Microscope Object Tracking via Image Segmentation and Movement Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current microscopy techniques face challenges in accurately tracking target objects through different section surfaces of a specimen in 3D renderings and successive images over time, due to limited field of view and noise in images, making it difficult to distinguish target objects from other objects and noise.
Innovation Solution
A method involving image segmentation techniques, such as using artificial neural networks, to identify and track target objects by determining similarity scores and movement vectors, allowing the microscope's field of view to be adjusted to maintain the target object within the field of view across section surfaces and time intervals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image segmentation is applied to identify target objects in noisy microscopy images, then tracking accuracy is improved, but processing time and computational complexity increase
Solution Approach 1:
The patent applies image segmentation techniques (such as artificial neural networks) to pre-process and identify target objects in noisy microscopy images before tracking. This preliminary action filters out non-relevant objects and noise, creating a cleaner dataset for subsequent tracking operations, thereby improving tracking accuracy despite the additional processing time required for segmentation
Solution Approach 2:
The patent segments the image processing task into distinct stages: first applying image segmentation to identify and separate target objects from background noise and other objects, then using the segmented results to guide tracking. This segmentation of the overall process allows for more precise target identification while managing computational complexity through staged processing
2Stability of the object's composition
If the microscope field of view is adjusted to track target objects across section surfaces, then object continuity is maintained, but the complexity of coordinate transformation and image alignment increases
Solution Approach 1:
The patent employs feedback mechanisms where the identified position of target objects in segmented images is used to adjust and update the field of view positioning for subsequent images. This feedback loop continuously refines the tracking accuracy across section surfaces, maintaining object continuity while systematically managing the complexity of coordinate transformations through iterative correction
Solution Approach 2:
The patent extends tracking from 2D images to 3D space by incorporating section surface information and applying coordinate transformations across multiple dimensions. This dimensional extension allows maintenance of object continuity through the depth of the specimen while systematically handling the increased complexity through structured spatial transformations
3Measurement precision
If image segmentation techniques are used to filter out noise and non-relevant objects, then signal-to-noise ratio is improved, but computational resources and processing time are consumed
Solution Approach 1:
The patent applies image segmentation techniques selectively to identify and process only the regions containing target objects, rather than processing the entire image uniformly. This partial action approach improves the signal-to-noise ratio by focusing computational resources on relevant areas while reducing overall processing requirements compared to full-image segmentation
Data Source
Figure 1A
Figure 1B
Figure 1C
AI summary
Object tracking using image segmentation is disclosed. A captured image of a specimen is obtained. A segmented image is generated based on the captured image. The segmented image indicates segments corresponding to objects of interest. One or more target objects are identified from the objects of interest in the segmented image. Objects of interest most similar, in position and/or shape, to target objects shown in a previous image may be identified. Alternatively, objects of interest that are associated with connecting vectors most similar to the connecting vectors that connect the target objects in a previous image may be identified. A movement vector is drawn from a target object position in the previous image to a target object position in the segmented image. A field of view of the microscope is moved, with respect to the specimen, according to the movement vector to capture another image of the specimen.