Search

Automated diagnosis of complex breast ultrasound images using deep learning has important clinical implications. In this study, we propose a novel breast cancer diagnosis framework, the CAS framework, based on lesion region to effectively recognize benign and malignant tumors. Firstly, train the segmentation network CRA-ENet on a public dataset with masks and save the trained weights. Then, the weights are used to segment clinical data without masks to obtain masks. Next, cover the masks on the corresponding original images to obtain ultrasound images that only include the lesion region. Finally, the SA-Net is employed to recognize benign and malignant tumors in images. Both the CRA-ENet and SA-Net are novel models explicitly designed for this study. The CRA-ENet is a network derived from ENet through pruning, convolutional replacement, and the incorporation of a hybrid attention mechanism, which can accurately segment breast tumor boundaries, allowing precise localization of lesion regions. The SA-Net is an innovative multi-branch parallel lightweight network that can enable effective artifact suppression and extraction of lesion-relevant features, significantly improving classification accuracy. In experimental, to validate the performance and efficiency of the proposed framework on clinical data provided by Yunnan Cancer Hospital based on the public dataset BUSI. Compared with simple classification utilizing only the classification network, the proposed approach improved accuracy, precision, and F1 score by 4.01%, 3.98%, and 4.00%, respectively, demonstrating the advantages of the framework.

Objective: The endoscopic findings associated with Helicobacter pylori –naïve status, current infection or past infection are an area of ongoing interest. Previous studies have investigated parameters with a potential diagnostic value. The aim of this study was to perform meta-analysis of the available literature to validate the diagnostic accuracy of mucosal features proposed in the Kyoto classification. Data sources: The databases of MEDLINE and Embase, clinicalTrials.gov and the Cochrane Library were systematically searched for relevant studies from October 1999 to October 2019. Methods: A bivariate random effects model was used to produce pooled diagnostic accuracy calculations for each of the studied endoscopic findings. Diagnostic odds ratios and sensitivity and specificity characteristics were calculated to identify significant predictors of H pylori status. Results: Meta-analysis included 4380 patients in 15 studies. The most significant predictor of an H pylori -naïve status was a regular arrangement of collecting venules (diagnostic odds ratio 55.0, sensitivity 78.3%, specificity 93.8%). Predictors of active H pylori infection were mucosal oedema (18.1, 63.7%, 91.1%) and diffuse redness (14.4, 66.5%, 89.0%). Map-like redness had high specificity for previous H pylori eradication (99.0%), but poor specificity (13.0%). Conclusion: The regular arrangement of collecting venules, mucosal oedema, diffuse redness and map-like redness are important endoscopic findings for determining H pylori status. This meta-analysis provides a tentative basis for developing future endoscopic classification systems. [ABSTRACT FROM AUTHOR]

Automated diagnosis of complex breast ultrasound images using deep learning has important clinical implications. In this study, we propose a novel breast cancer diagnosis framework, the CAS framework, based on lesion region to effectively recognize benign and malignant tumors. Firstly, train the segmentation network CRA-ENet on a public dataset with masks and save the trained weights. Then, the weights are used to segment clinical data without masks to obtain masks. Next, cover the masks on the corresponding original images to obtain ultrasound images that only include the lesion region. Finally, the SA-Net is employed to recognize benign and malignant tumors in images. Both the CRA-ENet and SA-Net are novel models explicitly designed for this study. The CRA-ENet is a network derived from ENet through pruning, convolutional replacement, and the incorporation of a hybrid attention mechanism, which can accurately segment breast tumor boundaries, allowing precise localization of lesion regions. The SA-Net is an innovative multi-branch parallel lightweight network that can enable effective artifact suppression and extraction of lesion-relevant features, significantly improving classification accuracy. In experimental, to validate the performance and efficiency of the proposed framework on clinical data provided by Yunnan Cancer Hospital based on the public dataset BUSI. Compared with simple classification utilizing only the classification network, the proposed approach improved accuracy, precision, and F1 score by 4.01%, 3.98%, and 4.00%, respectively, demonstrating the advantages of the framework. [ABSTRACT FROM AUTHOR]