Detection and diagnosis system of otosclerosis lesions based on small target detection neural network

Abstract

The invention belongs to the technical field of medical image processing, in particular to an otosclerosis lesion detection and diagnosis system based on a small target detection neural network. The system includes a feature extraction backbone network, a target detection and classification network, a noise-robust classification loss function, and a post-processing diagnosis system for multi-layer detection results; the feature extraction backbone network is a multi-level deep convolutional neural network, which is used to extract The feature map of the image; the target detection and classification network includes the above-mentioned feature extraction backbone network and region extraction network, region of interest pooling layer, and classification network to obtain the category of the region; noise robust classification loss function combined with cross entropy loss and mean absolute error The loss is less affected by the wrong labels in the training data; the present invention inputs the 3D temporal bone CT image layered into the network model, and after one forward propagation and post-processing, the lesion detection and diagnosis results can be obtained. The invention can reduce the influence of human factors and improve the efficiency and accuracy of clinical diagnosis.

Description

technical field [0001] The invention belongs to the technical field of medical image processing, and in particular relates to a system for detecting and diagnosing otosclerosis lesions, and more specifically, to a system for detecting and diagnosing otosclerosis lesions based on a small target detection neural network. Background technique [0002] Otosclerosis is a disease in which the labyrinthine dense lamellar bone is focally replaced by spongy new bone rich in cells and blood vessels. Otosclerosis can be divided into stapes type otosclerosis, cochlear type otosclerosis and mixed type otosclerosis according to the location and extent of the lesion. Cochlear otosclerosis is an advanced form of otosclerosis. It is not difficult to diagnose based on typical clinical manifestations and CT manifestations, and the treatment is limited to wearing hearing aids. The earliest and most common location of otosclerosis is in the front of the vestibular window, which will lead to sta...

AI Technical Summary

Problems solved by technology

[0007] However, the existing target detection network is usually unable to detect very small targets such as stapes. To solve this problem, the present invention provides a new noise-robust otosclerosis lesion detection and diagnosis system based on small target detection neural network, which can Fully combine the characteristics of training images, extract rich features, and realize the detection and diagnosis of otosclerotic areas at the same time

Images