Pupil Detection in Ophthalmic Laser Surgery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing pupil detection methods for ophthalmic laser surgery are unsatisfactory due to low accuracy, particularly in images captured before and after corneal applanation, where changes in eye shape and distortion occur.
Innovation Solution
A method involving the use of infrared images for pupil detection before applanation, where pixel intensity values are clustered into four clusters to determine the pupil boundary and center, and a method using color images in the HSV color space after applanation, with filtering and edge detection techniques to accurately detect the pupil.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If patient interface applanates the cornea to stabilize the eye for laser treatment, then the eye becomes stable for treatment, but the pupil becomes distorted and harder to detect accurately
Solution Approach 1:
The patent applies different detection parameters and algorithms for pre-applanation and post-applanation images. For pre-applanation images, it uses gradient computation with specific thresholding, while for post-applanation images, it uses correlation-based methods that account for the distorted pupil shape caused by applanation. This adaptive parameter change resolves the contradiction by optimizing detection accuracy for each specific eye state.
Solution Approach 2:
The system dynamically selects different detection algorithms and parameters based on whether the eye is in pre-applanation or post-applanation state. The pupil detection method transitions from gradient-based edge detection to correlation-based shape matching, adapting to the changing pupil morphology caused by applanation while maintaining accurate detection throughout the surgical process.
2Productivity
If gradient computation with heuristic thresholding is used for pupil detection, then the detection process is computationally efficient, but the detection accuracy is low
Solution Approach 1:
The patent segments the pupil detection process into distinct phases: pre-applanation detection using gradient methods for speed, and post-applanation detection using correlation methods for accuracy. By dividing the detection task according to the eye's state, it achieves both computational efficiency in the first phase and high accuracy in the second phase, resolving the contradiction between speed and precision.
3Device complexity
If fixed-radius circular mask convolution is applied for pupil detection, then the processing is simple and fast, but the accuracy is insufficient for distorted pupils after applanation
Solution Approach 1:
The system dynamically adjusts the detection approach based on the eye's state. For pre-applanation images with normal pupil shape, it uses simple fixed-radius circular mask convolution. For post-applanation images with distorted pupils, it transitions to correlation-based methods that can handle arbitrary shapes. This dynamic adaptation maintains simplicity when possible while achieving high accuracy when needed.
Data Source
AI summary
Two pupil detection methods implemented in an ophthalmic laser surgery system for detecting the pupil of the patient's eye both before and after the eye is applanated by a patient interface device. In the first method, an infrared image is captured before the eye is coupled to the patient interface. After excluding certain image artifacts, the pixel intensity is clustered into four clusters, and pixels belonging to the cluster with the lowest intensity are deemed pupil pixels, and pupil boundary and pupil center are determined accordingly. In the second method, a color image is captured after the eye is coupled to the patient interface. The color image is converted to the HSV color space, and only the hue channel is used for pupil detection. The hue channel image is filtered by median filtering and edge-preserving filtering (e.g. guided image filtering), then binarized, before pupil detection is performed.