Segmed and Verily Partner to Expand Real-World Imaging Data Access
Segmed has partnered with Verily to increase access to real-world medical imaging data. The collaboration aims to provide researchers and developers with the high-quality datasets needed to fuel AI model development and analytics for diagnostics and clinical research.
- Segmed's contribution to the partnership includes a repository of approximately 150 million de-identified imaging exams sourced from a global network of healthcare partners across five continents. - The medical imaging data will be integrated into Verily's Pre platform, an AI-native environment designed for harmonizing, governing, and analyzing diverse biomedical data. Researchers can access the datasets through the platform's Exchange and Workbench components. - The initial dataset made available focuses on breast cancer, providing a longitudinal cohort of 558 patients who underwent digital breast tomosynthesis exams, including 271 cases with biopsy-proven malignant lesions. - To ensure patient privacy and regulatory compliance, Segmed employs a de-identification process that has received an independent HIPAA Expert Determination. This process uses a combination of automated algorithms and human review to remove 18 unique patient identifiers from both image data and associated text. - Verily's Chief Revenue Officer, Bharat Rajagopal, noted that the addition of Segmed's curated multimodal data strengthens the Pre platform for research and discovery. The collaboration aims to lower barriers to entry for investigators in academia, life sciences, and healthcare AI. - This partnership is part of a broader trend of using AI to analyze complex medical imaging data like CT scans and MRIs to improve diagnostic accuracy and speed. However, the field faces challenges including data privacy concerns, the need for high-quality and diverse training data, and integration into existing clinical workflows.
