Visual Field Correction Using 3D Facial Contour Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing visual field tests struggle to accurately distinguish between peripheral visual field defects caused by ocular pathology and those caused by facial structures, leading to incorrect diagnoses and incomplete assessment of diseases like glaucoma, particularly in early stages where central and peripheral visual field losses may not be correlated.
Innovation Solution
A system and method using a convolutional neural network (CNN) to generate a 3D reconstruction of a subject's face from a 2D image, predict visual field defects from facial contours, and correct the acquired visual field by subtracting these defects to isolate those caused by ocular pathology, while optimizing head turn angles for comprehensive visual field mapping.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If 60-4 visual field testing pattern is used to assess far peripheral vision, then the coverage of peripheral visual field is improved, but the reliability of results deteriorates due to facial structure interference
Solution Approach 1:
The patent segments the visual field defects into two distinct components: those caused by facial structures and those caused by ocular pathology. By separately identifying and subtracting the facial structure component from the total measured defects, the system isolates the pathological component with higher reliability while maintaining comprehensive peripheral coverage.
Solution Approach 2:
The patent extracts the facial structure component from the total visual field measurement by using 3D facial reconstruction and ray-tracing algorithms to calculate expected defects from facial contours. This extracted component is then subtracted from the measured visual field to reveal the underlying ocular pathology with improved diagnostic accuracy.
2Reliability
If 24-2 or 30-2 field patterns are used for central vision assessment, then the reliability of central field measurement is improved, but the area of visual field coverage deteriorates
Solution Approach 1:
The patent merges multiple visual field testing patterns (24-2, 30-2, and 60-4) into a comprehensive assessment protocol. By combining the reliable central field measurement of 24-2/30-2 patterns with the extended peripheral coverage of 60-4 pattern, and then applying facial structure correction, the system achieves both reliable central measurement and comprehensive peripheral coverage simultaneously.
3Ease of operation
If facial structure impact is not corrected, then the simplicity of visual field testing is maintained, but the measurement precision deteriorates due to confounding factors
Solution Approach 1:
The patent performs preliminary 3D facial reconstruction and ray-tracing calculations before conducting the visual field test to predict where facial structures will cause defects. This pre-computed facial structure map is then subtracted from the actual visual field results, automatically correcting for facial interference without requiring additional manual adjustment or complex testing procedures during the actual exam.
Data Source
Figure 1
Figure 2
Figure 3A~3B
AI summary
Described herein is a system and method for correcting a visual field of a subject using a predicted visual field that includes visual field defects from facial contours of the subject. In an embodiment, the predicted visual field may be determined using a three-dimensional (3D) reconstruction of the face of the subject which may be generated from a two-dimensional (2D) image of the subject using a convolution neural network. A system and method for optimizing a head turn angle of a subject for determining a visual field of a subject is also described herein. An optimal head turn angle may be determined using the 3D reconstruction of the face of the subject. In some embodiments, after the visual field is maximized with positioning of the head, residual facial contour induced defects can be predicted and a final corrected field can be generated.