A deep learning-based algorithm for pulmonary tuberculosis detection in chest radiography.

In tuberculosis (TB), chest radiography (CXR) patterns are highly variable, mimicking pneumonia and many other diseases. This study aims to evaluate the efficacy of Google teachable machine, a deep neural network-based image classification tool, to develop algorithm for predicting TB probability of CXRs. The training dataset included 348 TB CXRs and 3806 normal CXRs for training TB detection. We also collected 1150 abnormal CXRs and 627 normal CXRs for training abnormality detection. For external validation, we collected 250 CXRs from our hospital. We also compared the accuracy of the algorithm to five pulmonologists and radiological reports. In external validation, the AI algorithm showed areas under the curve (AUC) of 0.951 and 0.975 in validation dataset 1 and 2. The accuracy of the pulmonologists on validation dataset 2 showed AUC range of 0.936-0.995. When abnormal CXRs other than TB were added, AUC decreased in both human readers (0.843-0.888) and AI algorithm (0.828). When combine human readers with AI algorithm, the AUC further increased to 0.862-0.885. The TB CXR AI algorithm developed by using Google teachable machine in this study is effective, with the accuracy close to experienced clinical physicians, and may be helpful for detecting tuberculosis by CXR.

Comparison of reversal of rat pulmonary fibrosis of nintedanib, pirfenidone, and human umbilical mesenchymal stem cells from Wharton's jelly.

BACKGROUND: The present study compared the effects of antifibrotic medications, pirfenidone, and nintedanib, with transplantation of human umbilical mesenchymal stem cells (HUMSCs) in restoring rat pulmonary fibrosis (PF). METHODS: A stable animal model was established via an intratracheal injection of 5 mg bleomycin (BLM). One single transplantation of 2.5× 107 HUMSCs or initiation of daily oral nintedanib/pirfenidone administration was performed on day 21 following BLM damage. RESULTS: Pulmonary function examination revealed that BLM rats exhibited a significant decrease in blood oxygen saturation and an increase in respiratory rates. While no significant improvements were found in BLM rats receiving nintedanib or pirfenidone, those who transplanted with HUMSCs showed a statistical amelioration in blood oxygen saturation and significant alleviation in respiratory rates. Quantification results revealed that a significant reduction in alveolar space and marked increases in substantial cell infiltration and collagen deposition in the left lungs of BLM rats. No significant alteration was observed in BLM rats administered nintedanib or pirfenidone. However, BLM rats transplanted with HUMSCs had a significant recovery in alveolar space and noticeable decreases in cell infiltration and collagen deposition. The inflammatory cell numbers in the bronchoalveolar lavage was increased in the BLM group. While the rats treated with nintedanib or pirfenidone had a lower cell number than the BLM group, a higher cell number was found as compared with the Normal group. In rats transplanted with HUMSCs, the cell number did not differ from the Normal group. CONCLUSIONS: Transplantation of HUMSCs could effectively treat PF as opposed to the administration of anti-fibrotic drugs with nintedanib or pirfenidone with a significant better result in lung volume, pathological changes, lung function, and blood oxygen saturation.