages and CT scans Using Different Pretrained-Networks in Tensorflow 2.x. 20 July 2020 [online].

  9. Pham, T.D. : A comprehensive study on classification of COVID-19 on computed tomography with pretrained convolutional neural networks. Sci Rep 10, 16942. 09 October 2020 [online].

About

In this project, I implemented a framework to try and infer COVID diagnosis from chest X-rays. I first trained a specific U-net to generate masks to segment specific areas of chest x-rays, and CNNs to infer Covid-19 diagnosis on the segmented areas of the lungs.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published