SINGAPORE - “Your project is going to be a template for future computer vision projects like this. You did a great job here.” – Machine Learning Professor, Singapore Management University
On the eve of 2nd of August 2021, we presented (AND NAILED) our project for our applied machine learning class. The term requirement for this course is to create a project using relevant use of case and real-world data. The conception of our study with a title "Detection of COVID-19 Disease Through CNN-Based Image Classification of Chest X-Ray and CT Scan Images" is a product of deliberate research on finding solutions to world problems that require immediate attention. Its objective is to find a faster and more convenient way of diagnosing COVID-19. Image classification of X-ray and CT scan images using machine learning techniques is a non-invasive method that could potentially offer a faster way of detecting the disease. The model can tell whether an X-Ray or CT scan image of the lungs is positive or negative for the disease in seconds depending on the computational power of the computer used. This method is not meant to replace the polymerase chain reaction (PCR) test which is the gold standard for diagnosing COVID-19. Instead, we thought that this method could be helpful in places where X-ray machines are more readily available than the PCR test. This can also be a good starting point for future research on the application of artificial intelligence to the healthcare system.
We hit a maximum accuracy of 99.5% and 97.1% for X-Ray and CT scan classification, respectively. It helped that we were able to find novel ways of explaining which parts of the X-Ray and CT scan images contribute greatly towards prediction. Moving forward, the plan is to collaborate with the professor to further enhance our study for possible publication and industry use. Our professor challenged us to build a single model capable of predicting both X-Ray and CT as a single dataset. We hope that this project could make a positive contribution to the pandemic-stricken world. Our group represents the Philippines, Russia, Italy, Vietnam and Singapore. (Text & Photos: Br. Kino Escolano FSC)