23rd Patent Award IVD Product Patent Analysis (II) | Aiwei Technology's Material Component Classification Method Based on Deep Neural Network Image Recognition Algorithm

Publication Time:

2022-08-13 00:00

Source:

Recently, the State Intellectual Property Office of China announced the winners of the 23rd China Patent Award, and Aowei Technology's invention patent—"A method and system for classifying formed elements in samples"—won the China Patent Excellence Award.

In the past, the classification of formed elements mainly relied on manual microscopic examination. Typically, the cells to be tested were first fluorescently labeled or stained, then images were acquired, and the formed elements were classified based on the image characteristics of the cells. This method is cumbersome and complex, unsuitable for testing large numbers of samples, susceptible to subjective factors, and difficult to accurately achieve fine classification of cells.

To improve the efficiency and accuracy of formed element classification, Aowei Technology has invented a method and system for classifying formed elements in samples to accurately classify formed elements. For the labeled samples to be tested, magnification is performed under a microscope, images are automatically acquired, and the formed elements in the images are segmented to obtain the first segmented image. Digital processing is performed using a deep neural network-based image recognition algorithm, extracting hundreds of morphological feature parameters such as size, shape, chromaticity, and texture for morphological analysis, achieving preliminary classification and recognition. The preliminarily classified formed elements undergo secondary segmentation and processing to obtain various fine classifications. This method achieves automatic and accurate identification and classification of formed elements, requiring no personnel attendance throughout the process. It completely solves the drawbacks of slow speed and low efficiency of previous manual microscopic examination, improving the convenience and accuracy of clinical applications.

Currently, this technology has been applied to the AVE-76 series urine formed element analyzers and AVE-26 series blood cell morphology analyzers, achieving automation, intelligence, and standardization of morphological microscopic examination. It performs fine classification of various formed elements, including red blood cell classification (normal red blood cells, large red blood cells, small red blood cells, thorny red blood cells, crenated red blood cells, ringed red blood cells, shadow red blood cells), white blood cell classification (pus cells, phagocytes), epithelial cell classification (squamous epithelial cells, superficial urothelial cells, intermediate urothelial cells, basal urothelial cells), crystal classification (calcium oxalate crystals, uric acid crystals, ammonium magnesium phosphate crystals, amorphous salt crystals), and cast classification (hyaline casts, red blood cell casts, white blood cell casts, granular casts, waxy casts, broad casts), etc., providing more valuable diagnostic indicators for clinical practice and improving the efficiency of morphological testing, freeing up manpower.