Automated diagnosis of COVID-19 using radiological modalities and Artificial Intelligence functionalities: A retrospective study based on chest HRCT database. (February 2023)

• Record Type: Journal Article
• Authors: Bhattacharjya, Upasana; Sarma, Kandarpa Kumar; Medhi, Jyoti Prakash; Choudhury, Binoy Kumar; Barman, Geetanjali
• Is Part Of: Biomedical signal processing and control. Volume 80(2023)Part 1
• Publication Date: 2023-02
• Subjects: COVID-19 -- Ground glass opacities -- Consolidation -- Crazy paving -- Halo Sign -- Machine Learning -- Deep learning; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2022.104297
• Languages: English
• ISSNs: 1746-8094

Abstract

Abstract: Background and Objective: The spread of coronavirus has been challenging for the healthcare system's proper management and diagnosis during the rapid spread and control of the infection. Real-time reverse transcription-polymerase chain reaction (RT-PCR), though considered the standard testing measure, has low sensitivity and is time-consuming, which restricts the fast screening of individuals. Therefore, computer tomography (CT) is used to complement the traditional approaches and provide fast and effective screening over other diagnostic methods. This work aims to appraise the importance of chest CT findings of COVID-19 and post-COVID in the diagnosis and prognosis of infected patients and to explore the ways and means to integrate CT findings for the development of advanced Artificial Intelligence (AI) tool-based predictive diagnostic techniques. Methods: The retrospective study includes a 188 patient database with COVID-19 infection confirmed by RT-PCR testing, including post-COVID patients. Patients underwent chest high-resolution computer tomography (HRCT), where the images were evaluated for common COVID-19 findings and involvement of the lung and its lobes based on the coverage region. The radiological modalities analyzed in this study may help the researchers in generating a predictive model based on AI tools for further classification with a high degree of reliability. Results: Mild to moderate ground glass opacities (GGO) with or without consolidation, crazy paving patterns, and halo signs were common COVID-19 related findings. A CT score is assigned to every patient based on the severity of lung lobe involvement. Conclusion: Typical multifocal, bilateral, and peripheral distributions of GGO are the main characteristics related to COVID-19 pneumonia. Chest HRCT can be considered a standard method for timely and efficient assessment of disease progression and management severity. With its fusion with AI tools, chest HRCT can be used as a one-stop platform for radiological investigation and automated diagnosis system. Highlights: The role of chest CT in diagnostic decision making. The typical initial CT findings in COVID-19. Chest CT imaging importance in diagnostic work-up and management in patients with COVID-19. Retrospective study of collected 188 cases of HRCT images. Application of deep learning in the field of COVID-19 radiological imaging.