A new automated system that involves deep learning technology enables detection of a COVID-19 lesion through the analysis of a computed tomography (CT) scan. This system, described in a study published in the journal Computers in biology and medicine, was carried out by researchers from UB, the EURECAT Technological Center of Catalonia and the Computer Vision Center (CVC).
The study “allowed us to verify the effectiveness of the system as a decision support tool for health professionals in their task of detecting COVID-19, and to measure the severity, extension and evolution. pneumonia caused by SARS-CoV-2, in the medium and long term â, notes the principal investigator of the study, Giuseppe Pezzano, researcher at UB and EURECAT Digital Health Unit.
Concretely, the operation of the system consists of “a first phase of pulmonary segmentation with the scanner to reduce the search area”, specifies Pezzano. âThen an algorithm is used to analyze the lung area and detect the presence of COVID-19. If it is positive, the image is processed to identify areas affected by the disease,â he adds.
The algorithm was tested in 79 volumes and 110 sections of CT scans that detected COVID-19 infection, obtained in three open-access image repositories. The researchers achieved an average precision for the segmentation of lesions caused by the virus of about 99%, with no false positives being seen during identification.
The method uses an innovative method to calculate the segmentation mask of medical images, which has provided good results in the segmentation of nodules in tomography images.
Some recently published studies “show that deep learning and computer vision algorithms have achieved better accuracy than cancer detection by experts in mammograms, stroke and heart attack prediction,” notes Petia Radeva , professor in the Department of Mathematics and Computer Science at UB. We couldn’t be left behind and so we worked on this technology to help physicians fight COVID-19 by providing them with high-precision data for medical image analysis in an objective, transparent and robust way. “, adds the expert, also in charge of Consolidated Research Group Computer Vision and Machine Learning at UB and principal researcher at the Computing Vision Center.
“This type of automated system represents an important tool for healthcare professionals to make more robust and precise diagnoses, because it can provide information that a human being could not measure”, emphasizes Oliver DÃaz, senior lecturer. at the Department of Mathematics and Computer Science at UB.
According to Vicent Ribas, head of the medical data analysis research line at EURECAT’s Digital Health Unit, âThe precision of this tool, demonstrated by the results of the study, opens the doors to its use for further development. other health applications, an area where the use of artificial intelligence is becoming increasingly useful. ”
Artificial intelligence can speed up stroke detection
Giuseppe Pezzano et al, CoLe-CNN +: Contextual learning – Convolutional neural network for the detection and segmentation of COVID-19-Ground-Glass-Opacities, Computers in biology and medicine (2021). DOI: 10.1016 / j.compbiomed.2021.104689
Quote: New Deep Learning Chest CT Scan System Enables Detection of COVID-19 Lesions (2021, December 1) Retrieved December 1, 2021 from https://medicalxpress.com/news/2021-12- thoracic-ct-scans-deep- authorize.html
This document is subject to copyright. Other than fair use for private study or research purposes, no part may be reproduced without written permission. The content is provided for information only.