Method for automatically segmenting left ventricle of SPECT three-dimensional reconstructed image

Abstract

The invention discloses a method for automatically segmenting a left ventricle of an SPECT three-dimensional reconstruction image, and the method comprises the steps: carrying out the equal-proportionreduction of the linear interpolation of an original SPECT chest three-dimensional image, and extracting the rigid registration parameter features of the reduced image through a feature extraction network; and automatically steering the SPECT image by utilizing the spatial transformation network and the parameter characteristics to obtain a prediction image of a standard view, obtaining a heart image from a cutting center part in the prediction image, and automatically segmenting the image through a UNET network to obtain a left ventricular structure segmentation result under the standard view. According to the invention, a deep learning network of multi-task learning is used to synchronously extract position features and semantic features of an image, and mutual supervision of dual-network features is used to achieve an effect of network integrated training, so that integrated automatic steering from different angles to a standard view, cardiac positioning and structural segmentationof a left ventricle are realized; and complexity and personal errors of manual steering and segmentation are reduced, full automation of image operation is achieved, and accuracy is improved.

Description

technical field [0001] The present invention relates to the fields of medical imaging and deep learning, in particular to a method for automatically segmenting the left ventricle of a SPECT three-dimensional reconstruction image based on a deep learning network. Background technique [0002] SPECT cardiac imaging is currently the gold standard for clinical diagnosis of cardiovascular diseases such as coronary heart disease and myocardial ischemia, as well as efficacy evaluation and prognosis judgment. Lesions, providing more detailed functional activity information of myocardial tissue. When performing SPECT clinically, a series of operations and analyzes are required on the reconstructed SPECT images. The calculation of the left ventricular ejection coefficient is an important indicator for evaluating cardiac function, which requires segmentation of the left ventricular cavity and ventricular wall , calculated by extracting the volume of the left ventricular cavity at diff...

AI Technical Summary

Problems solved by technology

[0003] The manual operation of the physician is often required to convert the image from the conventional RA view to the standard SA view of the heart for clinical analysis. This subjective operation is likely to introduce random errors and affect the accuracy of the analysis, and it takes a long time for manual operation.

For left ventricular image segmentation, the current common clinical nuclear medicine cardiac image analysis software mostly uses conventional image processing segmentation methods, such as the segmentation method based on the centerline of the left ventricular wall, the segmentation method based on the left ventricular model, and the segmentation method based on the heart atlas. methods, segmentation methods based on threshold or k-means clustering, etc., due to the low resolution of SPECT images, and heart images are easily affected by breathing and heartbeat motion, resulting in blurred image motion boundaries, the current segmentation methods often have low segmentation accuracy when performing segmentation. , The problem of inaccurate extraction of segmentation edges further affects the accuracy of quantitative analysis

Images