Developing and verifying automatic detection of active pulmonary tuberculosis from multi-slice spiral CT images based on deep learning.

OBJECTIVE: Diagnosis of tuberculosis (TB) in multi-slice spiral computed tomography (CT) images is a difficult task in many TB prevalent locations in which experienced radiologists are lacking. To address this difficulty, we develop an automated detection system based on artificial intelligence (AI) in this study to simplify the diagnostic process of active tuberculosis (ATB) and improve the diagnostic accuracy using CT images. DATA: A CT image dataset of 846 patients is retrospectively collected from a large teaching hospital. The gold standard for ATB patients is sputum smear, and the gold standard for normal and pneumonia patients is the CT report result. The dataset is divided into independent training and testing data subsets. The training data contains 337 ATB, 110 pneumonia, and 120 normal cases, while the testing data contains 139 ATB, 40 pneumonia, and 100 normal cases, respectively. METHODS: A U-Net deep learning algorithm was applied for automatic detection and segmentation of ATB lesions. Image processing methods are then applied to CT layers diagnosed as ATB lesions by U-Net, which can detect potentially misdiagnosed layers, and can turn 2D ATB lesions into 3D lesions based on consecutive U-Net annotations. Finally, independent test data is used to evaluate the performance of the developed AI tool. RESULTS: For an independent test, the AI tool yields an AUC value of 0.980. Accuracy, sensitivity, specificity, positive predictive value, and negative predictive value are 0.968, 0.964, 0.971, 0.971, and 0.964, respectively, which shows that the AI tool performs well for detection of ATB and differential diagnosis of non-ATB (i.e. pneumonia and normal cases). CONCLUSION: An AI tool for automatic detection of ATB in chest CT is successfully developed in this study. The AI tool can accurately detect ATB patients, and distinguish between ATB and non- ATB cases, which simplifies the diagnosis process and lays a solid foundation for the next step of AI in CT diagnosis of ATB in clinical application.

Impact of localized fine tuning in the performance of segmentation and classification of lung nodules from computed tomography scans using deep learning.

Background: Algorithm malfunction may occur when there is a performance mismatch between the dataset with which it was developed and the dataset on which it was deployed. Methods: A baseline segmentation algorithm and a baseline classification algorithm were developed using public dataset of Lung Image Database Consortium to detect benign and malignant nodules, and two additional external datasets (i.e., HB and XZ) including 542 cases and 486 cases were involved for the independent validation of these two algorithms. To explore the impact of localized fine tuning on the individual segmentation and classification process, the baseline algorithms were fine tuned with CT scans of HB and XZ datasets, respectively, and the performance of the fine tuned algorithms was tested to compare with the baseline algorithms. Results: The proposed baseline algorithms of both segmentation and classification experienced a drop when directly deployed in external HB and XZ datasets. Comparing with the baseline validation results in nodule segmentation, the fine tuned segmentation algorithm obtained better performance in Dice coefficient, Intersection over Union, and Average Surface Distance in HB dataset (0.593 vs. 0.444; 0.450 vs. 0.348; 0.283 vs. 0.304) and XZ dataset (0.601 vs. 0.486; 0.482 vs. 0.378; 0.225 vs. 0.358). Similarly, comparing with the baseline validation results in benign and malignant nodule classification, the fine tuned classification algorithm had improved area under the receiver operating characteristic curve value, accuracy, and F1 score in HB dataset (0.851 vs. 0.812; 0.813 vs. 0.769; 0.852 vs. 0.822) and XZ dataset (0.724 vs. 0.668; 0.696 vs. 0.617; 0.737 vs. 0.668). Conclusions: The external validation performance of localized fine tuned algorithms outperformed the baseline algorithms in both segmentation process and classification process, which showed that localized fine tuning may be an effective way to enable a baseline algorithm generalize to site-specific use.