Machine learning twin network for food safety screening
Innovative machine learning twin network technology that differentiates normal from abnormal food images, enhancing food safety screening. Originally developed for mammogram analysis, it's now adapted for food safety with a TRL of 4.
Overview
The machine learning twin network technology provides an advanced solution for enhancing food safety screening by differentiating normal from abnormal food images. Initially developed for medical imaging, specifically in screening mammograms, this technology leverages a differential image amplifier that identifies slight differences between two images. With proven accuracy in medical applications, it now offers promising potential for transforming food safety protocols by providing more precise and reliable screening outcomes.
Technical specifications
- Twin Network: A machine learning architecture designed to detect subtle differences between paired images, improving identification accuracy.
- Differential Image Amplifier: Enhances the ability to discern slight variations, crucial for detecting food safety issues.
- Proven Accuracy: Achieved a sensitivity of 93% and specificity of 91% in medical imaging, showcasing its effectiveness and reliability.
- Adaptation for Food Safety: The twin network will be trained on food safety ground truth data to identify abnormal food conditions, aiming to surpass current gold standard tests.
Technology readiness level
This technology is currently at TRL 4, having been validated in a laboratory setting for medical imaging. Future validation is planned to adapt this solution for food safety applications, with a study duration of 1 to 2 years.
About University of Connecticut
UConn is a comprehensive public research university and the state’s flagship, with its main campus in Storrs and an academic medical center in Farmington. Industry engages through UConn Tech Park—anchored by the Innovation Partnership Building—with shared labs, pilot‑scale testbeds, and advanced prototyping for collaborators. Integration with UConn Health enables clinical collaborations and translation, while Hartford and Stamford campuses connect companies to talent along the New York–Boston corridor; a statewide extension network supports technology adoption. Research is supported by competitive federal funding from agencies such as NIH, NSF, DOE, USDA, and DOD. A dedicated technology commercialization office and the Technology Incubation Program support IP, licensing, and startup incubation across multiple sites.
Inquire about this partnering listing
Industry applications
delivered to your inbox.