Search

In this study, we developed a deep learning model to identify patients with tongue cancer based on a validated dataset comprising oral endoscopic images. We retrospectively constructed a dataset of 12,400 verified endoscopic images from five university hospitals in South Korea, collected between 2010 and 2020 with the participation of otolaryngologists. To calculate the probability of malignancy using various convolutional neural network (CNN) architectures, several deep learning models were developed. Of the 12,400 total images, 5576 images related to the tongue were extracted. The CNN models showed a mean area under the receiver operating characteristic curve (AUROC) of 0.845 and a mean area under the precision-recall curve (AUPRC) of 0.892. The results indicate that the best model was DenseNet169 (AUROC 0.895 and AUPRC 0.918). The deep learning model, general physicians, and oncology specialists had sensitivities of 81.1%, 77.3%, and 91.7%; specificities of 86.8%, 75.0%, and 90.9%; and accuracies of 84.7%, 75.9%, and 91.2%, respectively. Meanwhile, fair agreement between the oncologist and the developed model was shown for cancer diagnosis (kappa value = 0.685). The deep learning model developed based on the verified endoscopic image dataset showed acceptable performance in tongue cancer diagnosis. [ABSTRACT FROM AUTHOR]

Background: Morphologic features of the proximal femur reminiscent of those seen in patients with femoroacetabular impingement (FAI) have been reported among asymptomatic individuals in Western populations, but whether this is the case in Asian populations is unknown.Questions/purposes: The purpose of this study was to determine the prevalence of radiographic findings in the proximal femur that are consistent with FAI in asymptomatic Korean volunteers.Methods: Two hundred asymptomatic volunteers with no prior hip surgery or childhood hip problems underwent three-view plain radiographs (pelvis AP view, Sugioka view, and 45° Dunn view) of both hips. There were 146 hips from male volunteers and 254 hips from female volunteers in the study. The mean age of all participants was 34.7 years (range, 21-49 years). Cam-type morphologic features were defined as the presence of the following on one or more of the three views: pistol-grip morphologic features, an osseous bump at the femoral head-neck junction, flattening of the femoral head-neck offset, or alpha angle greater than 55°. Pincer-type morphologic features were determined by radiographic signs, including crossover sign, deficient posterior wall sign, or lateral center-edge angle greater than 40°.Results: The prevalence of cam-type morphologic features seen on at least one radiograph was 38% (male, 57%; female, 26%). The prevalence of cam-type features (at least one positive cam-type feature) was 2.0% (male, 6%; female, 0%) on the pelvic AP view, 24% (male, 36%; female, 17%) on the Sugioka view, and 30% (male, 47%; female, 20%) on the 45° Dunn view. The prevalence of pincer-type morphologic features (at least one positive pincer-type feature) was 23% (male, 27%; female, 21%) on the pelvic AP view.Conclusion: The prevalence of FAI-related morphologic features in asymptomatic Asian populations was comparable to the prevalence in Western populations. Considering the high prevalence of radiographic hip findings reminiscent of FAI in asymptomatic Asian populations, it will be important to determine whether FAI-related morphologic features are a cause of hip pain when considering surgery in Asian patients. [ABSTRACT FROM AUTHOR]

Executive dysfunction assessed by the Wisconsin Card Sorting Test has been observed in patients with mesial temporal lobe epilepsy (TLE); however, the mechanism of executive impairment is unclear. We therefore investigated the potential contribution of the hippocampus toward executive dysfunction in 85 mesial TLE patients, and in a control group of 34 patients with neocortical TLE. Standardized regression-based methodology was used for correcting test-retest bias. We found that 56% of mesial TLE patients had impaired sorting performance, with 30% showing severe impairment. A lower full-scale intelligence quotient, older age, longer duration, and preoperative use of topiramate were significantly correlated with poorer preoperative sorting abilities. However, neither these variables nor postoperative discontinuation/reduction of topiramate were related to postoperative changes in sorting abilities. Only preoperative sorting abilities were negatively correlated with postoperative changes in executive performance. That is, patients with better preoperative executive ability showed greater postoperative deterioration in executive function, whereas those with poorer preoperative sorting ability had greater improvement in executive function after mesial temporal resection. There were no differences in card sorting ability between the mesial and neocortical TLE groups. In conclusion, our results suggest that impairment of card sorting performance in the mesial TLE group cannot be explained only by hippocampal dysfunction, but that other processes (possibly epileptic discharges propagated to the extratemporal area) must play a predominant role. Copyright © 2007 S. Karger AG, Basel [ABSTRACT FROM AUTHOR]