Fibrobronchoscopy intubation aid decision making method based on deep learning

Abstract

The invention discloses a fibrobronchoscopy intubation aid decision making method based on deep learning. The method comprises the steps that an intubation video is collected and divided into image frames; a decision instruction set is formulated; operation instructions are taken as category labels of image decision to label the image frames frame by frame; original images are processed to make adata set; a training set and a verification set are extracted; feature extraction is performed on a training sample, and high-level feature maps are obtained step by step; results after the feature maps are conversed are sent into a Naive Bayes classifier and a softmax activation function for calculation; the training sample is input, a loss function loss is set, and separate training is performedon an operation instruction decision model and a Naive Bayes classifier model separately; the training step is repeated to cross-validate a network model. According to the method, an end-to-end method is adopted, and a decision result is directly obtained by inputting the images, so that the algorithm cost is greatly reduced, the decision speed is high, and the real-time performance is very high.

Description

technical field [0001] The invention belongs to the field of deep learning and image processing, and in particular relates to an auxiliary decision-making method for fiberoptic bronchoscopic intubation based on deep learning. Background technique [0002] Before some operations that require general anesthesia, in order to ensure the safety of the patient and avoid suffocation due to loss of consciousness or tracheal obstruction, the patient needs to be intubated to provide oxygen supply to the patient's lungs. And the most widely used method at present is to utilize the method for carrying out tracheal intubation to the patient by the method for overcoating the trachea of ​​bronchofiberoscope. The bronchoscope is a tube-shaped medical device with a lens on its head, which can be bent up and down perpendicular to the field of view plane where the front lens is located through the push rod on the rear handle; at the same time, by twisting the doctor's wrist The bronchoscope l...

AI Technical Summary

Problems solved by technology

At present, there is a shortage of anesthesiologists in my country, a serious imbalance in the ratio of anesthesiologists to patients, and the status quo of overworked anesthesiologists.

On the other hand, due to the long working hours and the need for rich work experience in intubation work, it is inevitable that the anesthesiologist will make misoperations during the tracheal intubation process, or the intubation is too long, resulting in hypoventilation of the patient, and an assistant is urgently needed. Method helps anesthesiologists perform intubation

[0004] At present, all intubation operations are performed manually by anesthesiologists, and such a complicated operation requires doctors to have rich medical knowledge and clinical experience

Due to the differences in individual conditions of patients, the tracheal environment in the body is also different, and certain physiological or pathological conditions will make it difficult to perform artificial intubation

To sum up, there are two difficulties in the existing manual intubation, one is the recognition of the picture taken by the bronchoscopic lens and the judgment of the moving direction of the lens; the other is the cumbersome lens control operation based on the picture

Images