Medical Imaging Data Strategies for Catalyzing AI Medical Device Innovation

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Health Informatics

Background:

• Access to quality medical imaging data drives innovation in AI for patient care.
• Seamless communication among stakeholders is vital for establishing data quality standards.
• Heterogeneity in AI medical devices and data complexity pose challenges to clinical translation.

Purpose of the Study:

• To discuss challenges in medical imaging data quality and AI translation.
• To propose strategies for a data platform to ensure data quality and diversity.
• To envision a platform catalyzing AI-enabled medical device innovation.

Main Methods:

• Discussion of data characteristics impacting AI development (size, labels, diversity, sequestration, drift).
• Exploration of strategies for a data platform, including protocols, checklists, and guidelines.
• Consideration of study design, performance metrics, and data analytics monitoring.

Main Results:

• Identified challenges in data size, labels, diversity, sequestration, reuse, and drift.
• Proposed a data platform with quality assurance protocols and diversity assessment tools.
• Outlined strategies for efficient data usage and monitoring.

Conclusions:

• A comprehensive data platform can address critical challenges in medical imaging AI.
• Such a platform can foster efficient development and evaluation of AI medical devices.
• This approach aims to accelerate the clinical translation of safe and effective AI technologies.