Zonguldak Bülent Ecevit University, Department of Electrical-Electronics Engineering, Zonguldak, 67100, Turkey.
Comput Biol Med. 2022 Sep;148:105934. doi: 10.1016/j.compbiomed.2022.105934. Epub 2022 Aug 2.
World Health Organization has described the real-time reverse transcription-polymerase chain reaction test method for the diagnosis of the novel coronavirus disease (COVID-19). However, the limited number of test kits, the long-term results of the tests, the high probability of the disease spreading during the test and imaging without focused images necessitate the use of alternative diagnostic methods such as chest X-ray (CXR) imaging. The storage of data obtained for the diagnosis of the disease also poses a major problem. This causes misdiagnosis and delays treatment. In this work, we propose a hybrid 3D reconstruction method of CXR images (CXRI) to detect coronavirus pneumonia and prevent misdiagnosis on CXRI. We used the digital holography technique (DHT) for obtaining a priori information of CXRI stored in created digital hologram (CDH). In this way, the elimination of the storage problem that requires high space was revealed. In addition, Discrete Orthonormal S-Transform (DOST) is applied to the reconstructed CDH image obtained by using DHT. This method is called CDH_DHT_DOST. A multiresolution spatial-frequency representation of the lung images that belong to healthy people and diseased people with the COVID-19 virus is obtained by using the CDH_DHT_DOST. Moreover, the genetic algorithm (GA) is adopted for the reconstruction process for optimization of the CDH image and then DOST is applied. This hybrid method is called CDH_GA_DOST. Finally, we compare the results obtained from CDH_DHT_DOST and CDH_GA_DOST. The results show the feasibility of reconstructing CXRI with CDH_GA_DOST. The proposed method holds promises to meet demands for the detection of the COVID-19 virus.
世界卫生组织(WHO)已经描述了用于诊断新型冠状病毒病(COVID-19)的实时逆转录-聚合酶链反应检测方法。然而,检测试剂盒数量有限、检测结果需要较长时间、检测过程中疾病传播的可能性较高且影像学检查缺乏聚焦图像,这使得需要使用替代的诊断方法,如胸部 X 射线(CXR)成像。此外,用于诊断疾病的数据存储也存在很大问题,这可能导致误诊和治疗延误。在这项工作中,我们提出了一种 CXR 图像的混合 3D 重建方法(CXRI),用于检测冠状病毒性肺炎并防止 CXRI 误诊。我们使用数字全息技术(DHT)获取存储在创建的数字全息图(CDH)中的 CXRI 的先验信息。这样,就揭示了消除需要高空间存储的问题。此外,离散正交 S-变换(DOST)应用于使用 DHT 获得的重建 CDH 图像。这种方法称为 CDH_DHT_DOST。通过使用 CDH_DHT_DOST,获得了属于健康人和 COVID-19 病毒感染者的肺部图像的多分辨率空间频率表示。此外,采用遗传算法(GA)进行重建过程,以优化 CDH 图像,然后应用 DOST。这种混合方法称为 CDH_GA_DOST。最后,我们比较了 CDH_DHT_DOST 和 CDH_GA_DOST 获得的结果。结果表明,使用 CDH_GA_DOST 重建 CXRI 是可行的。该方法有望满足 COVID-19 病毒检测的需求。