Scholarlink

Early and precise discovery of COVID-19-related anomalies in lung computed tomography (CT) pictures is pivotal for preventive measures and treatment. Nonetheless, outwardly dissecting lung CT pictures is tedious. This study uses artificial intelligence (AI) to improve the demonstrative capacities of radiologists by dissecting processed tomography pictures. A profound learning model named ARUCOVID-Net is presented in this review, intended to precisely recognize the level of impacted regions and foresee the seriousness of coronavirus lung sores in chest CT filters. A public dataset containing physically clarified lung and disease veils from 20 COVID patients has been used to prepare the proposed framework (ARUCOVID-Net), which comprises two organizations. The main brain organization, ATTRESU-Net1, is utilized to recognize lung parenchyma. Thus, a semantic progressive division approach (ATTRESU-Net1 and ATTRESU-Net2) is used on the fragmented lungs to unequivocally recognize explicit locales impacted by COVID sores. ATTRESU-Net addresses an improved rendition of the U-Net idea consolidating a consideration system and residual units. The seriousness levels of COVID patients—gentle, moderate, extreme, or basic—are surveyed utilizing the Mamdani fuzzy inference system (MFIS) in light of the level of contamination locales. The ARUCOVID-Net has achieved a high level of precision in localizing these lesions, with a DSC of 77.8% and IOU 65.9% compared to a baseline U-Net, with enhance-mean absolute error of 1.06% in quantifying infected lung regions. Therefore, this work serves as the initial stage in evaluating the severity and predicting a future outcome of COVID-19 patients. It has the potential for quantitative assessment that aids doctors in evaluating the severity and monitoring the advancement of the disease.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Estimation of the Severity Level of COVID-19 Contagion in Chest CT Images Using Attention-Residual U-Net Structure and Fuzzy Logic Model

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Payman Hussein Hussan,
  • Israa Hadi Ali

摘要

Early and precise discovery of COVID-19-related anomalies in lung computed tomography (CT) pictures is pivotal for preventive measures and treatment. Nonetheless, outwardly dissecting lung CT pictures is tedious. This study uses artificial intelligence (AI) to improve the demonstrative capacities of radiologists by dissecting processed tomography pictures. A profound learning model named ARUCOVID-Net is presented in this review, intended to precisely recognize the level of impacted regions and foresee the seriousness of coronavirus lung sores in chest CT filters. A public dataset containing physically clarified lung and disease veils from 20 COVID patients has been used to prepare the proposed framework (ARUCOVID-Net), which comprises two organizations. The main brain organization, ATTRESU-Net1, is utilized to recognize lung parenchyma. Thus, a semantic progressive division approach (ATTRESU-Net1 and ATTRESU-Net2) is used on the fragmented lungs to unequivocally recognize explicit locales impacted by COVID sores. ATTRESU-Net addresses an improved rendition of the U-Net idea consolidating a consideration system and residual units. The seriousness levels of COVID patients—gentle, moderate, extreme, or basic—are surveyed utilizing the Mamdani fuzzy inference system (MFIS) in light of the level of contamination locales. The ARUCOVID-Net has achieved a high level of precision in localizing these lesions, with a DSC of 77.8% and IOU 65.9% compared to a baseline U-Net, with enhance-mean absolute error of 1.06% in quantifying infected lung regions. Therefore, this work serves as the initial stage in evaluating the severity and predicting a future outcome of COVID-19 patients. It has the potential for quantitative assessment that aids doctors in evaluating the severity and monitoring the advancement of the disease.