No abstract

Lung cancer remains the most commonly diagnosed cancer and the leading cause of death from cancer. Recent research shows that the human eye can provide useful information about one’s health status, but few studies have revealed that the eye’s features are associated with the risk of cancer. The aims of this paper are to explore the association between scleral features and lung neoplasms and develop a non-invasive artificial intelligence (AI) method for detecting lung neoplasms based on scleral images. A novel instrument was specially developed to take the reflection-free scleral images. Then, various algorithms and different strategies were applied to find the most effective deep learning algorithm. Ultimately, the detection method based on scleral images and the multi-instance learning (MIL) model was developed to predict benign or malignant lung neoplasms. From March 2017 to January 2019, 3923 subjects were recruited for the experiment. Using the pathological diagnosis of bronchoscopy as the gold standard, 95 participants were enrolled to take scleral image screens, and 950 scleral images were fed to AI analysis. Our non-invasive AI method had an AUC of 0.897 ± 0.041(95% CI), a sensitivity of 0.836 ± 0.048 (95% CI), and a specificity of 0.828 ± 0.095 (95% CI) for distinguishing between benign and malignant lung nodules. This study suggested that scleral features such as blood vessels may be associated with lung cancer, and the non-invasive AI method based on scleral images can assist in lung neoplasm detection. This technique may hold promise for evaluating the risk of lung cancer in an asymptomatic population in areas with a shortage of medical resources and as a cost-effective adjunctive tool for LDCT screening at hospitals.

No abstract

Ultrasound (US) examination has been commonly utilized in clinical practice for assessing the rheumatoid arthritis (RA) activity, which is hampered by low intra-observer and inter-observer agreement as well as considerable time and expense to train experienced radiologists. Here, we present Rheumatoid ArthriTIs kNowledge Guided (RATING) deep learning model that scores RA activity and generates interpretable features to assist radiologists' decision-making. RATING model was developed using paired grey-scale US and color Doppler US images, and tested both in a prospective setting and on power Doppler US images collected from an external medical center. RATING model generalized well across settings, predicting the EOSS combined score with an accuracy of 86.1% (95% confidence interval (CI)=82.5%-90.1%) in the prospective setting and 85.0% (95% CI=80.5%-89.1%) on the US images collected from the external medical center. Prospective experiments demonstrated that RATING model improved the combined score accuracy of radiologists from 41.4% to 64.0% on average. Automated AI models for the assessment of RA may facilitate US RA examination and provide support for clinical decision-making.