Urine erythrocyte lesion identification and statistics method and system based on improved Faster R-cnn

Abstract

The invention belongs to the field of artificial intelligence auxiliary diagnosis of medical pathology, and discloses an urine erythrocyte lesion identification and statistics method and a system based on improved Faster R-cnn, and the method comprises the following steps: S1, collecting a plurality of sample urine erythrocyte images of a nephrotic patient; S2, labeling the types of the urine redcells in the sample urine red cell image, generating a sample by using an adversarial generative network, and performing data enhancement preprocessing after the sample is generated to serve as training data; S3, inputting the training data into an improved Faster R-cnn neural network for training; wherein a 1 * 1 convolution layer is added into a basic network of the improved Faster R-cnn neuralnetwork, and a BN layer in the network is replaced by a GN layer; and S4, inputting the to-be-identified urine erythrocyte image into the network to obtain the classification of all urine erythrocytesin the image, and carrying out statistics. According to the method, a quicker and more accurate erythrocyte classification malformation rate statistical result can be obtained.

Description

technical field [0001] The invention belongs to the field of artificial intelligence-assisted diagnosis of medical pathology, and in particular relates to a method and system for identifying and counting urinary red blood cell lesions based on an improved Faster R-cnn. Background technique [0002] In recent years, artificial intelligence technology represented by deep learning has developed significantly, and its applications in the fields of pathology and medical imaging diagnosis have also attracted more and more attention. Deep learning utilizes multi-layer artificial neural networks, which combine low-level features to form more abstract high-level features to discover distributed feature representations of data. Deep learning can better reflect the essential characteristics of data, and its effect is significantly better than artificially designed features such as expert systems. At present, deep learning has been extensively studied in the fields of medical imaging s...

AI Technical Summary

Problems solved by technology

However, the traditional image processing and feature extraction methods require human participation in the extraction of features, which has great subjective characteristics. Only based on the above method to identify and count urine red blood cells will cause missed detection and false detection.

Images