Pleural thickening on screening chest X-rays: a single institutional study.

Although pleural thickening is a common finding on routine chest X-rays, its radiological and clinical features remain poorly characterized. Our investigation of 28,727 chest X-rays obtained from annual health examinations confirmed that pleural thickening was the most common abnormal radiological finding. In most cases (92.2%), pleural thickening involved the apex of the lung, particularly on the right side; thus, it was defined as a pulmonary apical cap. Pleural thickening was more common in males than in females and in current smokers or ex-smokers than in never smokers. The prevalence increased with age, ranging from 1.8% in teenagers to 9.8% in adults aged 60 years and older. Moreover, pleural thickening was clearly associated with greater height and lower body weight and body mass index, suggesting that a tall, thin body shape may predispose to pleural thickening. These observations allowed us to speculate about the causative mechanisms of pleural thickening that are attributable to disproportionate perfusion, ventilation, or mechanical forces in the lungs.

The incidence and risk factors of asymptomatic primary spontaneous pneumothorax detected during health check-ups.

BACKGROUND: Patients with primary spontaneous pneumothorax (PSP) usually complain of sudden-onset dyspnea and pleuritic chest pain. However, asymptomatic PSP has been incidentally detected on chest X-rays. In this study, we analyzed the incidence, characteristics, risk factors, and prognosis of asymptomatic PSP detected during regular medical check-ups in university students. METHODS: In this study, 101,709 chest X-rays were performed during medical check-ups for students at the University of Tokyo between April 2011 and March 2016. Among them, 43 cases of asymptomatic PSP (0.042%) were detected. We calculated the lung collapse rate of pneumothorax using Kircher's method. We also analyzed risk factors associated with asymptomatic PSP using characteristics inspected in medical check-ups. RESULTS: The incidence of asymptomatic PSP was significantly higher in men than in women (0.050% vs 0.018%). Multivariate analysis revealed an association of younger age, greater height, lower body mass index, and greater height growth per year with an increased risk of asymptomatic PSP in male students. Mild lung collapse (<10%) was present in 22 of 43 students with asymptomatic PSP; among these, eight students eventually underwent an invasive therapy. CONCLUSIONS: The prevalence of asymptomatic PSP among university students was as high as 0.042%. In addition to known risk factors for conventional PSP, greater height growth was a risk factor for asymptomatic PSP. Careful follow-up is very important because a considerable number of patients with mild lung collapse eventually require an invasive medical procedure.

Quantitative evaluation model of variable diagnosis for chest X-ray images using deep learning.

The purpose of this study is to demonstrate the use of a deep learning model in quantitatively evaluating clinical findings typically subject to uncertain evaluations by physicians, using binary test results based on routine protocols. A chest X-ray is the most commonly used diagnostic tool for the detection of a wide range of diseases and is generally performed as a part of regular medical checkups. However, when it comes to findings that can be classified as within the normal range but are not considered disease-related, the thresholds of physicians' findings can vary to some extent, therefore it is necessary to define a new evaluation method and quantify it. The implementation of such methods is difficult and expensive in terms of time and labor. In this study, a total of 83,005 chest X-ray images were used to diagnose the common findings of pleural thickening and scoliosis. A novel method for quantitatively evaluating the probability that a physician would judge the images to have these findings was established. The proposed method successfully quantified the variation in physicians' findings using a deep learning model trained only on binary annotation data. It was also demonstrated that the developed method could be applied to both transfer learning using convolutional neural networks for general image analysis and a newly learned deep learning model based on vector quantization variational autoencoders with high correlations ranging from 0.89 to 0.97.