Li X, et al. - In this retrospective, multicohort, diagnostic study using ultrasound images sets from three hospitals in China, researchers investigated whether use of deep convolutional neural network (DCNN) models could enhance the diagnostic accuracy of thyroid cancer by analysing sonographic imaging data from clinical ultrasounds. The training set comprised 131,731 ultrasound images from 17,627 patients with thyroid cancer and 180,668 images from 25,325 controls from the thyroid imaging database at Tianjin Cancer Hospital. Only individuals with suspected thyroid cancer had pathological examination to confirm diagnosis and images from anatomical sites that were judged as not having cancer were excluded from the training set. In the detection of thyroid cancer patients, a high performance was displayed by the model in the validation sets tested, with area under the curve values of 0·947 (95% CI 0·935–0·959), 0·912 (95% CI 0·865–0·958), and 0·908 (95% CI 0·891–0·925) for the Tianjin internal validation set, for the Jilin external validation set, and for the Weihai external validation set, respectively. Compared with a group of skilled radiologists, similar sensitivity and improved specificity were demonstrated by the DCNN model as far as detection of thyroid cancer patients was concerned.