Protein Recognition using Optical Band Extraction
The technology is tuned specifically to produce reliable images for the enhancement of clinical interpretability.
Sidebar
Overview:
Researchers at the University of Louisville have developed a process using machine learning and artificial intelligence techniques for automating the interpretation of Immunofixation Electrophoresis (IFE) gels. IFE is commonly used to detect and characterize multiple myeloma, and its interpretation requires significant time, effort and expertise from medical staff. This UofL technology is designed to analyze IFE gel scans to identify the presence of diagnostic proteins and provide a more efficient means of triage in a manner that reduces inter-operator variability.
Highlights:
- PROBE represents a step in the direction of diagnostic AI automation in the clinical laboratory.
- Potential to automatically identify specific diagnostic proteins for the interpretability and transparency of model outputs.
- Trained with over 4,000 archived serum IFE gel scan images and achieved an AUROC of 1.00 on validation data and 0.99 on test data for predicting positive vs. negative serum IFE samples.
Benefits:
- The only known automated gel IFE interpretation software with confidence regularization and interpretable outputs.
- PROBE is poised to be integrated with routine clinical laboratory workflows to aid in triaging patients with potential plasma cell disorders.
- The technology is tuned specifically to produce reliable images for the enhancement of clinical interpretability.
Applications:
Market applications include but are not limited to electrophoresis IFE research; electrophoresis applications; diagnostic software; monoclonal gammopathy screening; artificial intelligence laboratory automation.
IP Status:
Filed (available upon request for update about the current position)
Inventors:
- Mark Linder