Background/Aims: Gastric antral vascular ectasia (GAVE) is a rare but important cause of gastrointestinal (GI) bleeding. The clinical course of GAVE is not well-known, and recurrent bleeding from GAVE is a therapeutic challenge. Therefore, we investigated the clinical course of GAVE and identified the risk factors for bleeding from it. Methods: We retrospectively reviewed the records of patients diagnosed with GAVE using upper GI endoscopy at Asan Medical Center between January 2004 and December 2019 and evaluated the clinical course and risk factors for bleeding from GAVE. Results: Of the 348 patients (mean age, 62.3 ± 10.7 years; male, 62%), bleeding from GAVE occurred in 123 (35%) patients during follow-up (median, 17.3 months; interquartile range [IQR], 4.2–46.6). GI bleeding from GAVE was significantly associated with Child–Pugh class B or C liver cirrhosis (odds ratio [OR], 2.55; 95% confidence interval [CI], 1.57–4.16), chronic kidney disease (CKD) (OR, 2.77; 95% CI, 1.52–5.07), use of antithrombotic agents (OR, 2.34; 95% CI, 1.13–4.82), and involvement of the duodenal bulb (OR, 3.21; 95% CI, 1.76–5.86). Rebleeding occurred in 39 of 123 patients (32%), in whom CKD (OR, 2.55; 95% CI, 1.12–5.81) was significantly associated with rebleeding. Endoscopic hemostasis was most commonly performed using argon plasma coagulation, and the median number of endoscopic hemostasis performed was 2 (IQR, 1–3). Conclusions A careful follow-up for bleeding is needed in GAVE patients with liver cirrhosis, CKD, use of antithrombotic agents, and duodenal bulb involvement.

Purpose: Sentinel lymph node biopsy (SLNB) using dye and isotope (DUAL) is recommended over the dye-only (DYE) method after neoadjuvant chemotherapy (NCT) due to potentially lower false-negative rates. However, the long-term outcome of either method is unclear. We aimed to compare the long-term oncological outcomes of DYE versus DUAL SLNB methods in patients who received NCT. Materials and Methods: In this retrospective cohort study, 893 patients who underwent SLNB following NCT and had pathologically negative lymph nodes were included. After propensity score matching for cT, cN, and pT categories, 280 patients were in the DYE group and 560 in the DUAL group. Indigo carmine was used for dye and Tc-99m antimony trisulfate for isotope mapping. Results: Median follow-up was 75.6 months in the DYE group and 83.0 months in the DUAL group. Mean (±standard deviation) number of harvested sentinel nodes was 6.7 (±3.9) and 6.7 (±3.3) in the DYE and DUAL groups (p=0.875). Five-year distant metastasisfree survival was 95.2% in DYE group and 93.3% in DUAL group (hazard ratio [HR], 1.45; 95% confidence interval [CI], 0.82 to 2.57; p=0.192). Disease-free survival (HR, 0.97; 95% CI, 0.69 to 1.50; p=0.914) and overall survival (HR, 0.98; 95% CI, 0.56 to 1.69; p=0.954) were not significantly different. Axillary recurrence rate was 1.8% and 2.5% in DYE and DUAL groups (p=0.647). Conclusion Long-term oncological outcomes did not significantly differ between DYE and DUAL SLNB methods. The dye-only method can be safely recommended for breast cancer patients who received NCT.

Background: This study aimed to evaluate the applicability of deep learning technology to thyroid ultrasound images for classification of thyroid nodules. Methods: This retrospective analysis included ultrasound images of patients with thyroid nodules investigated by fine-needle aspiration at the thyroid clinic of a single center from April 2010 to September 2012. Thyroid nodules with cytopathologic results of Bethesda category V (suspicious for malignancy) or VI (malignant) were defined as thyroid cancer. Multiple deep learning algorithms based on convolutional neural networks (CNNs) —ResNet, DenseNet, and EfficientNet—were utilized, and Siamese neural networks facilitated multi-view analysis of paired transverse and longitudinal ultrasound images. Results: Among 1,048 analyzed thyroid nodules from 943 patients, 306 (29%) were identified as thyroid cancer. In a subgroup analysis of transverse and longitudinal images, longitudinal images showed superior prediction ability. Multi-view modeling, based on paired transverse and longitudinal images, significantly improved the model performance; with an accuracy of 0.82 (95% confidence intervals [CI], 0.80 to 0.86) with ResNet50, 0.83 (95% CI, 0.83 to 0.88) with DenseNet201, and 0.81 (95% CI, 0.79 to 0.84) with EfficientNetv2_ s. Training with high-resolution images obtained using the latest equipment tended to improve model performance in association with increased sensitivity. Conclusion CNN algorithms applied to ultrasound images demonstrated substantial accuracy in thyroid nodule classification, indicating their potential as valuable tools for diagnosing thyroid cancer. However, in real-world clinical settings, it is important to aware that model performance may vary depending on the quality of images acquired by different physicians and imaging devices.

Background: This study aimed to evaluate the applicability of deep learning technology to thyroid ultrasound images for classification of thyroid nodules. Methods: This retrospective analysis included ultrasound images of patients with thyroid nodules investigated by fine-needle aspiration at the thyroid clinic of a single center from April 2010 to September 2012. Thyroid nodules with cytopathologic results of Bethesda category V (suspicious for malignancy) or VI (malignant) were defined as thyroid cancer. Multiple deep learning algorithms based on convolutional neural networks (CNNs) —ResNet, DenseNet, and EfficientNet—were utilized, and Siamese neural networks facilitated multi-view analysis of paired transverse and longitudinal ultrasound images. Results: Among 1,048 analyzed thyroid nodules from 943 patients, 306 (29%) were identified as thyroid cancer. In a subgroup analysis of transverse and longitudinal images, longitudinal images showed superior prediction ability. Multi-view modeling, based on paired transverse and longitudinal images, significantly improved the model performance; with an accuracy of 0.82 (95% confidence intervals [CI], 0.80 to 0.86) with ResNet50, 0.83 (95% CI, 0.83 to 0.88) with DenseNet201, and 0.81 (95% CI, 0.79 to 0.84) with EfficientNetv2_ s. Training with high-resolution images obtained using the latest equipment tended to improve model performance in association with increased sensitivity. Conclusion CNN algorithms applied to ultrasound images demonstrated substantial accuracy in thyroid nodule classification, indicating their potential as valuable tools for diagnosing thyroid cancer. However, in real-world clinical settings, it is important to aware that model performance may vary depending on the quality of images acquired by different physicians and imaging devices.

Background: This study aimed to evaluate the applicability of deep learning technology to thyroid ultrasound images for classification of thyroid nodules. Methods: This retrospective analysis included ultrasound images of patients with thyroid nodules investigated by fine-needle aspiration at the thyroid clinic of a single center from April 2010 to September 2012. Thyroid nodules with cytopathologic results of Bethesda category V (suspicious for malignancy) or VI (malignant) were defined as thyroid cancer. Multiple deep learning algorithms based on convolutional neural networks (CNNs) —ResNet, DenseNet, and EfficientNet—were utilized, and Siamese neural networks facilitated multi-view analysis of paired transverse and longitudinal ultrasound images. Results: Among 1,048 analyzed thyroid nodules from 943 patients, 306 (29%) were identified as thyroid cancer. In a subgroup analysis of transverse and longitudinal images, longitudinal images showed superior prediction ability. Multi-view modeling, based on paired transverse and longitudinal images, significantly improved the model performance; with an accuracy of 0.82 (95% confidence intervals [CI], 0.80 to 0.86) with ResNet50, 0.83 (95% CI, 0.83 to 0.88) with DenseNet201, and 0.81 (95% CI, 0.79 to 0.84) with EfficientNetv2_ s. Training with high-resolution images obtained using the latest equipment tended to improve model performance in association with increased sensitivity. Conclusion CNN algorithms applied to ultrasound images demonstrated substantial accuracy in thyroid nodule classification, indicating their potential as valuable tools for diagnosing thyroid cancer. However, in real-world clinical settings, it is important to aware that model performance may vary depending on the quality of images acquired by different physicians and imaging devices.