Network model training method and device and focus positioning method and device

Abstract

The invention discloses a network model training method and device, a focus positioning method and device, a computer readable storage medium and electronic equipment. The network model training method comprises the steps of determining a medical image including a focus, wherein the medical image is a chest orthographic image; determining focus positioning information and tissue positioning information corresponding to the medical image based on the medical image; generating training data based on the focus positioning information, the tissue positioning information and the medical image; anddetermining an initial network model, and training the initial network model based on the training data to generate a positioning model, the positioning model being used for determining positioning information corresponding to the chest orthographic image including the lesion. According to the embodiment of the invention, the method can improve the precision and robustness of the generated positioning model, and improves the positioning precision and robustness of subsequent focus positioning. In addition, the training speed can be greatly improved.

Description

technical field [0001] The present disclosure relates to the technical field of image processing, and specifically relates to a network model training method and device, a lesion localization method and device, a computer-readable storage medium, and electronic equipment. Background technique [0002] With the rapid development of medical imaging technology and image processing technology, lesion localization technology based on medical imaging has increasingly become an important means of adjuvant therapy. However, since the existing lesion localization method mainly realizes lesion localization by continuously predicting a large number of high-precision medical images (such as CT image sequences including hundreds of pieces), its localization speed is low. Moreover, because the existing lesion localization methods are easily affected by factors such as image contrast and brightness, their localization accuracy and robustness are poor. Especially when the existing lesion l...

AI Technical Summary

Problems solved by technology

However, because the existing lesion localization method mainly achieves lesion localization by continuously predicting a large number of high-precision medical images (such as a sequence of hundreds of CT images), its localization speed is relatively low.

Moreover, because the existing lesion localization methods are easily affected by factors such as image contrast and brightness, their localization accuracy and robustness are poor.

Especially when the existing lesion localization method is applied to medical images containing lungs, respiratory motion artifacts will directly affect the positioning accuracy, and even directly lead to localization failure

Images