Machine Learning, CNN, Convolutional Neural Network, Neural Network, Covid-19, Chest-CT


Artificial Intelligence and Robotics | Computer Sciences


Dr. Greg Wolffe


Currently, the most widely used diagnostic tool for COVID-19 is the RT-PCR nasal swab test recommended by the CDC. However, some studies have shown that chest CT scans have the potential to be more accurate and are also capable of detecting the virus in its earlier stages. Unfortunately, CT results are not instantaneously available as it may be days before a radiologist can review the scan. This delay is one of the factors preventing the widespread use of CT scans for COVID detection. To address the delay, this project investigated Convolutional Neural Networks, an advanced form of machine learning used for image classification. CNNs have proven very effective at extracting patterns from images and have been used to detect clinical signs of COVID. The goal of this project was to develop an improved CNN that could accurately predict whether a patient is COVID-19 positive based on their CT scan. This could potentially provide a valuable prescreening tool for overwhelmed radiologists.