RESUMO
Objectives: Localized autoimmune pancreatitis is difficult to differentiate from pancreatic ductal adenocarcinoma on endoscopic ultrasound images. In recent years, deep learning methods have improved the diagnosis of diseases. Hence, we developed a special cross-validation framework to search for effective methodologies of deep learning in distinguishing autoimmune pancreatitis from pancreatic ductal adenocarcinoma on endoscopic ultrasound images. Methods: Data from 24 patients diagnosed with localized autoimmune pancreatitis (8751 images) and 61 patients diagnosed with pancreatic ductal adenocarcinoma (20,584 images) were collected from 2016 to 2022. We applied transfer learning to a convolutional neural network called ResNet152, together with our innovative imaging method contributing to data augmentation and temporal data process. We divided patients into five groups according to different factors for 5-fold cross-validation, where the ordered and balanced datasets were created for the performance evaluations. Results: ResNet152 surpassed the endoscopists in all evaluation metrics with almost all datasets. Interestingly, when the dataset is balanced according to the factor of the endoscopists' diagnostic accuracy, the area under the receiver operating characteristic curve and accuracy were highest at 0.85 and 0.80, respectively. Conclusions: It is deduced that image features useful for ResNet152 correlate with those used by endoscopists for their diagnoses. This finding may contribute to sample-efficient dataset preparation to train convolutional neural networks for endoscopic ultrasonography-imaging diagnosis.
RESUMO
We investigated the association of salt intake with lifestyle-related diseases and also the association of habitually consumed foods with salt intake. A cross-sectional study was conducted using data from a baseline survey of 2,129 residents of Yonezawa city (980 males and 1,149 females), Yamagata prefecture. The residents were divided into three groups based on their estimated daily salt intake: low, medium, and high. In both genders, the prevalence of hypertension and diabetes increased in the order of high > medium > low salt intake (trend p<0.001). Similar trends were observed in the prevalence of hyperlipidemia in females and metabolic syndrome in males. The prevalence of diabetes in the high salt intake group was significantly higher than that in the control group (matched from the low and medium salt intake groups), even when confounding factors were excluded by propensity score matching (p<0.01). Network analysis showed that the low salt intake group had a greater tendency to habitually consume various vegetables than the high salt intake group. Our findings reveal that the prevalence of lifestyle-related diseases increased with higher salt intake. We speculate that a dietary shift to multiple vegetable consumption could have salt-lowering effects.
RESUMO
BACKGROUND/AIM: The current study aimed to identify biomarkers for differentiating between patients with oral cancer (OC) and healthy controls (HCs) on the basis of the comprehensive proteomic analyses of saliva samples by using liquid chromatography-mass spectrometry (LC-MS/MS). PATIENTS AND METHODS: Unstimulated saliva samples were collected from 39 patients with OC and from 31 HCs. Proteins in the saliva were comprehensively analyzed using LC-MS/MS. To differentiate between patients with OC and HCs, a multiple logistic regression model was developed for evaluating the discriminatory ability of a combination of multiple markers. RESULTS: A total of 23 proteins were significantly differentially expressed between the patients with OC and the HCs. Six out of the 23 proteins, namely α-2-macroglobulin-like protein 1, cornulin, hemoglobin subunit ß, Ig k chain V-II region Vk167, kininogen-1 and transmembrane protease serine 11D, were selected using the forward-selection method and applied to the multiple logistic regression model. The area under the curve for discriminating between patients with OC and HCs was 0.957 when the combination of the six metabolites was used (95% confidence interval=0.915-0.998; p<0.001). Furthermore, these candidate proteins did not show a stage-specific difference. CONCLUSION: The results of the current study showed that six salivary proteins are potential non-invasive biomarkers for OC screening.