According to recent studies, patients with COVID-19 have different feature characteristics on chest X-ray (CXR) than those with other lung diseases. This study aimed at evaluating the layer depths and degree of fine-tuning on transfer learning with a deep convolutional neural network (CNN)-based COVID-19 screening in CXR to identify efficient transfer learning strategies. The CXR images used in this study were collected from publicly available repositories, and the collected images were classified into three classes: COVID-19, pneumonia, and normal. To evaluate the effect of layer depths of the same CNN architecture, CNNs called VGG-16 and VGG-19 were used as backbone networks. Then, each backbone network was trained with different degrees of fine-tuning and comparatively evaluated. The experimental results showed the highest AUC value to be 0.950 concerning COVID-19 classification in the experimental group of a fine-tuned with only 2/5 blocks of the VGG16 backbone network. In conclusion, in the classification of medical images with a limited number of data, a deeper layer depth may not guarantee better results. In addition, even if the same pre-trained CNN architecture is used, an appropriate degree of fine-tuning can help to build an efficient deep learning model.

Lung cysts are commonly seen on computed tomography (CT), and cystic lung diseases show a wide disease spectrum. Thus, correct diagnosis of cystic lung diseases is a challenge for radiologists. As the first diagnostic step, cysts should be distinguished from cavities, bullae, pneumatocele, emphysema, honeycombing, and cystic bronchiectasis. Second, cysts can be categorized as single/localized versus multiple/diffuse. Solitary/localized cysts include incidental cysts and congenital cystic diseases. Multiple/diffuse cysts can be further categorized according to the presence or absence of associated radiologic findings. Multiple/diffuse cysts without associated findings include lymphangioleiomyomatosis and Birt-Hogg-Dubé syndrome. Multiple/diffuse cysts may be associated with ground-glass opacity or small nodules. Multiple/diffuse cysts with nodules include Langerhans cell histiocytosis, cystic metastasis, and amyloidosis. Multiple/diffuse cysts with ground-glass opacity include pneumocystis pneumonia, desquamative interstitial pneumonia, and lymphocytic interstitial pneumonia. This stepwise radiologic diagnostic approach can be helpful in reaching a correct diagnosis for various cystic lung diseases.

Kim C, et al. Korean J Radiol. 2020 Jul;21(7):838-850. https://doi.org/10.3348/kjr.2019.0711

Obstructive sleep apnoea (OSA) is a common form of sleep disordered breathing. Untreated OSA might accelerate atherosclerosis, potentially increasing the cardiovascular disease burden in patients. The present study aimed to evaluate the association between objectively measured OSA severity and the presence of subclinical systemic atherosclerosis using noninvasive measurements, including tomographic quantification of the calcium burden.A total of 2157 participants of the Korean Genome and Epidemiology Study, who were free of structural heart disease and underwent both in-home polysomnography and chest computed tomography, were cross-sectionally analysed. Participants were divided into three groups based on the severity of OSA: no OSA (apnoea–hypopnoea index (AHI) <5 events·h⁻¹, n=1096), mild OSA (AHI 5– <15 events·h⁻¹, n=700) and moderate-to-severe OSA (AHI ≥15 events·h⁻¹, n=361). Calcium deposits in the thoracic aorta and coronary arteries were measured by the Agatston score.Participants with moderate-to-severe OSA were 1.6 times (95% CI 1.18–2.15 times; p=0.002) more likely to have ascending thoracic aorta calcification (≥100 units) than those without OSA, after adjustment for cardiovascular risk factors. In addition, the association between moderate-to-severe OSA and ascending thoracic aorta calcification of subjects with higher epicardial fat volume was slightly stronger than that in patients without OSA and in the lowest epicardial fat volume tertile (OR 2.11, 95% CI 1.30–3.43).Severity of OSA in the general population was independently associated with subclinical systemic atherosclerosis. These findings highlight the potential importance of severe OSA, especially in subjects with higher epicardial fat, as a possible predictive factor for systemic atherosclerosis and cardiovascular disease.

Kim M, et al. Yonsei Med J. 2009 Jun;50(3):385-390. http://dx.doi.org/10.3349/ymj.2009.50.3.385