UTECH gets U.S. patent for AI-assisted landmark detection in clinical video

UTECH Products Inc. and EndoSoft® said the U.S. Patent and Trademark Office issued a patent for multithreaded video processing that uses cloud-based machine learning to detect landmarks and lesions during clinical procedures. The technology is designed to keep live procedure video responsive while adding AI-assisted annotations in real time.

Why it matters: - The patent adds intellectual property around AI-assisted clinical documentation, procedural video analysis, cloud computing, and lesion detection. - The technology is aimed at helping clinicians identify anatomy and lesions during a procedure, which could improve real-time workflow and documentation.

What happened: - UTECH Products Inc. and EndoSoft® said the U.S. Patent and Trademark Office issued U.S. Patent No. 12,608,803 B2 on July 15, 2026. - The patent is titled “Multithreading Video Processing with Cloud Services to Detect and Analyze Landmarks and Objects During a Clinical Procedure.” - The patent covers systems and methods for processing live video streams during clinical procedures using multithreaded local processing and cloud-based machine learning. - The technology is designed to identify anatomical landmarks and lesions in endoscopy video and return visual annotations to the clinical display while the procedure is underway.

The details: - The system separates live video display from cloud-based analysis. - A primary processing thread keeps video displaying in real time. - A secondary processing thread sends selected frames to cloud services for AI analysis. - The architecture is designed to preserve responsive viewing while AI-assisted annotations run in the background. - The invention builds on UTECH’s earlier patented real-time, cloud-assisted video processing technology. - The patent focuses on clinical landmark identification in gastrointestinal procedures. - Landmarks named in the filing include the cecum, terminal ilium, ileocecal valve, hepatic flexure, splenic flexure, sigmoid colon, rectum, Z-line, pylorus, and the second and third part of the duodenum. - The patent also describes machine learning models for lesion detection, lesion sizing, and diagnostic analysis.

Between the lines: - The filing shows UTECH and EndoSoft are pushing to make AI work alongside, not after, the procedure. - The multithreading approach suggests a design priority on keeping the live feed uninterrupted while offloading analysis to the cloud. - The patent language points to a broader strategy around procedural intelligence, not just image recognition. - Zohair Jaffri, V.P. of Product Development at EndoSoft LLC, said the patent reflects continued investment in AI-assisted clinical documentation and procedural intelligence. - Jaffri said the goal is to help clinicians capture meaningful procedural information as care is being delivered.

What's next: - The patent may support further product development across AI-assisted video analysis and clinical documentation tools. - UTECH’s patent portfolio now has broader coverage across artificial intelligence, procedural video analysis, cloud-integrated computing, and clinical documentation technology. - EndoSoft® continues to market AI applications through the Argus® Ecosystem and other endoscopy software tools. - EndoSoft® also offers EHR, video and image middleware, DICOM/PACS, nursing, infusion, patient engagement, scheduling, and inventory management software.

The bottom line: - UTECH and EndoSoft are betting that real-time AI can make endoscopy video more useful during the procedure, not just after it ends.