Artificial intelligence wound evaluation area measuring and calculating method and device

Abstract

The invention relates to an artificial intelligence wound evaluation area measurement and calculation method and device. The method comprises the following steps: S1, constructing a feature extractionnetwork and a convolutional neural network for wound evaluation area measurement and calculation; S2, shooting a wound picture and labeling the wound picture by using a deep learning labeling tool; S3, generating a wound picture training set based on the labeled wound pictures; s4, inputting the wound picture training set into a feature extraction network and a convolutional neural network for image data training, and generating a training model; and S5, based on the trained feature extraction network and the convolutional neural network, performing wound evaluation area measurement and calculation on the shot wound picture to obtain wound information. The method has the advantages that through an artificial intelligence algorithm, human resources are saved, and efficiency and accuracy are improved.

Description

【Technical Field】 [0001] The invention relates to the field of wound image recognition, in particular to a method and device for measuring and calculating the area of ​​artificial intelligence wound assessment. 【Background technique】 [0002] At present, clinical medical staff use a straightedge to measure the wound size, record the length and width of the wound, and calculate its area. Disadvantages of this method: (1) The measuring tool needs to touch the wound, which may cause cross-infection; (2) The smallest unit of ruler measurement is centimeters. In the later period of judging the progress of the wound, the centimeter measurement is not fine enough; (3) For wounds with irregular edges, it is difficult to measure length and width. [0003] There is currently a "wound measurement" APP ("wound measurement record"), whose functions include: taking pictures and archiving of wounds, calculating area, and inputting basic patient data. Disadvantages of this method: (1) The area c...

AI Technical Summary

Problems solved by technology

Disadvantages of this method: (1) The measurement tool needs to be in contact with the wound, and there is a risk of cross-infection; (2) The smallest unit of ruler measurement is centimeters, and the centimeter measurement is not fine enough in the later stage to judge the progress of the wound; (3) For wounds with irregular wound edges, it is more difficult to measure the length and width

Disadvantages of this method: (1) The calculation of the area is based on manual calculation of the length and width of the wound; (2) The wound assessment only stays in the calculation of the wound area; (3) The clinical use of medical staff is not much

Disadvantages of this device: only the accurate measurement of the wound area is achieved, but in other aspects of wound assessment, such as the patient's whole body assessment, local assessment, etc., the assessment cannot be achieved

After searching, it was found that the wound assessment equipment and system based on artificial intelligence algorithm had not been reported

[0006] The multi-receptive pyramid network PSPNet (Pyramid Scene Parsing Network) is a multi-scale estimation network that integrates context information improved on the Feature Pyramid Network (FPN), and introduces more context information. When the segmentation layer has more When global information is used, the probability of mis-segmentation will be lower; this idea is currently applied in many image fields, and there are many ways to introduce more context information, such as: 1. Increase the receptive field of the separation layer, which The first method is the most intuitive, the wider the field of view, the more things you can see; there are many ways to increase the receptive field, such as dilated convolution, which is successfully applied to the deeplab algorithm; the global mean Pooling operation, PSPNet's global mean pooling operation is also a way to increase the receptive field; 2. The fusion of deep features and shallow features increases the semantic information of shallow features, so that there is enough for shallow segmentation. Context information, as well as detailed information of the target, this approach has existed in FCN as early as, but there is a certain room for optimization including the selection of fusion strategy and segmentation layer

Images