Pulmonary lobe segmentation method based on full convolutional neural network

Abstract

The invention provides a pulmonary lobe segmentation method based on a full convolutional neural network. According to the technical scheme, the method comprises: constructing a pulmonary lobe segmentation data set; obtaining a 3D bounding box of the lung organ; preprocessing the data in the lung 3D bounding box; Inputting the data blocks into a full convolutional neural network for training; andinputting the data blocks into the trained network for prediction. Due to the fact that the full convolutional neural network is adopted, end-to-end training and prediction are achieved, manual intervention is not needed, and the prediction speed is high. Moreover, segmentation is carried out in the lung bounding box, the interference of 3D bounding box external information of the lung on lung lobe segmentation is eliminated, and the integrity and details of lung lobe segmentation are obviously superior to those of a traditional method. Lung lobe region segmentation can be well realized for lung CT data with obvious diseases, so that technical support is provided for further quantitative and qualitative evaluation of lung diseases. Compared with a traditional algorithm, the method has theadvantages that the pulmonary lobe segmentation precision is obviously improved, and full-automatic pulmonary lobe segmentation is realized.

Description

technical field [0001] The invention relates to the field of medical image processing, in particular to a lung lobe segmentation method based on a fully convolutional neural network. Background technique [0002] In recent years, due to the rapid development of medical imaging and computer technology, it has become a general trend to better integrate computer technology into medical imaging. High-definition, high-contrast CT images are usually used in the diagnosis of lung diseases. Observing the structural and functional characteristics of the lungs with the help of chest CT is an important auxiliary means for various lung diseases in clinical practice. In order to provide doctors with reliable diagnostic data and facilitate early detection and treatment of patients' conditions, follow-up processing of chest CT images is usually required. , to extract and segment lung tissue images. [0003] At present, many segmentation methods have been applied to lung segmentation. The...

AI Technical Summary

Problems solved by technology

This method can extract image features of some data, but requires a large number of training samples, the segmentation results are highly dependent on samples and features, and the processing time is long

(4) The method based on image registration and shape model, which generally works better, but it is affected by the training set data will lead to large variability in results, it is difficult to establish a model, and the amount of calculation is large, resulting in slow speed and difficult Meet the real-time requirements of clinical applications

[0004] In summary, due to the influence of many factors such as various separation steps, many traditional segmentation methods are difficult to segment the ideal effect, and the robustness is not strong. It needs to be improved to improve the segmentation speed and accuracy to meet the requirements of medical diagnosis on the lungs. image request

Images