Automated computer-based screening process for strabismus.
Patent Information
- Application Number
- TH2303003405
- Authority / Receiving Office
- TH · TH
- Patent Type
- Utility models
- Current Assignee / Owner
- Filing Date
- 2023-11-20
- Publication Date
- 2026-10-01
- Estimated Expiration
- 2029-11-19
Smart Images
Figure 00000002_0000
Abstract
Claims
OCR09WP01 / 09 / 25691. The automated computer-based strabismus screening procedure involves the following steps: a. Receiving facial images of the test subject under natural light conditions and at normal computer usage distance without corneal reflection. b. Image extraction and verification: The processor extracts the facial images obtained from step a, and then extracts the left DL(1) and right DR(4) peripheral areas of the eye and performs a visual posture verification. c. Regression calculation: The verified peripheral area images from step b are fed into an artificial intelligence model to identify the central position of the pupil. d. Diagnosis of strabismus: The processor calculates the geometric deviation difference of the coordinates obtained from step c and compares it to the standard reference normal values to determine the specific characteristics of the strabismus as described in step c.The following are parallel calculations of pupil coordinates in two directions to check for cross-referencing: • Calculation of the left pupil coordinate (6) from the DL(1) image and the right pupil coordinate (3) from the DR(4) image. • Calculation of the right pupil coordinate (2) from the DL(1) image and the left pupil coordinate (5) from the DR(4) image. In step 1, the processor finds the difference between the coordinate positions obtained from the analysis of both directions for each eye to interpret the strabismus condition.
2. An automated computer-based strabismus screening process according to claim 1, in step 2, the image data of the left and right eye periphery areas must meet the geometric proportions check criteria with a limited area difference within the ratio range of 0.9 to 1.1, and the process automatically discards image data in frames with ratios outside this criterion to prevent errors from head turning or facial tilting.
3. An automated computer-based strabismus screening process according to claim 1, in step 3, the artificial intelligence model in step 4.
4. An automated computer-based neural network (CNR) screening process based on claim 1, where in step 1, the reference normal distance threshold is set to a specific constant value of 6.21 for the left eye coordinate difference and 9.24 for the right eye coordinate difference.
5. An automated computer-based strabismus screening process based on claim 1, where the process from step 1 to step 2 is continuously iterative in a time series loop to analyze statistical trends for screening latent strabismus in subjects experiencing eye strain from prolonged visual fatigue.