COVID-19: Denoising COVID-19 CT Scans With Scalable Open Source Software
Computed Tomography scans (or “CT scans”) have been shown to be better detectors of COVID-19 than the traditional reverse-transcription polymerase chain reaction (RT-PCR) test, especially at early stages of infection. The radiation dose associated with a thorax CT scan is high. A low dose method of CT scans exists, but resultant image quality is markedly lower than a traditional scan. We present a method for denoising CT scan images based on scalable open source software. We find our method performs “well enough”, considering our aims to create a production application with “off-the-shelf” components, but would like to do follow on research comparing it to other denoising methods. We would also like to combine our results with other research to create a COVID-19 detection test based which is entirely automated and based on CT scans only.