Enterprise AI Analysis
Artificial intelligence applied in identifying left ventricular walls in myocardial perfusion scintigraphy images: Pilot study
Authors: Solange Amorim Nogueira, Fernanda Ambrogi B. Luz, Thiago Fellipe O. Camargo, Julio Cesar S. Oliveira, Guilherme Carvalho Campos Neto, Felipe Brazao F. Carvalhaes, Marcio Rodrigues C. Reis, Paulo Victor Santos, Giovanna Souza Mendes, Rafael Maffei Loureiro, Daniel Tornieri, Viviane M. Gomes Pacheco, Antonio Paulo Coimbra, Wesley Pacheco Calixto
This paper proposes the use of artificial intelligence techniques, specifically the nnU-Net convolutional neural network, to improve the identification of left ventricular walls in images of myocardial perfusion scintigraphy, with the objective of improving the diagnosis and treatment of coronary artery disease. The methodology included data collection in a clinical environment, followed by data preparation and analysis using the 3D Slicer Platform for manual segmentation, and subsequently, the application of artificial intelligence models for automated segmentation, focusing on the efficiency of identifying the walls of the left ventricular. A total of 83 clinical routine exams were collected, each exam containing 50 slices, which is 4,150 images. The results demonstrate the efficiency of the proposed artificial intelligence model, with a Dice coefficient of 87% and an average Intersection over Union of 0.8, reflecting high agreement with the manual segmentations produced by experts and surpassing traditional interpretation methods. The internal and external validation of the model corroborates its future applicability in real clinical scenarios, offering a new perspective in the analysis of myocardial perfusion scintigraphy images. The integration of artificial intelligence into the process of analyzing myocardial perfusion scintigraphy images represents a significant advancement in diagnostic accuracy, promoting substantial improvements in the interpretation of medical images, and establishing a foundation for future research and clinical applications, such as artifact correction.
Executive Impact Summary
Our AI-powered solution for left ventricular wall identification achieves high accuracy, significantly enhancing diagnostic precision and streamlining medical imaging workflows.
0
Dice Coefficient (External Validation)
0
Average Intersection over Union
0
Clinical Exams Analyzed
0
Total Images Processed
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Coronary Artery Disease & MPS Context
Coronary Artery Disease (CAD) remains a leading cause of mortality globally, with early detection being crucial for effective treatment. Myocardial Perfusion Scintigraphy (MPS) is a vital non-invasive imaging technique used to diagnose and assess CAD by visualizing myocardial blood flow.
However, MPS images are susceptible to various artifacts arising from patient factors, equipment, or technical issues, which can complicate interpretation. The integration of Artificial Intelligence (AI), particularly deep learning, offers a transformative approach to overcome these challenges, enhancing diagnostic accuracy and efficiency in cardiology.
AI-Enhanced Myocardial Perfusion Scintigraphy Workflow
Our methodology outlines a comprehensive process for identifying left ventricular walls in MPS images using AI. Starting from patient preparation and image acquisition, the workflow integrates advanced computational systems for data management, anonymization, and manual segmentation, culminating in automated AI-driven analysis.
Enterprise Process Flow: AI in MPS Image Analysis
Examination Equipment
→
PACS
→
Gateway
→
Anonymizer
→
VNA
→
DsM (Data Management)
→
3D Slicer (Manual Segmentation)
→
AI Model (nnU-Net Training)
→
AI Model (Prediction)
→
APP Viewer (Results & Comparison)
This streamlined process ensures data integrity, patient privacy, and facilitates the development and validation of robust AI models for medical image interpretation.
Performance & Expert Validation
The nnU-Net model demonstrated exceptional performance in identifying left ventricular walls. With a Dice coefficient of 88.46% and an IoU of 0.8 in external validation, the model significantly aligns with expert manual segmentations, showcasing its reliability and potential to surpass traditional methods.
0
AI Model's Dice Coefficient (External Validation)
Reflecting high agreement with manual segmentations, indicating superior diagnostic accuracy.
| Agreement Scale | AI Model (Evaluations) | Manual Mask (Evaluations) |
|---|---|---|
| I agree 100% with the mask | 4 | 4 |
| I agree 80% with the mask | 5 | 4 |
| I agree 60% with the mask | 0 | 1 |
Qualitative assessment by nuclear medicine specialists revealed high agreement, with most evaluations indicating 80% to 100% concordance between AI-generated and manually created masks. This validation confirms the model's practical efficiency and effectiveness in clinical settings.
Advancing Diagnostic Capabilities
This pilot study successfully demonstrated the feasibility and efficiency of applying AI to enhance the identification of left ventricular walls in MPS images. The robust performance of the nnU-Net model, validated both internally and externally, signifies a substantial advancement in medical image interpretation.
The research provides a solid foundation for future work, including artifact correction and broader clinical applications. The integration of AI promises to revolutionize diagnostic accuracy, offering invaluable support to medical professionals in the fight against cardiovascular disease.
Calculate Your Potential AI ROI
Estimate the efficiency gains and cost savings your enterprise could achieve by integrating advanced AI solutions.
Your AI Implementation Roadmap
A clear path to integrating AI, from discovery to sustained impact and innovation.
Phase 1: Discovery & Strategy
In-depth analysis of current workflows, identification of AI opportunities, and development of a tailored implementation strategy.
Phase 2: Pilot & Validation
Deployment of a proof-of-concept AI solution, rigorous testing, and validation against key performance indicators.
Phase 3: Full-Scale Integration
Seamless integration of the validated AI solution into your existing enterprise systems and processes.
Phase 4: Optimization & Scaling
Continuous monitoring, performance optimization, and strategic scaling of AI capabilities across your organization.
Ready to Transform Your Enterprise with AI?
Connect with our AI specialists to discuss how these insights can be applied to your specific business challenges.
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