A method for automated appearance comparison of new medications in hospitals

Abstract

This invention proposes an automated comparison method for identifying the appearance of newly imported drugs and existing drugs in hospitals. Through high-resolution image capture, feature point detection, dynamically updated databases, and OCR (Optical Character Recognition) and other feature comparison methods, it can quickly calculate the similarity of shape, size, and color and output visualized results, effectively reducing the risk of packaging confusion and improving medication safety and efficiency. It can assist managers in preventing similar packaging and avoiding clinical medication errors.

Claims

1. A method for comparing the appearance of new drugs used in hospitals, employing image recognition technology to determine the similarity between a newly introduced drug and a plurality of existing drugs, comprising: Step 1: capturing at least one appearance image of the newly introduced drug placed at a positioning angle using at least one high-resolution camera; Step 2: preprocessing the appearance image; Step 3: obtaining reference image data of a plurality of existing drugs from a drug database; Step 4: extracting a feature descriptor from the feature region of the newly introduced drug and the reference image data using a feature point detection algorithm; Step 5: comparing the feature descriptor of the newly introduced drug with the reference image data. The existing drug's characteristic descriptor is compared to calculate a similarity score; Step 6: Text information on the appearance of the new drug is extracted using optical character recognition technology, and weight is assigned to this text information when calculating the similarity score; if the text keywords match perfectly, the similarity score is increased; if the brand name is different or the dosage label is inconsistent, the overall score is decreased; then the similarity score is expressed as an overall percentage, and the similarity score and the visualized comparison results are output, and the reference drugs that are closest to the new drug are presented in order of similarity for use by a hospital's drug management decision-makers.

2. The method as described in claim 1 further includes a step of dynamically updating the drug database, whereby when a new drug is added or an existing drug is removed from the market or its packaging is redesigned, the corresponding reference images and feature descriptors are promptly deleted or replaced, so that the database can remain up-to-date and ensure the timeliness and accuracy of subsequent comparison results.

3. The method as described in claim 1, wherein the measurement of shape and size includes at least one of the length, width, thickness or curvature of the new drug, and the geometric parameters are calculated after obtaining the edge contour of the packaging or tablet through edge detection, and used as a weighting factor for similarity assessment.

4. The method as described in claim 1, wherein in the preprocessing of step 2 or the similarity assessment of step 5, the method further includes quantitative analysis of the color distribution on the packaging or bare tablet surface of the new drug and the reference drug, and incorporates the color distribution similarity into the overall assessment to improve the accuracy of the comparison results.

5. The method as described in claim 1, wherein in step 6, a visual comparison result is generated, the new drug and the existing drugs in the reference image data are displayed side by side, and similarities are connected by lines.

6. The method as described in claim 1, wherein in step 6, text information on the appearance of the new drug is extracted using optical character recognition technology, and the text information is included in the similarity assessment.

7. The method as described in claim 5, wherein in step 6, a selection is added to enhance the presentation.

8. The method described in claim 6, wherein the keyword comparison is performed on brand name, drug name, main English letters or dosage numbers, and weights are assigned according to pre-set keyword rules; when the brand is the same and the dosage is consistent, the similarity score is increased, and when the text differences are large, the similarity score is decreased.

Images