Clinical Endoscopy最新文献

Background: Lymphadenopathy presents diagnostic challenges, particularly for the mediastinal and intra-abdominal lymph nodes (LNs). Endoscopic ultrasonography (EUS) has emerged as a tool for LN detection; however, its accuracy varies. To enhance the diagnostic performance and minimize medical costs, assisting LN assessment using EUS is necessary. Machine learning (ML) offers potential for medical image analysis. This study aimed to develop an ML model for classifying mediastinal and intra-abdominal LNs using gastrointestinal EUS.

Methods: EUS images of mediastinal and intra-abdominal LNs were randomly split into training and validation datasets. U-Net was selected for LN segmentation, and six deep-learning architectures were combined with the k-nearest-neighbor algorithm for LN classification. Physicians, comprising one expert group and one trainee group, reviewed the validation dataset and made individual diagnoses. A logistic regression model was generated based on LN features. We compared the diagnostic yields of ML, expert and trainee groups, logistic regression analysis, and a combination of the various methods mentioned above for diagnosing LNs.

Results: In total, 93 patients were enrolled, providing 630 images. The ResNet-50+logistic regression analysis+expert group achieved the best F1 score and sensitivity of 0.89 and 100.0%, respectively. Paired comparisons revealed that the combination outperformed both experts and trainees in terms of the area under the curve (p<0.01).

Conclusions: ML assists in predicting the mediastinal and intra-abdominal LNs based on gastrointestinal EUS images, particularly when combined with expert expertise and logistic regression models.

In recent years, the field of gastrointestinal endoscopy has grown significantly in Vietnam. Although Vietnamese gastrointestinal endoscopy still lags behind developed countries, such as Japan and Korea, the advancement throughout the country has been rapid. Current advanced gastrointestinal endoscopy techniques from around the world have been implemented in Vietnam. The number of endoscopists has also significantly increased. These advancements, particularly in interventional endoscopy, have primarily resulted from investments in equipment and tools, strategic personnel training, and international collaborations. Since the establishment of the Vietnamese Federation for Digestive Endoscopy in 2011, numerous international collaborations and training activities have accelerated the development of interventional gastrointestinal endoscopy in Vietnam.

Background: Endoscopic ultrasound and conventional computed tomography (CT) are useful for preoperative assessment of subepithelial tumors (SETs). However, surgical approaches are sometimes changed intraoperatively owing to unexpected discrepancies between the planned and actual visualizations of tumors because preoperative images are typically two-dimensional.

Methods: In this study, we evaluated the feasibility of morphological evaluation using three-dimensional (3D) reconstruction of SETs and its utility in preoperative assessments. We included 15 lesions with a diameter of 1 to 5 cm that were evaluated by CT and pathologically diagnosed as mesenchymal tumors. We examined the feasibility of 3D reconstruction of lesions by evaluating sphericity using CT images with reference to circularity, which was measured from endoscopic ultrasound still images. Furthermore, the predictability of planned surgery determined using 3D images was investigated.

Results: The median lesion diameter was 22 mm. There were 10, 3, and 2 lesions of gastrointestinal stromal tumors, leiomyomas, and schwannomas, respectively. 3D reconstruction was feasible for all lesions, with a median sphericity of 0.85, aligning with the median circularity (0.88). The predictability of the 3D-based surgical approach was 90%.

Conclusions: 3D reconstruction of SETs is feasible and useful for preoperative determination of the surgical approach.

Effective endoscopy training begins by assessing the trainee's experience and identifying their skill level: beginners, learners, independent practitioners, or experts. Beginners focus on basic tasks, such as cecal intubation, while advanced trainees refine efficiency and complex techniques. Training prioritizes conscious competence through deliberate practice, reflection, and verbalizing actions; this enhances mindfulness and procedural expertise. Clear communication, standardized terminology, and constructive feedback ensure safety, confidence, and skill retention. SMART objectives-specific, measurable, achievable, relevant, and timely-help structure sessions for skill development and mastery. Simulation-based models support training at all the levels. Beginners benefit from cost-effective low-fidelity bench models and virtual reality (VR) simulators, which offer realistic tactile feedback and customizable scenarios. Studies have shown that both low- and high-fidelity models can effectively teach basic skills, although VR is preferred for foundational training. Advanced trainees utilize animal-based models for therapeutic interventions, three-dimensional printed models for pathology-specific practice, and hybrid models that combine VR and physical elements for enhanced realism. Augmented reality and haptic feedback systems refine advanced skills, but face developmental and cost challenges. Mentored live patient models excel in real-world decision-making, but raise ethical concerns. Training is tailored to individual needs, and competency-based training ensures mastery at each stage, from beginners to advanced practitioners.

Approximately 90% of cases of gastric cancer (GC) are caused by Helicobacter pylori infection, and screening esophagogastroduodenoscopy is effective for secondary prevention of GC. Endoscopic findings of the stomach due to H. pylori infection vary widely, and the risk of GC varies according to each finding. GC risk is evaluated by combining endoscopic and histopathological findings. In the operative link on gastritis assessment and operative link on gastric intestinal metaplasia assessment staging, GC risk is determined by histopathological evaluation. In the endoscopic grading of gastric intestinal metaplasia, Kyoto classification, and modified Kyoto classification, the risk is considered based on endoscopic findings. However, evaluating endoscopic findings is challenging because the evaluation varies depending on the skill of the endoscopist. Similarly, histopathological findings can be assessed differently by different pathologists. Histopathological evaluation by biopsy carries a risk of bleeding; thus, simpler and less-invasive risk stratification methods are desirable. Artificial intelligence for risk stratification, which has the potential for improved accuracy and consistency, has been developed for endoscopic and histopathological evaluations. Appropriate GC risk stratification would benefit the economy and patients, and further evaluation of surveillance intervals tailored to individual risks is warranted.

Background: Surveillance strategies for small grade 1 rectal neuroendocrine tumors (G1 rNETs) after incomplete endoscopic resection (ER) remain controversial. We evaluated the long-term outcomes of patients with G1 rNET ≤1 cm after ER who did and did not undergo complete resection.

Methods: We retrospectively evaluated 441 patients with G1 rNETs measuring ≤1 cm after ER between 2011 and 2022. Patients were divided into complete and incomplete resection groups according to histopathological evaluation. Logistic regression analysis identified the risk factors for incomplete resection after ER.

Results: The mean follow-up intervals were 38.6 and 45.7 months in all patients and the incomplete resection group, respectively. No recurrences were observed during the follow-up period. The mean lesion size was 5.5 mm and the complete resection rate was 80.5% (n=355). In the logistic regression analysis, lesion size 5.1 to 10 mm (odds ratio [OR], 2.3; 95% confidence interval [CI], 1.245-4.203; p=0.008), multiple lesions (OR, 8.3; 95% CI, 1.247-54.774; p=0.029), and retroflexion view during the procedure (OR, 4.0; 95% CI, 1.668-9.615; p=0.002) were independent risk factors for incomplete resection.

Conclusions: The prognosis of G1 rNET ≤1 cm after ER was very good, regardless of the histopathological results.