MECOS combines are designed for microscopic analysis of a wide range of biomaterials in medical, research, veterinary, sanitary laboratories of various profiles.
In scanner mode, automatic combine techniques form representative digital images of slides (SDIs) and provide an ergonomic workplace for visual viewing of SDIs on the monitor.
A group of techniques automatically generates a gallery of images of objects of analysis in accordance with the requirements of regulated manual microscopy techniques. Galleries of analysis objects are provided in the "preclassification" mode with possible adjustment by the user of the composition of the galleries.
AMSM© technology of MECOS combines maximizes the quality and speed of production of SDIs, performing automatic adaptation to the properties of the slide. Preclassification galleries are formed by automatic combine analyzers - trained neural networks.
The collective cloud resource http://mecosvirt.ru allows laboratories to receive remote telemedicine consultations of the SDIs and the results of preclassification. The resource http://mecosvirt.ru also provides access to collective activities of the professional community to serve the flow of requests for consultations, advanced training, quality control, creation of archives of diagnostic cases, and conferences.
During operation, the combine generates statistics on the accuracy of preclassification. Combines also receive up-to-date statistics on the accuracy of preclassification for the team of operated combines, coming from the cloud monitor of the manufacturer MecosMonitor. Based on accuracy statistics, the administration, taking into account the current conditions and tasks of the laboratory, decides on the need to control the entire SDI, control only preclassification galleries or possibly apply an automatic technique without user control.
MECOS combines are active self-learning systems that use the period of operation to increase their own competence and accuracy of preclassification.
Users, performing regular control of the results of preclassification of slide flows of laboratories, send the SDIs and the results of verified control of preclassification galleries through MecosMonitor to the database of combine artificial intelligence LOSCAM©. LOSCAM periodically conducts retraining of neural networks of analyzers using an ever-increasing database of control results as a training sample. With increasing preclassification accuracy, LOSCAM upgrades the analyzers of operating combines through MecosMonitor.
The use of AMSM, MecosMonitor and LOSCAM technologies for the first time creates conditions for the formation of full-scale representative training and testing samples of digital slides that are not available before the stage of mass operation. The professional community of users is involved in the process of developing and making decisions about options for controlling automatic methods. It is possible to develop / modernize automatic methods that are at any level of responsibility for the result.