Around 1590, Dutch spectacle maker Jansen created the world's first microscope using a simple concave and convex lens, shifting human vision from the macroscopic to the microscopic and opening the door to morphological examination. In 1650, Dutchman Leeuwenhoek's first optical microscope was born, and through his invented 300x magnification optical microscope, he successfully observed human red blood cells for the first time. Since then, the methods and means of observing the microscopic world using microscopes have been used to this day, laying the foundation for the germination and development of cytology, microbiology, and parasitology. For a long time, morphological examination has heavily relied on manual operation, but manual operation is slow, cumbersome, and inconsistent, and the test results depend on the operator's sense of responsibility and testing skills. With the increasing number of specimens today, traditional manual microscopic examination is clearly unable to meet the clinical testing needs.

Figure 1: Development History of Morphological Microscopy

With the comprehensive improvement of computer hardware performance and algorithm computing power, the rapid iteration of machine vision and convolutional neural network technology, and the rapid development of artificial intelligence technology in the medical field, cell morphological microscopy is a core area of artificial intelligence application. Aiwei Technology is the earliest starter and has the richest experience and achievements. Aiwei's original innovative "machine vision technology" and "automation of medical microscope morphological testing" and other key common technologies are internationally leading, ending the history of morphological testing relying solely on manual completion and filling many international and domestic technological gaps. It has successively achieved automated, intelligent, standardized, and standardized detection of formed elements in urine, stool, vaginal discharge, and blood specimens, and is a typical representative of the application of AI in morphological recognition in China and abroad.

Figure 2: How Intelligent is AI Microscopy?

Deep learning algorithms, computing power, and big data are the three major elements of artificial intelligence. Any factor affecting the above three elements may affect the accuracy and stability of cell recognition. Currently, Aiwei Technology has achieved installations of more than 10,000 units in more than 6,000 hospitals, with an annual specimen testing volume exceeding 110 million person-times, providing massive data for AI training. Secondly, in order to improve computing power, Aiwei Technology has built multiple high-performance servers that can simultaneously conduct model training under different parameters, greatly improving the computing speed. Aiwei Technology's 4th generation AI recognition system is based on the deep learning method of convolutional neural networks, combined with intelligent image acquisition technology, and after training the model with big data, it has built deep learning models for the recognition of formed elements in urine, stool, vaginal discharge, and blood specimens. The model uses a convolutional method to gradually mine shallow, intermediate, and deep features in the image, with stronger robustness and generalization ability, ensuring that the detection results are accurate, stable, and reliable. Currently, the 4th generation AI recognition system has been widely used in its launched automated analysis products and solutions for urine, stool, genital secretions, and blood specimens, greatly reducing the workload of operators, improving work efficiency, and gaining widespread recognition.

Figure 3: Aiwei Microscopy Family

AVE-7200 Series Modular Urine Analysis Production Line

Testing Items

Physical Examination: Includes color, turbidity, specific gravity, conductivity, and osmotic pressure.

Dry Chemistry: 14 items, can provide urinary microalbumin/creatinine (ACR).

Morphology: Classification and quantitative counting of white blood cells, epithelial cells, casts, and crystals in urine, analysis of red blood cell morphology, and provision of red blood cell shape, size, and chromaticity curves, as well as size-chromaticity scatter plots to assist in judging the source of hematuria.

AVE-2100 Series Modular Blood Cell Morphology Automated Analysis Production Line

Testing Items

White Blood Cells: Divided into neutrophils, lymphocytes, monocytes, eosinophils, and basophils, calculating the percentage of each type of cell, and pre-classifying abnormal white blood cells and immature granulocytes.

Red Blood Cells: Can be divided into normal red blood cells, macrocytes, microcytes, spherocytes, target cells, stomatocytes, etc., calculating the proportion of each type of cell, describing the morphology of red blood cells, and pre-classifying immature red blood cells.

Platelets: Can be divided into normal platelets, large/giant platelets, and platelet aggregation, and provide alerts when platelet aggregation exceeds the warning value.

Parasites: Can provide alerts for blood parasite (malaria, filariasis, trypanosomiasis, leishmaniasis, etc.) infections.

Reclassification: After manual review, cells that have been abnormally classified or unidentified cells can be reclassified.

AVE-32 Series Genital Secretion Analyzer

Testing Items

Dry Chemistry: Can detect pH, hydrogen peroxide, lactic acid, leukocyte esterase, sialidase, β-glucuronidase, N-acetylglucosaminidase, proline aminopeptidase, coagulase, oxidase, and alkaline phosphatase to assist in judging the type of vaginitis.

Morphology: Includes red blood cells, white blood cells, epithelial cells, clue cells, Trichomonas vaginalis, cocci, bacilli, fungi,

pathogenic microorganisms, automatically reports cleanliness, can be expanded to provide indicators such as bacterial density, diversity, dominant bacteria, Nugent score, Donders score, and can provide dynamic video of Trichomonas detection to improve the detection rate of Trichomonas.

AVE-56 Series Fully Automated Stool Analyzer

Testing Items

Physical Examination: Automatically identifies color and characteristics.

Morphology: Can detect all formed elements in stool specimens under a microscope, including red blood cells, white blood cells, crystals, ova, bacteria, fungi, fat globules, starch granules, etc.

Chemistry and Immunology: Fecal occult blood (supports hemoglobin immunology, transferrin immunology, hemoglobin-immuno double linkage, hemoglobin-transferrin immuno double linkage), rotavirus, adenovirus, Helicobacter pylori, calprotectin, lactoferrin, etc.

From replacing manual microscopic examination to single-machine automation, then to full-process automated assembly lines, and finally upgrading to AI deep learning and big data cloud management, after more than 20 years of clinical practice and accumulation, Aiwei Technology, as a representative in the field of morphological microscopy manufacturing, has demonstrated the development history of morphological microscopy. With the continuous maturity and deep integration of advanced technologies such as artificial intelligence and big data, intelligent morphological testing will achieve intelligent management of laboratory equipment and data, greatly improving testing efficiency, reducing the workload, lowering operating costs, and promoting standardization. From intelligent hardware to cloud computing platforms, from data analysis to machine learning algorithms, everything will help laboratories achieve greater intelligence and efficiency, making clinical laboratories the "most powerful brain" in the diagnostic decision-making process, improving medical efficiency, and safeguarding the health of the people.