Search PubMed⌕ Search
Home/Archive/PubMed

PubMed · 42720264

Development and Crossover Evaluation of an Artificial Intelligence-Assisted System for Solid Pancreatic Lesion Detection and Pancreatic Parenchyma Recognition in Endoscopic Ultrasonography (With Video).

Toshio Fujisawa·Takamichi Kuwahara·Tatsuya Sato·Yusuke Takasaki·Nozomi Okuno·Ryunosuke Hakuta·Ko Tomishima·Shin Haba·Naminatsu Takahara·Yukiko Ito·Yousuke Nakai·Kazuo Hara·Hiroyuki Isayama
View original source ↗Export metadata ↓

Abstract

BACKGROUND AND STUDY AIMS: Pancreatobiliary endoscopic ultrasonography (EUS) is technically demanding, and supervised training opportunities are limited. We developed an artificial intelligence (AI) overlay system for detecting solid pancreatic lesions (SPL) and recognizing pancreatic parenchyma (PP) and evaluated its effect on reader performance. PATIENTS AND METHODS: Across six centers, two deep learning-based models were trained using expert-annotated EUS frames. We then conducted a randomized, two-sequence, two-period crossover reader study in which eight endosonographers (five novices and three experts) interpreted image sets with and without AI assistance. The primary endpoint was superiority of sensitivity for SPL detection among novices; key secondary endpoints included specificity and PP recognition. RESULTS: From 118 patients, 120 SPL-positive/negative image sets and 160 PP-positive/negative image sets were constructed. Among novices, AI assistance improved SPL detection sensitivity (88.7% vs. 76.8%, p&#x2009;<&#x2009;0.001) and accuracy (86.4% vs. 78.7%), while specificity met the predefined noninferiority criterion (84.2% vs. 80.5%, p&#x2009;<&#x2009;0.001). For PP recognition, sensitivity increased numerically (86.3% vs. 83.3%) but did not meet the predefined superiority criterion (p&#x2009;=&#x2009;0.095); specificity met the noninferiority criterion (87.8% vs. 81.0%), and accuracy increased from 82.1% to 87.0%. Among experts, sensitivity was maintained for both tasks, whereas specificity increased with AI assistance. CONCLUSIONS: AI assistance improved SPL detection among novice endosonographers. For PP recognition, sensitivity increased without reaching statistical superiority, whereas specificity met the predefined noninferiority criterion. These findings support a potential adjunctive role for AI in EUS interpretation.

Explore related subjects

HumansEndosonographyCross-Over StudiesFemaleArtificial IntelligenceMalePancreatic NeoplasmsMiddle AgedAgedSensitivity and SpecificityPancreasPancreatic Diseasesartificial intelligenceendoscopic ultrasonographynovicepancreatic parenchymasolid pancreatic lesionJournal ArticleMulticenter StudyRandomized Controlled TrialVideo-Audio Media

Keep this discovery

BibTeXRIS

Toshio Fujisawa, Takamichi Kuwahara, Tatsuya Sato, Yusuke Takasaki, Nozomi Okuno, Ryunosuke Hakuta, Ko Tomishima, Shin Haba, Naminatsu Takahara, Yukiko Ito, Yousuke Nakai, Kazuo Hara, Hiroyuki Isayama. 2026. Development and Crossover Evaluation of an Artificial Intelligence-Assisted System for Solid Pancreatic Lesion Detection and Pancreatic Parenchyma Recognition in Endoscopic Ultrasonography (With Video).. https://doi.org/10.1111/den.70282

Cite the original work for its findings. Save a collection to share your selection of sources.

Discover connections

Aged subjectArtificial Intelligence subjectCross-Over Studies subjectFemale subjectHumans subjectJournal Article subjectMale subjectMiddle Aged subjectMulticenter Study subjectRandomized controlled trials study

Connections use source metadata and explicit phrase matches, not verified experimental comparisons.

KEEP EXPLORING

Related citations

PubMed2026

Endoscopic Ultrasound-Guided Versus Transjugular Portal Pressure Measurements: Systematic Review and Meta-Analysis.

PURPOSE: Published reviews of endoscopic ultrasound-guided portal pressure gradient (EUS-PPG) have emphasized feasibility and safety. We performed a systematic review and meta-analysis specifically to evaluate how closely EUS-based portal pressure measurements track invasive comparator measurements in prospective paired studies and to summarize agreement, technical success, and adverse events. METHODS: We searched major databases through January 2026 for prospective cohorts reporting same-patient EUS-based portal pressure measurement and invasive hemodynamic measurements. Correlations were pooled with random-effects models and analyzed separately for studies comparing EUS-PPG with hepatic venous pressure gradient (HVPG) and studies comparing EUS-based portal measurements with direct portal venous pressure. Agreement and threshold discordance were summarized descriptively. RESULTS: Six prospective cohorts (127 attempted procedures) were included. In studies using HVPG as the comparator, the pooled correlation was 0.82 (95% CI, 0.72-0.89; I2&#x2009;=&#x2009;0%). In studies comparing EUS-based portal measurements with direct portal venous pressure, the pooled correlation was 0.86 (95% CI, 0.72-0.93; I2&#x2009;=&#x2009;16.9%). Technical success was 95.3%. EUS-PPG-attributed adverse events occurred in 2.4% of procedures, with no procedure-related deaths. Agreement data were limited. Reported limits of agreement were wide (approximately -&#xa0;6 to&#x2009;+&#x2009;7&#xa0;mmHg), and discrepancies of 5&#xa0;mmHg or greater occurred in 4 of 30 paired measurements. CONCLUSIONS: EUS-based portal pressure measurement is feasible and shows a strong association with invasive hemodynamic comparators, but the evidence base remains small (six cohorts, 127 attempted procedures). Further study will be necessary to establish patient-level agreement, procedural reproducibility, EUS-specific clinically significant portal hypertension thresholds, and whether HVPG-based decision thresholds can be transferred to EUS-derived measurements.

Humans↗
PubMed2026

Efficacy and safety of endoscopic ultrasound-guided choledochoduodenostomy compared with endoscopic ultrasound-guided gallbladder drainage for palliation of malignant distal biliary obstruction: a systematic review and meta-analysis.

BACKGROUND AND AIMS: When ERCP is not feasible or fails in the palliation of malignant distal biliary obstruction (MDBO), EUS-guided choledochoduodenostomy (EUS-CDS) and EUS-guided gallbladder drainage (EUS-GBD) are viable alternatives. We conducted a systematic review and meta-analysis comparing the safety and efficacy of the 2 techniques for the palliation of MDBO. METHODS: Multiple databases were searched through November 2025 for studies that reported outcomes of EUS-CDS and EUS-GDB in patients with MDBO. A meta-analysis was performed to determine pooled proportions and relative risk (RR) with 95% CIs. We compared the rates of technical and clinical success, overall adverse events (AEs), and lumen-apposing metal stent dysfunction. A random-effects model was used for our meta-analysis, and heterogeneity was assessed using the I2 (%) statistics. RESULTS: Five studies (352 patients; EUS-CDS: 193 and EUS-GBD: 159) were included. Technical success was 93.3% (95% CI, 70.6-98.8) for EUS-CDS and 95.9% (95% CI, 90.0-98.4) for EUS-GBD (RR, 1.02; 95% CI, 0.94-1.10; P = .6). Clinical success was 90.1% (95% CI, 84.7-93.7) versus 86.6% (95% CI, 80.3-91.0) (RR, 0.97; 95% CI, 0.90-1.05; P = .4). There were no significant differences in overall AEs (19.7% vs 17.6%; RR, 0.93; 95% CI, 0.58-1.48; P = .8), severe AEs (11.0% vs 8.3%; RR, 0.69; 95% CI, 0.33-1.44; P = .3), or stent dysfunction (15.0% vs 14.5%; RR, 0.95; 95% CI, 0.35-2.58; P = .9). CONCLUSIONS: EUS-GBD appears comparable to EUS-CDS in terms of technical and clinical success, AEs, and stent dysfunction. Further prospective studies are warranted to corroborate our findings.

Humans↗
PubMed2026

Comparative efficacy and safety of bi-flanged metal stents versus lumen-apposing metal stents for endoscopic drainage of pancreatic fluid collections: a systematic review and meta-analysis.

INTRODUCTION: Pancreatic fluid collections (PFCs), particularly walled-off necrosis, are common complications of acute pancreatitis that often require endoscopic drainage. Bi-flanged metal stents (BFMS; NAGI; Taewoong Medical, Gyenoggi-do, Korea) and electrocautery-enhanced lumen-apposing metal stents (LAMS; AXIOS, Boston Scientific Corporation, Marlborough, Massachusetts, USA) are frequently used, but comparative data remain limited. This study aims to compare the efficacy and safety of BFMS and LAMS in the endoscopic drainage of PFCs. METHODS: This meta-analysis followed the Cochrane Handbook for Systematic Reviews of Interventions and Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Comprehensive database searches were conducted through November 2024 to identify studies comparing BFMS and LAMS for endoscopic ultrasound-guided drainage of PFCs. Outcomes were pooled using a random-effects model with RevMan Web, and statistical significance was defined as a P value less than 0.05. RESULTS: Three studies ( n &#x2005;=&#x2005;627; 329 BFMS and 298 LAMS) met inclusion criteria. No significant differences were observed between BFMS and LAMS for technical success [odds ratio (OR): 1.16; 95% confidence interval (CI): 0.55-2.43] or clinical success (OR: 0.97; 95% CI: 0.51-1.87). Similarly, there were no differences in walled-off necrosis recurrence (OR: 2.01; 95% CI: 0.17-24.02), number of direct endoscopic necrosectomy sessions (OR: 0.52; 95% CI: 0.05-5.12), or mean number of endoscopic procedures (mean difference: 0.18, 95% CI: 2.08-2.45). Adverse events were also comparable between groups, including bleeding (OR: 0.64), infection (OR: 1.18), stent migration (OR: 1.83), and stent occlusion/dysfunction (OR: 1.71). CONCLUSION: BFMS and LAMS provide equivalent efficacy and safety in the endoscopic ultrasound-guided drainage of PFCs. Either stent type represents a viable therapeutic option. Further large-scale prospective studies are warranted to refine stent selection strategies.

Humans↗