Embryo Image Selection for Morphokinetic Parameter Annotation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The process of establishing values for parameters related to embryo development from time-lapse images in IVF is time-intensive and requires skilled clinical assessment, with automated methods lacking the reliability of human expertise.
Innovation Solution
A method that automatically selects images from a series of time-lapse embryo images based on predefined parameters, allowing users to quickly establish values for morphokinetic parameters and other characteristics through user input, with the option to scroll through images and provide clinical judgments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual annotation of embryo development parameters is performed by skilled embryologists, then reliability of clinical assessment is maintained, but time consumption increases significantly
Solution Approach 1:
The patent introduces an automated image analysis system that acts as an intermediary between the time-lapse imaging system and the embryologist. This system automatically processes images, detects developmental events, and pre-annotates parameters, reducing the time burden on embryologists while maintaining assessment reliability through their final verification.
Solution Approach 2:
The system enables self-service annotation by automatically analyzing images and generating parameter values without requiring continuous embryologist intervention. The automated detection of cell divisions, timing events, and morphological changes allows the system to annotate parameters independently, with embryologists only needing to review and confirm results.
2Productivity
If automated methods are used to establish parameter values, then time efficiency is improved, but reliability of assessment deteriorates due to lack of skilled clinical judgment
Solution Approach 1:
The system performs preliminary automated annotation of embryo parameters before the embryologist conducts final assessment. By pre-processing images and generating initial parameter values automatically, the system prepares data in advance, allowing embryologists to focus their expertise on verification and complex judgment cases rather than manual measurement of all parameters.
Solution Approach 2:
The automated system performs partial annotation by detecting only certain developmental events and parameters automatically, while leaving other parameters requiring clinical judgment to be annotated by embryologists. This hybrid approach combines the speed of automation with the reliability of human expertise for critical assessments.
3Measurement precision
If users manually navigate through entire series of time-lapse images to establish parameter values, then comprehensive assessment is achieved, but productivity decreases
Solution Approach 1:
The system automatically pre-processes the entire series of time-lapse images before the embryologist begins assessment. It detects developmental events, timestamps them, and organizes images by event type in advance, so that when the embryologist reviews the data, all relevant information is already prepared and easily accessible, eliminating the need to manually search through hundreds of images.
Solution Approach 2:
The patent segments the large series of time-lapse images into smaller, organized groups based on detected developmental events. Instead of presenting the embryologist with a continuous sequence of hundreds of images, the system divides them into event-specific segments (e.g., all images around 2-cell division, all images around 4-cell division), making comprehensive assessment more manageable and efficient.
Data Source
AI summary
Apparatus and methods for helping a user establish values (e.g. timings) for a plurality of parameters of interest (e.g. cell divisions) relating to the development of an embryo from a series of images of the embryo are described. For each parameter of interest an image is selected for display to a user seeking to establish a value for the parameter of interest. For example, the selected image may be an image predicted to be an image reflecting the value for the parameter of interest. For example, the selected image may be based on a calculated timing for a particular developmental event. The timing may be calculated from a numerical analysis of the images or maybe predetermined. If the user is unable to determine a value for the parameter of interest from the selected image, the user may scroll through neighboring images until the user can determine a value for the parameter of interest. A value for the proud of interest may then be established in response to user input, for example a user providing an indication that a timing associated with a currently displayed image from the series of images should be taken to be the value of the parameter of interest. The different parameters of interest may be established in an iterative manner in which an initial image for display to a user is selected for each parameter of interest based on the parameter of interest.