Implementation of a deep learning–based computer-aided detection system in clinical practice improved the diagnostic performance for detecting pneumothorax on follow-up chest radiographs after perc... Background Accurate detection of pneumothorax on chest radiographs, the most common complication of percutaneous transthoracic needle biopsies (PTNBs), is not always easy in practice. A computer-ai...

Kang M, et al. J Korean Med Sci. 2021 Jan;36(5):e46. https://doi.org/10.3346/jkms.2021.36.e46

Sepsis is a potentially life-threatening condition that occurs when the body’s response to infection leads to widespread inflammation and tissue damage. Negative cultures can make it difficult for clinicians to make a diagnosis and may raise questions about the validity of the definition of sepsis. In addition, the clinical distinctions between burn patients with blood culture-positive and -negative sepsis are also poorly understood. Therefore, this study aimed to examine the clinical differences between blood culture-positive and -negative sepsis in burn patients in order to improve the understanding of the pathophysiology and epidemiology of sepsis in this population. This study had a retrospective design, and the participants were adults aged ≥18 years. Patients diagnosed with sepsis were divided into two groups based on their blood culture results within 1 week of sepsis diagnosis. We enrolled 1643 patients admitted to our institution’s burn intensive care unit between January 2010 and December 2021. pH, platelet count, bicarbonate and haematocrit were significant in both the positive and negative groups. However, lymphocyte, red cell distribution width and blood urea nitrogen were significant only in the positive group, whereas lactate dehydrogenase was significant only in the negative group. Acinetobacter baumannii, Pseudomonas aeruginosa, and Klebsiella pneumonia are common gram-negative bacterial species, and Staphylococcus aureus and Staphylococcus epidermidis are common gram-positive bacterial species seen in burn patients with positive blood cultures. Carbapenem resistance was found to be associated with an unfavourable prognosis in gram-negative bacteria, with the exception of P. aeruginosa. pH, platelet count, bicarbonate and haematocrit were routine biomarkers that demonstrated statistical significance in both groups. Lactate dehydrogenase was significant in the blood-negative group, while red cell distribution width, blood urea nitrogen and lymphocyte count were significant in the positive group. Furthermore, the most common causes of sepsis are gram-negative bacteria, including A. baumannii, K. pneumoniae and P. aeruginosa. Additionally, resistance to carbapenems is associated with unfavourable outcomes.

Diagnosing ground-glass opacity (GGO) pulmonary lesions poses challenges. This study evaluates the utility of radial probe endobronchial ultrasound-guided transbronchial lung biopsy (RP-EBUS-TBLB) in diagnosing GGO pulmonary lesions. A total of 1651 RP-EBUS procedures were performed during the study period. This study analyzed 115 GGO lesions. The EBUS visualization yield was 80.1%. Of 115 lesions, 69 (60%) were successfully diagnosed. The average size of diagnosed lesions was significantly larger than that of undiagnosed lesions (21.9 ± 7.3 vs. 17.1 ± 6.6 mm, p < 0.001). Diagnostic yield varied by lesion size: 50.0% for lesions 30 mm. The mixed blizzard sign on EBUS appeared in 60.6% of mixed GGO lesions, with no cases in pure GGO lesions. Multivariable analyses showed that lesion size (odds ratio [OR], 1.10; 95% confidence interval [CI], 1.00–1.16; p < 0.001) and mixed blizzard sign on EBUS (OR, 20.92; CI, 7.50–58.31; p < 0.001) were significantly associated with diagnostic success. Pneumothorax and hemoptysis occurred in 1.7% and 2.6% of patients, respectively. RP-EBUS-TBLB without fluoroscopic guidance is a viable diagnostic approach for GGO pulmonary lesions with acceptable complications.

Objective: Prior studies on models based on deep learning (DL) and measuring the cardiothoracic ratio (CTR) on chest radiographs have lacked rigorous agreement analyses with radiologists or reader tests. We validated the performance of a commercially available DL-based CTR measurement model with various thoracic pathologies, and performed agreement analyses with thoracic radiologists and reader tests using a probabilistic-based reference. Materials and Methods: This study included 160 posteroanterior view chest radiographs (no lung or pleural abnormalities, pneumothorax, pleural effusion, consolidation, and n = 40 in each category) to externally test a DL-based CTR measurement model. To assess the agreement between the model and experts, intraclass or interclass correlation coefficients (ICCs) were compared between the model and two thoracic radiologists. In the reader tests with a probabilistic-based reference standard (Dawid–Skene consensus), we compared diagnostic measures—including sensitivity and negative predictive value (NPV)—for cardiomegaly between the model and five other radiologists using the non-inferiority test. Results: For the 160 chest radiographs, the model measured a median CTR of 0.521 (interquartile range, 0.446–0.59) and a mean CTR of 0.522 ± 0.095. The ICC between the two thoracic radiologists and between the model and two thoracic radiologists was not significantly different (0.972 versus 0.959, p = 0.192), even across various pathologies (all p-values > 0.05). The model showed non-inferior diagnostic performance, including sensitivity (96.3% versus 97.8%) and NPV (95.6% versus 97.4%) (p < 0.001 in both), compared with the radiologists for all 160 chest radiographs. However, it showed inferior sensitivity in chest radiographs with consolidation (95.5% versus 99.9%; p = 0.082) and NPV in chest radiographs with pleural effusion (92.9% versus 94.6%; p = 0.079) and consolidation (94.1% versus 98.7%; p = 0.173). Conclusion: While the sensitivity and NPV of this model for diagnosing cardiomegaly in chest radiographs with consolidation or pleural effusion were not as high as those of the radiologists, it demonstrated good agreement with the thoracic radiologists in measuring the CTR across various pathologies.

When evaluated by a panel of expert thoracic radiologists, artificial intelligence–generated chest radiograph reports showed an overall clinical acceptance rate of 87.6%, with acceptance rates differing by clinical context.