CN-115984130-B - Method for acquiring perfusion parameter map and lesion area and readable storage medium

CN115984130BCN 115984130 BCN115984130 BCN 115984130BCN-115984130-B

Abstract

The application relates to a method for acquiring a perfusion parameter map and a lesion area and a computer readable storage medium, wherein the method for acquiring the perfusion parameter map and the lesion area comprises the steps of acquiring three-dimensional dynamic CT perfusion images based on brain CT perfusion images at different moments, filtering each voxel in the three-dimensional dynamic CT perfusion images to obtain a filtered three-dimensional dynamic CT perfusion image, and filtering each current voxel according to the different moments by utilizing the spatial similarity and the temporal intensity similarity of voxels adjacent to the current voxel, and acquiring the perfusion parameter map and the lesion area according to the filtered three-dimensional dynamic CT perfusion images. The application utilizes the spatial similarity and the time intensity similarity to carry out filtering, considers not only the spatial similarity information, but also the time information of the CT perfusion images which are continuously scanned, improves the signal-to-noise ratio and the calculation accuracy of the images, and obtains the perfusion parameter map and the lesion area which are obtained in the subsequent flow more reliably.

Inventors

XIANG JIANPING
LIU XIN
HE JINGSONG
SHAN YEJIE

Assignees

杭州脉流科技有限公司

Dates

Publication Date: 20260512
Application Date: 20221216

Claims (9)

1. A method for acquiring a perfusion parameter map and a lesion area, comprising: Obtaining three-dimensional dynamic CT perfusion images based on brain CT perfusion images at different moments, wherein the three-dimensional dynamic CT perfusion images at each moment are formed by sequentially carrying out image registration, skull removal and lateral ventricle removal on the brain CT perfusion images at the moment; Filtering each voxel in the three-dimensional dynamic CT perfusion image to obtain a filtered three-dimensional dynamic CT perfusion image, wherein the filtering comprises the steps of filtering each current voxel by utilizing the spatial similarity and the time intensity similarity of the adjacent voxels of the current voxel at different moments, wherein the time intensity similarity is the similarity between curves of the CT values of the current voxel and the adjacent voxels along with the time; and acquiring a perfusion parameter map and a lesion area according to the filtered three-dimensional dynamic CT perfusion image.
2. The method of obtaining perfusion parameter maps and lesion areas according to claim 1, wherein the temporal intensity similarity is obtained by: In which, in the process, For the current voxel, As the neighboring voxels to the current voxel, As a standard deviation of the time-intensity similarity, As a function of the index of the values, Is a voxel With adjacent voxels Mean square error of the time-intensity curve of (c).
3. The method of obtaining a perfusion parameter map and a lesion area according to claim 2, wherein the mean square error is obtained by: wherein T is the total number of different moments of each layer of the three-dimensional dynamic CT perfusion image, Is spatially located at The intensity value of voxel p at time t, Is spatially located at The intensity value of voxel n at time t.
4. The method of obtaining perfusion parameter maps and lesion areas according to claim 1, wherein the spatial similarity is obtained by: In which, in the process, Is a voxel With adjacent voxels Is used for the distance of euclidean distance, Is the spatial similarity standard deviation.
5. The method of acquiring perfusion parameter maps and lesion areas according to claim 1, wherein the filtering process is performed by: ; Wherein, the For the normalized coefficient X, Y, Z is the preset radius of the filter kernel, For the current voxel, The neighboring voxels of the current voxel, Is a voxel And voxels (voxel) Is used for the time intensity similarity of the (c) to the (c), Is a voxel And voxels (voxel) Is used for the spatial similarity of the (c) and (d), For three spatial position coordinates of voxel n, For three spatial position coordinates of voxel p, As a variable for the accumulation of the variables, Is the space position The intensity value at the time instant t is, The intensity value at time t after filtering for voxel p.
6. The method for acquiring a perfusion parameter map and a lesion area according to claim 1, wherein acquiring the perfusion parameter map according to the filtered three-dimensional dynamic CT perfusion image specifically comprises: Obtaining a tissue time density curve, an artery input function and a vein output function according to the filtered three-dimensional dynamic CT perfusion image; Correcting the arterial input function by utilizing the venous output function to obtain a corrected arterial input function; Obtaining a residual function according to the tissue time density curve and the corrected arterial input function; obtaining a perfusion parameter map from the residual function, the perfusion parameter map including cerebral blood flow Parameter map and residual function peak time And (5) a parameter diagram.
7. The method of claim 1, wherein obtaining a lesion region comprises obtaining an infarcted core region, comprising: obtaining a three-dimensional connected domain of the low perfusion region by utilizing a first threshold value for each layer Sequentially carrying out binarization, closing operation and hole filling on the parameter map, and combining to obtain a three-dimensional connected domain of the low perfusion region; Obtaining the normal state Dividing each layer of brain CT perfusion image into left and right half brains, classifying the left and right half brains into an abnormal side and a normal side according to the ratio of voxels with the peak time of the residual function being larger than a first threshold value to the left and right half brains, and calculating to obtain the average value of cerebral blood flow of the voxels with the normal side being smaller than a second threshold value, wherein the average value is taken as the normal Reference is made to a reference standard; obtaining infarcted core area, obtaining relative Parameter diagrams, said relative The parameter diagram is the cerebral blood flow of each voxel and the normal The relative values are compared by a third threshold value with reference to a ratio graph of the reference And sequentially carrying out binarization, closing operation and hole filling on the parameter map, combining the parameter map and the three-dimensional connected domain of the low perfusion region to obtain a three-dimensional intersection region, and sequentially carrying out closing operation and hole filling on the three-dimensional intersection region of each layer to obtain an infarct core region.
8. The method of claim 7, wherein obtaining a lesion region comprises obtaining a penumbra region, comprising: and combining the three-dimensional connected domain of the low perfusion region and the infarcted core region to obtain the penumbra region.
9. A computer readable storage medium having stored thereon a computer program, characterized in that the computer program when executed by a processor realizes the steps of the method of acquiring perfusion parameter maps and lesion according to any one of claims 1-8.

Description

Method for acquiring perfusion parameter map and lesion area and readable storage medium Technical Field The present application relates to the field of medical image processing, and in particular, to a method and readable storage medium for acquiring perfusion parameter maps and lesion areas. Background With the development of medical imaging technology and computer technology, brain CT perfusion imaging (CTP) has become an important imaging method for currently examining acute ischemic stroke. By quantitatively analyzing the CT perfusion image, relevant cerebral hemodynamic perfusion parameters of a patient, such as cerebral blood volume (cerebral blood volume, CBV), cerebral blood flow (cerebral blood flow, CBF), average transmission time (MEAN TRANSIENT TIME, MTT) and the like, can be obtained, and normal brain tissues, pathological tissues (infarct cores), reperfusion ischemic tissues (ischemic penumbra) and the like can be identified according to the perfusion parameter map. The CT perfusion image is rapidly and accurately processed, a perfusion parameter map is obtained, and the positions and the ranges of the infarct core and the ischemic penumbra are identified, so that the CT perfusion image is very important for treating the patients suffering from cerebral apoplexy. However, CT perfusion images are affected by noise, and the signal-to-noise ratio is low. The main sources of noise in CT perfusion images are quantum noise, noise introduced by inherent limitations of CT hardware systems, noise introduced in the image reconstruction process, and the like. Noise in the CT perfusion image is unavoidable, and the noise can seriously affect the accuracy of the calculation of the parameter map, thereby affecting the identification of the infarct core and the ischemic penumbra region. Therefore, it is necessary to perform filtering and noise reduction on the CT perfusion image before calculating the parameter map. Most of the current post-processing methods of CT perfusion images adopt methods such as Gaussian filtering or bilateral filtering to carry out noise reduction treatment on the images, but the filtering methods only consider the spatial domain and the value domain information, and ignore the time information of the CT perfusion images which are continuously scanned, so that the noise reduction effect is limited. Disclosure of Invention In view of the foregoing, it is desirable to provide a method for acquiring a perfusion parameter map and a lesion area. The application discloses a method for acquiring a perfusion parameter map and a lesion area, which comprises the following steps: Obtaining three-dimensional dynamic CT perfusion images based on brain CT perfusion images at different moments; Filtering each voxel in the three-dimensional dynamic CT perfusion image to obtain a filtered three-dimensional dynamic CT perfusion image, wherein the filtering comprises the steps of filtering each current voxel by utilizing the spatial similarity and the time intensity similarity of voxels adjacent to the current voxel at different moments; and acquiring a perfusion parameter map and a lesion area according to the filtered three-dimensional dynamic CT perfusion image. Optionally, the temporal intensity similarity is obtained by: Wherein p is the current voxel, n is the adjacent voxel of the current voxel, sigma 1 is the standard deviation of the time-intensity similarity, exp is an exponential function, and MSD (p, n) is the mean square error of the time-intensity curve of the voxel p and the adjacent voxel n; optionally, the mean square error is obtained by: Wherein, T is the total number of different moments of each layer of the three-dimensional dynamic CT perfusion image, I (p (x, y, z, T)) is the intensity value of a voxel p at the position (x, y, z) in space at the moment T, and I (n (ζ, eta, zeta, T)) is the intensity value of a voxel n at the position (ζ, eta, zeta) in space at the moment T. Optionally, the spatial similarity is obtained by: Where d (p, n) is the Euclidean distance between voxel p and adjacent voxel n, and σ 2 is the spatial similarity standard deviation. Optionally, the filtering process is performed by the following formula: Wherein L (p) is a normalized coefficient, X, Y, Z is a preset radius of the filter kernel, p is a current voxel, n is a neighboring voxel of the current voxel, ts (p, n) is a temporal intensity similarity of the voxel p and the voxel n, ds (p, n) is a spatial similarity of the voxel p and the voxel n, (ζ, η, ζ) is three spatial position coordinates of the voxel n, (x, y, z) is three spatial position coordinates of the voxel p, I, j, k is a variable for summation, I (ζ+i, η+j, ζ+k, t)) is an intensity value of the spatial position (ζ+i, η+j, ζ+k) at time t, and F (p (x, y, z, t)) is an intensity value of the voxel p after filtering at time t. Optionally, acquiring a perfusion parameter map according to the filtered three-dimensional dynamic CT perfusion im