Preeti
This study aimed to develop an AI algorithm for detecting tuberculosis (TB) using freely available open TB chest X-ray datasets for training. The study was approved by the institutional review board and adhered to ethical guidelines. Two separate training datasets were used to train the algorithm to detect TB patterns and other abnormalities on chest X-rays. The algorithm was developed using Google Teachable Machine, a free online AI software that employs transfer learning to train image classification models. The study utilized high-performance hardware for training the algorithm.
To validate the accuracy of the algorithm, clinical CXR data from the hospital were collected and used for external validation. The dataset included TB, normal cases, and pneumonia/other disease cases. The algorithm’s performance was compared to that of five clinical physicians, and radiology reports were also analyzed for TB detection sensitivity. Statistical analyses, including ROC curve analysis, sensitivity, specificity, positive predictive value, negative predictive value, and overall accuracy, were performed to evaluate the algorithm’s performance.
The study found that the AI algorithm showed promising accuracy in detecting TB on chest X-rays compared to clinical physicians. The algorithm’s performance was evaluated using various metrics, and the results indicated its potential for accurate TB detection. The study highlights the importance of AI in improving TB diagnosis and underscores the need for further research and validation in clinical settings.
Source link
Source link: https://www.nature.com/articles/s41598-024-65703-z
GIPHY App Key not set. Please check settings