Aman Verma at SKF India Digital Private Limited has successfully created a prediction model which detects if a patient is COVID positive or not by analyzing the Chest X-Ray of the patient. This is achieved by using a ResNet-18 model and train it on a COVID-19 Radiography Dataset. This dataset has nearly 3000 chest X-Ray scans which are categorized in three different classes viz. Normal, Viral Pneumonia and COVID-19. His objective was to create was to create an image classification model that can predict Chest X-Ray scans that belong to one of the three classes with a reasonably high accuracy.
This model was developed with programming in python with a theoretical knowledge and practical implementation of Convolutional Neural Networks and optimization techniques such as gradient descent. The project completion certificate was also awarded to Aman by Coursera for the same.
- Create custom Dataset and DataLoader in PyTorch
- Train a ResNet-18 model in PyTorch to perform Image Classification
- Importing Libraries
- Creating Custom Dataset
- Image Transformations
- Prepare Dataloader
Testing & Inference Steps
- Data Visualization
- Creating the model
- Training the model
- Final Results
Results: Detecting COVID-19 with the help of Chest X-Ray
Skills gained while working on this project – Deep Learning, Machine Learning, Statistical Classification, Medical Imaging, and, PyTorch.
At the end of this project we can say that, Aman’s knowledge in programming with Python came in handy along with the theoretical understanding of Convolutional Neural Networks, and optimization techniques such as gradient descent. According to him, this dataset, and the model that he trained while working on this project, can not be used to diagnose COVID-19 or Viral Pneumonia as a stand-alone diagnosis mechanism.