Pneumonia picture calibrating device and method capable of self-adaptively adjusting size of receptive field

Abstract

The invention relates to medical instruments, the field of deep learning convolutional neural networks and the field of target detection and positioning. In order to improve chest X-ray film diagnosisefficiency, the pneumonia picture verification device capable of self-adaptively adjusting the size of the receptive field comprises an X-ray machine and a computer, a picture shot by the X-ray machine is input into the computer, and the computer comprises a feature extraction network processing module, a feature pyramid module, a classification sub-branch module and a regression sub-branch module; the ResNet50 and the ResNet101 are respectively combined with the selective kernel convolution to form an SK-ResNet50 network and an SK-ResNet101 network, and the SK-ResNet50 network and the SK-ResNet101 network are used as feature extraction networks; the extracted features are input into a feature pyramid module to be processed, and a feature pyramid outputs a feature map; the classificationsub-branch module outputs a detection score of the prediction box; and the regression sub-branch module outputs the position of a prediction box, wherein the prediction box is a predicted pneumonia lesion area. The method is mainly applied to design and manufacturing occasions.

Description

technical field [0001] The invention relates to the fields of medical equipment, deep learning convolutional neural network, and target detection and positioning. The dynamic selection unit is combined with the target detection network for improvement, so that the neuron can adapt itself according to the multi-scale input information, that is, the size of the target. Adjust the size of the receptive field to more accurately realize the task of pneumonia detection and localization in chest X-ray images. Specifically, it relates to a pneumonia detection device and a positioning method that can adaptively adjust the size of the receptive field. Background technique [0002] Pneumonia is a serious lung disease, which is the inflammation of the alveoli caused by bacteria, viruses, fungi, etc. If it is not diagnosed and treated in time, it will deteriorate rapidly and lead to heart failure, empyema, lung abscess, myocarditis or toxic encephalitis, etc. other illnesses. Globally,...

AI Technical Summary

Problems solved by technology

Wu et al. designed a device for predicting X-ray pneumonia results based on convolutional neural networks, in which the classification network uses ResNet50, and the detection model is the faster proposed regional convolutional neural network (Faster R-CNN), but the detection model used is Two-stage detection model, so the model complexity is high and the detection speed is slow

Amit et al. first input the X-ray image into the calibration candidate region (ROI Align) classifier, then use the Faster R-CNN model to segment and predict the prediction frame, and adjust the threshold during the training process to improve the result, but the result accuracy is not high.

Images