Urine Sample Object Counting via Dual-Path Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for analyzing urine samples, such as urinary sediment analysis, face challenges in accurately detecting and counting cells and bacteria due to issues like cell overlaps, stacks, and blurry images, which affect the reliability and reproducibility of the results.
Innovation Solution
A method and system that utilize digitalized images of urine samples processed onto a slide, employing object detection and semantic segmentation models to classify and count objects. The method includes receiving digitalized images, detecting connected components, classifying them, and using object detection models to identify countable components and semantic segmentation models to handle uncountable components, thereby providing accurate object counts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If brightfield optical technology is used to digitalize microscope slides, then the method is cost-effective and simple, but cell overlaps, stacks, and blurry images reduce measurement precision
Solution Approach 1:
The patent applies segmentation by dividing the image analysis into two distinct pathways: connected component segmentation for separating overlapping cells, and pixel-level semantic segmentation for handling uncountable clusters. This segmentation strategy resolves the technical contradiction by maintaining the simplicity of brightfield imaging while improving measurement precision through sophisticated image processing techniques that handle overlaps and blurriness.
Solution Approach 2:
The patent employs parameter changes by transforming the image data through multiple processing stages including connected component analysis, classification into countable/uncountable categories, and application of different counting strategies. These parameter transformations enable accurate object counting despite the limitations of brightfield optical technology, resolving the contradiction between method simplicity and measurement precision.
2Device complexity
If traditional segmentation methods are used on overlapped cell images, then the process is simple, but classification accuracy deteriorates due to cell clusters and stacks
Solution Approach 1:
The patent implements a two-level segmentation approach: first segmenting images into connected components, then classifying each component as countable or uncountable. For uncountable components, pixel-level semantic segmentation is applied. This hierarchical segmentation resolves the contradiction by maintaining process simplicity at the component level while achieving high classification accuracy through specialized handling of complex clusters.
Solution Approach 2:
The patent transitions from object-level segmentation to pixel-level segmentation for uncountable components, adding a dimensional aspect to the analysis. This dimensional change enables accurate classification of cell clusters and stacks that cannot be resolved by traditional single-level segmentation, thereby improving measurement precision without excessive complexity.
3Measurement precision
If object detection models are applied to countable components, then accurate counting is achieved, but processing time increases for large populations
Solution Approach 1:
The patent segments the population into countable and uncountable components, applying different processing strategies to each. Countable components use object detection models for accurate counting, while uncountable components use pixel-level segmentation with area-based estimation. This segmentation resolves the contradiction by applying computationally intensive methods only where necessary, maintaining accuracy while improving overall processing efficiency for large populations.
Solution Approach 2:
The patent applies full object detection modeling only to countable components rather than all components. For uncountable components, a simplified pixel-level approach is used. This partial application of the more intensive method resolves the contradiction between measurement precision and productivity by optimizing the balance between accuracy and processing speed based on component characteristics.
Data Source
AI summary
A method for classifying and counting objects recoverable from a urine sample processed onto a slide. The method includes the following steps: receiving at least one digitalized image of the whole slide; detecting connected components by segmentation of the image of the whole slide; classifying the detected connected components into countable connected components and uncountable connected components using a classifier; for the countable connected components using an object detection model to obtaining the number of objects for each class; for the uncountable components using a semantic segmentation model to obtaining the number of objects for each class; summing up the number of objects for each class obtained from the semantic segmentation model and the object detection model outputting a number of objects for each class.