CT pneumonia focus automatic processing system based on deep learning

Abstract

The invention discloses a CT pneumonia focus automatic processing system based on deep learning. The CT pneumonia focus automatic processing system comprises the steps that 1, preprocessing is carried out, wherein an original lung CT three-dimensional image is input into an nnU-Net network for preprocessing; step 2, three-dimensional sampling: extracting mutually overlapped local slice images in an interested region of the lung; 3, performing lung lobe segmentation: performing lung lobe segmentation on the preprocessed image by using a DenseVNet network, and performing region growth by using a middle point of a block as a seed point after a result is obtained; 4, pneumonia focus segmentation and identification: taking left and right lung lobes as interested areas, finding out a minimum bounding box, cutting, inputting the cut image into a 3D Unet segmentation network, and then connecting a classification network for automatic identification and diagnosis; and step 5, focus quantification: counting the distribution of pneumonia focus in the branch leaves to obtain the specific distribution of pneumonia in each lung lobe. The system has the advantages that the integrated operation of automatic 3D segmentation, quantification and identification can be carried out, so that doctors can be assisted in diagnosis.

Description

technical field [0001] The invention relates to the technical field of medical image processing and auxiliary diagnosis, in particular to an automatic processing system for CT pneumonia lesions based on deep learning. Background technique [0002] Pneumonia is a common respiratory system disease. In the diagnosis of clinical pneumonia, CT imaging is a commonly used diagnostic method. The examination is convenient and intuitive. Especially for new coronary pneumonia, CT imaging is often recommended as the first choice for diagnosis. In clinical practice, diagnosis is usually made by clinicians looking at CT images, which is time-consuming and has a certain degree of subjectivity. Due to the rapid progress of pneumonia and the long diagnosis cycle, misdiagnosis and missed diagnosis will have a great impact on patients. Therefore, an intelligent diagnosis system is needed to assist doctors in detecting pneumonia lesions, and improve the efficiency of clinical diagnosis while im...

AI Technical Summary

Problems solved by technology

However, artificially designed features such as radiomics are greatly affected by changes in the image background

Images