Enterprise AI Analysis
Artificial intelligence-assisted endoscopic ultrasound diagnosis of esophageal subepithelial lesions
This study details the development and evaluation of an Artificial Intelligence (AI) model for endoscopic ultrasound (EUS) diagnosis of esophageal subepithelial lesions (SELs). Leveraging YOLOv8s-seg and MobileNetv2, the AI system demonstrates high potential for detecting lesions and accurately identifying their originating layers. Notably, it significantly enhances the diagnostic capabilities of junior endoscopists and offers a considerable speed advantage over human diagnosis, thereby improving diagnostic efficiency and reducing subjective bias in clinical practice. The model's accuracy in distinguishing between critical originating layers (second/third vs. fourth) is comparable to senior endoscopists, suggesting its utility as a valuable 'secondary observer' in EUS examinations.
Revolutionizing Esophageal SEL Diagnosis
Our AI model offers a significant leap forward in diagnostic precision and efficiency for esophageal subepithelial lesions, addressing critical challenges faced by endoscopists. This translates directly into improved patient outcomes, optimized resource allocation, and a standardized approach to complex EUS interpretations.
76.5%
Diagnostic Accuracy
0.01s/image
Diagnosis Speed
+15% Accuracy
Junior Endoscopist Boost
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
92.2%
The AI model achieved a precision of 92.2% in lesion detection, indicating high accuracy in identifying actual lesions among all detected entities.
Originating Layer Recognition Accuracy Comparison
| Diagnostician | Accuracy (%) | Significance (P-value) |
|---|---|---|
| AI model | 55.2% (48.0–62.2%) | Reference |
| All senior endoscopists | 59.3% (51.8–65.9%) | 0.057 |
| All junior endoscopists | 45.1% (37.8–52.0%) | 0.043* |
*P<0.05 compared to AI model. AI model performance is comparable to senior endoscopists and significantly outperforms junior endoscopists in ternary classification.
76.5%
For the critical distinction between layers 2/3 vs. layer 4, the AI model achieved 76.5% accuracy, comparable to senior endoscopists (74.9–79.8%) and superior to junior endoscopists (65.6–66.7%).
EUS-AI Model Development Process
1445 Patients, 1605 EUS Images Collected
→
Lesion Detection Model (YOLOv8s-seg)
→
Originating Layer Identification Model (MobileNetv2)
→
Training, Validation, Test Split (8:1:1)
→
Model Evaluation & Comparison
The AI model was developed using a comprehensive dataset of esophageal SEL EUS images. The process involved distinct models for lesion detection and layer identification, rigorously split into training, validation, and test sets to ensure robust evaluation.
1445
Total number of patients with esophageal SELs included in the study, contributing to a robust dataset for AI model training and validation.
Algorithm Selection for Lesion Detection
| Algorithm | mAP@0.5 | F1-Score |
|---|---|---|
| YOLOv8s-seg | 0.832 | 81.9% |
| YOLOv8 | 0.785 | 78.2% |
| YOLOv5s | 0.751 | 74.0% |
YOLOv8s-seg was chosen for its superior performance in both detection and segmentation.
Enhanced Diagnostic Accuracy for Junior Endoscopists
Scenario: A junior endoscopist typically struggles with accurate SEL layer identification due to limited experience, leading to potential misdiagnosis or delayed treatment decisions.
Solution: The EUS-AI model provides real-time, objective analysis. In binary classification (layers 2/3 vs. 4), the AI's 76.5% accuracy significantly surpasses junior endoscopists' 65.6–66.7%, effectively bridging the experience gap.
Outcome: With AI assistance, junior endoscopists can achieve diagnostic accuracy comparable to senior colleagues, reducing inter-observer variability and improving initial management decisions, ultimately leading to better patient care and accelerated learning.
κ=0.430
Interobserver agreement among senior endoscopists for ternary classification (κ=0.430) was higher than for junior endoscopists (κ=0.227), highlighting the need for standardized tools like AI.
400x
The AI model diagnoses an image in 0.01s, compared to 4.03s for senior endoscopists and 6.78s for junior endoscopists, offering a massive speed advantage.
Calculate Your Potential AI Savings
Estimate the efficiency gains and cost savings your enterprise could realize by integrating our AI-powered diagnostic solutions. Adjust the parameters below to see a customized impact.
Annual Cost Savings
$0
Annual Hours Reclaimed
0
Your AI Implementation Journey
A structured approach to integrating AI into your diagnostic workflow, ensuring a seamless transition and maximum impact.
Phase 1: Discovery & Customization
Initial consultation, data assessment, and tailoring the AI model to your specific EUS image characteristics and clinical protocols.
Phase 2: Integration & Training
Seamless integration of the AI system with existing PACS/EUS platforms, followed by comprehensive training for your clinical team.
Phase 3: Pilot Deployment & Optimization
Launch a pilot program, gather feedback, and fine-tune the AI model for optimal performance within your operational environment.
Phase 4: Full-Scale Rollout & Continuous Support
Expand AI deployment across all relevant departments, with ongoing monitoring, updates, and expert support to ensure sustained value.
Ready to Transform Your Diagnostic Capabilities?
Unlock the full potential of AI-assisted EUS for esophageal SELs. Schedule a personalized consultation to discuss how our solution can integrate with your practice.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat