CT image detection method and device based on deep learning and control equipment

Abstract

The invention discloses a CT image detection method based on deep learning. The method comprises the steps of preprocessing an original image to obtain a to-be-detected image; establishing a trainingneural network detection model to obtain a neural network CT image detection model; inputting the to-be-detected image into the neural network CT image detection model; Obtaining detection results, wherein the input to-be-detected image comprises cross section direction information. Different from an existing single-input and single-output model training mode, input is changed into five-dimensional data from a one-dimensional data image of a single slice, output is also changed into two-channel mask output from single output, model precision and focus fitting degree of a prediction result areimproved, precision reaches expert level, and clinical application can be carried out. Moreover, in the process, a neural network CT image detection model is established by using big data in advance,during judgment, a CT scanning image of a patient is directly input, the model is used for calculation, the model output response speed is high, and disease delay is not likely to be caused.

Description

technical field [0001] The present invention relates to the field of image detection, in particular to a CT image detection method, device and control equipment based on deep learning. Background technique [0002] Cancer is one of the most common causes of death due to disease in the world, and the number of people who die from cancer remains high every year. Due to the serious threat cancer poses to human beings, people have made many efforts to carry out early diagnosis and treatment. With the development of imaging technology, three-dimensional CT (computerized tomography) has become a routine clinical method for cancer diagnosis. With the emergence of high-resolution CT scanners in recent years, more and more CT image data are generated. Surgeons and radiologists often need to manually label thousands of CT scan slice images, which is a heavy burden for them. Therefore, efficient medical aided diagnosis algorithms have become an urgent need. [0003] But so far, no aux...

AI Technical Summary

Problems solved by technology

[0003] But so far there is no auxiliary diagnostic algorithm that can reach the clinical level, the main reason is that the accuracy of the algorithm is not high and the generalization ability is not strong.

For example, the level of the detection algorithm for liver cancer mainly depends on the results of the Liver TumorSegmentation Challenge. Among them, the fitting degree of liver segmentation can reach up to 0.96, but under the premise of low false alarm rate, the fitting degree of lesions can only reach 0.702 at most, and the precision Difficult to reach the level of experts, difficult to carry out clinical application

Images