TO: Prof. Ellis
FROM: Neil Domingo
DATE: 3/3/2021
SUBJECT: 500-Word Summary of Article About Utilizing Artificial Intelligence In Fighting COVID-19
The following is a 500-word summary of a peer-reviewed article about the use of artificial intelligence in medical imaging during the COVID-19 pandemic. The journal’s goal is to further discuss the use of medical imaging with artificial intelligence in fighting against COVID-19 and discuss machine learning methods in the imaging workflow. Medical imaging such as CT scans and X-rays have been found to play a critical role in restraining the transmission of COVID-19. CT scans is one of the imaging-based diagnoses that is used for COVID-19 and includes three stages: pre-scan acquisition, image acquisition, and disease diagnosis. Artificial Intelligence contributes to the fight against COVID-19 as it allows for safer, accurate, and efficient imaging solutions. Imaging facilities, and workflows should be considered important to reduce the risks and save lives from COVID-19. According to authors “AI-empowered image acquisition can significantly help automate the scanning procedure and also reshape the workflow with minimal contact to patients, providing the best protection to the imaging technicians” (F. Shi et al., 2020, pg 4). The use of contactless and imaging acquisition is necessary to reduce the risks of technicians and patients being infected as there is contact between them. Artificial intelligence can be used to help the contactless scanning as it will be able to identify the pose and shape of a patient by using data from visual sensors. Scan range, the start and end point of a CT scan, can be estimated by the use of visual sensors with artificial intelligence, and scanning efficiency can be improved. A mobile CT platform with artificial intelligence implemented, is an example of an scanning automated workflow allowing for the prevention of unnecessary interaction between technicians and patients. The patient positioning algorithm will capture the patient’s pose.
Segmentation is crucial in image processing and analysis in order to assess COVID-19 as it covers the region of interest (ROIs) (organs that are affected by COVID-19/ infected areas). CT produces high-quality 3D images, and ROIs can be segmented into it. Proposals such as human knowledge,and machine learning methods can be integrated with a segmentation network in order to allow for adequate training data for segmentation tasks. Image segmentation allows radiologists to accurately identify lung infection, and analyzing and diagnosing COVID-19.
Patients that are suspected of COVID-19 are in need of diagnosis and treatment, and with COVID-19 being similar to pneumonia, in which AI-assisted diagnosis using medical images can be highly beneficial. Deep learning models were proposed such as ResNet50 to detect COVID-19 through X-ray images. The ResNet50 model contains two tasks: classification between COVID/non-COVID and anomaly detection (allows for optimization of the COVID-19 score that is used for classification). Studies have separated COVID-19 patients from non-COVID-19 patients, with the help of artificial intelligence and the reading time of radiologists was reduced by 65%.
With many studies proposing CT-based COVID-19 diagnosis show promising results, it is important for early detection and predictions of severity. It is challenging for artificial intelligence to be used in a procedure regarding the incubation period and infectivity. X-rays and CT scans are not often available for COVID-19 applications which slows down any artificial intelligence methods from continually being researched and developed.
Reference
F. Shi et al., “Review of Artificial Intelligence Techniques in Imaging Data Acquisition, Segmentation, and Diagnosis for COVID-19,” in IEEE Reviews in Biomedical Engineering, vol. 14, pp. 4-15, 2021, doi: 10.1109/RBME.2020.2987975.
