Self-training system of ultrasonic artificial intelligence model

Abstract

The invention provides a self-training system for an ultrasonic artificial intelligence model, and relates to the technical field of model self-training, and the system comprises an ultrasonic scanning device which is internally integrated with a factory artificial intelligence model; the self-training system is used for hospital staff to input a plurality of medical annotation images obtained by annotating a plurality of local medical images scanned by the ultrasonic scanning equipment; storing a factory model network corresponding to the factory artificial intelligence model and a plurality of pre-configured model networks; calling each model network according to a model construction instruction input by a hospital worker to update the factory model network to obtain an updated model network, and training the updated model network according to each medical annotation image to obtain an optimization model; and configuring the optimization model to the corresponding ultrasonic scanning equipment to replace the factory artificial intelligence model. The method has the beneficial effects that hospital workers can conveniently train the artificial intelligence model which meets the personalized requirements of hospitals and is relatively high in precision based on the local medical images.

Description

technical field [0001] The invention relates to the technical field of model self-training, in particular to a self-training system of an ultrasonic artificial intelligence model. Background technique [0002] Ultrasound is the most commonly used examination equipment at the grassroots level. Compared with detection methods such as CT and X-ray, ultrasound diagnosis has the characteristics of real-time scanning, non-invasive, non-radiation, low price, etc., and it can be moved and the scanning method is flexible. With the rapid development of AI technology, AI data analysis and deep learning technologies are also making efforts to promote smarter and more accurate medical equipment. [0003] In the prior art, a certain ultrasound-related AI function is generally used as a function of ultrasound equipment and is directly provided to hospitals for use. It cannot solve the problem of hospitals in different regions and hospitals with different patient distributions. When using ...

AI Technical Summary

Problems solved by technology

[0003] In the existing technology, an ultrasound-related AI function is usually provided directly to hospitals as a function of ultrasound equipment, which cannot solve the problem of accurate recognition when using the same ultrasound AI model in hospitals in different regions and with different patient distribution. high and low

Since the hospital cannot customize or modify the relevant parameters of the AI ​​function, or retrain the AI ​​model or optimize the AI ​​model based on the ultrasound images scanned by the hospital, when the accuracy of model recognition is low, it is necessary to wait for the equipment manufacturer to provide the AI ​​model optimize

However, when equipment manufacturers optimize their AI models, firstly, they cannot obtain ultrasound images from various hospitals, and secondly, even if they obtain ultrasound images scanned by many different hospitals, they cannot effectively train a model that is suitable for all hospitals. At the same time, the hospital waits for a long time for equipment manufacturers to provide optimized AI models, which reduces the efficiency of the use of equipment AI functions, thereby reducing the use value of equipment

Images