Image-based verification of pharmaceutical contents in robotic preparation systems

The use of structured label templates with OCR and machine learning addresses the challenge of non-standardized pharmaceutical labels, improving the accuracy and safety of robotic substance identification in automated preparation systems.

JP2026520491APending Publication Date: 2026-06-23EQUASHIELD MEDICAL
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
EQUASHIELD MEDICAL
Filing Date
2024-05-30
Publication Date
2026-06-23

AI Technical Summary

Technical Problem

Automated pharmaceutical preparation systems face challenges in accurately identifying and verifying pharmaceutical substances due to non-standardized and visually complex label formats, which can lead to misidentification and potential harm to patients.

Method used

A method and system that utilize structured label templates incorporating both text content and graphical appearance, combined with optical character recognition (OCR) and machine learning, to accurately identify and verify pharmaceutical substances by comparing container labels with reference templates, ensuring reliable substance identification.

Benefits of technology

Enhances the accuracy and reliability of pharmaceutical substance identification, reducing the risk of misidentification and ensuring safe preparation by robots, even in the presence of label variations and imperfections.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2026520491000001_ABST
    Figure 2026520491000001_ABST
Patent Text Reader

Abstract

Label image information is used for the automated preparation of pharmaceutical formulations using pharmaceutical substances, as well as for the automated identification and / or verification of the identification of those pharmaceutical substances. In some examples, the text content of the substance container is identified using OCR. This corresponds to an important part of the label information for human inspection. As a cross-check, one or more additional visual aspects of the label, e.g., aspects of the graphical appearance other than the text content itself, are evaluated. Selectively, the risk of misidentification is assessed. Selectively, an incomplete match to a template is considered acceptable because the actual risk of misidentification is low, and / or a close but potentially ambiguous template match is considered not to have a sufficiently low risk, and as a result, mitigation measures are implemented.
Need to check novelty before this filing date? Find Prior Art