Automatic analysis system for magnetic resonance imaging and operation method thereof

Abstract

The present invention proposes an automatic analysis system for magnetic resonance imaging and its operating method, which includes the following steps: obtaining multiple images of the subject's brain from a nuclear magnetic resonance imaging machine; obtaining contrast-enhanced T1-weighted images and T2-weighted images from the multiple images, and preprocessing the multiple images; calculating the ratio of the T2-weighted image to the contrast-enhanced T1-weighted image to produce a contrast-enhanced image; performing unsupervised clustering on the region of interest of the contrast-enhanced image to separate the cystic part and the non-cystic part, and calculating multiple related feature parameters based on this; analyzing the volume change of the tumor after radiosurgery on the brain tumor corresponding to the region of interest; performing linear regression analysis on the multiple feature parameters and the volume change of the tumor as a basis for prognostic evaluation. Through the technical solution of the present invention, a regression model can be reliably established to evaluate the prognosis of tumor treatment.

Description

Technical Field

[0001] The present invention relates to a system and an operating method thereof, and in particular to an automatic analysis system for magnetic resonance imaging and an operating method thereof. Background Art

[0002] Abnormal proliferation of tissues and cells, when they grow to a large number, can become a tumor. Tumors can be benign or malignant. For example, acoustic neuroma is generally a benign, slow-growing tumor. Treatment options include radiosurgery, such as Gamma Knife radiosurgery. However, in some patients, the tumor actually grows larger after radiosurgery, complicating prognosis.

[0003] Therefore, how to provide an automatic analysis system and its operation method has become an important issue. Summary of the Invention

[0004] The present invention provides an automatic analysis system and an operating method thereof, which improve the problems of the prior art.

[0005] In one embodiment of the present invention, the automatic magnetic resonance imaging analysis system of the present invention includes a memory and a processor. The memory stores at least one instruction, and the processor is communicatively coupled to the memory. The processor is configured to access and execute at least one instruction to: obtain multiple images of a subject's brain from a magnetic resonance imaging machine; derive contrast-enhanced T1-weighted images and T2-weighted images from the multiple images and pre-process the images; calculate the ratio of the T2-weighted image to the contrast-enhanced T1-weighted image to generate a contrast-enhanced image; perform unsupervised clustering on a region of interest (ROI) in the contrast-enhanced image to separate cystic and non-cystic regions, and calculate multiple related feature parameters accordingly; analyze tumor volume changes after radiosurgery in a brain tumor corresponding to the ROI; and perform linear regression analysis between the multiple feature parameters and the tumor volume changes to serve as a basis for prognostic assessment.

[0006] In one embodiment of the present invention, after preprocessing and homogeneity correction of multiple images, the T2-weighted image is aligned with the contrast-enhanced T1-weighted image, and brain segmentation is performed on both the T2-weighted image and the contrast-enhanced T1-weighted image, and gray matter regions, white matter regions, cerebrospinal fluid regions, bone regions, and soft tissue regions are generated respectively.

[0007] In one embodiment of the present invention, the ratio of the T2-weighted image to the contrast-enhanced T1-weighted image is calculated to generate a contrast-enhanced image that satisfies the following relationship:

[0008] SI(T2w / T1wC)=(SI(T2w) / WM mean SI(T2w)) / (SI(T1wC) / WM mean SI(T1wC)),

[0009] Wherein, SI(T2w / T1wC) is the signal intensity of contrast-enhanced images, SI(T2w) is the signal intensity of T2-weighted images, WM mean SI(T2w) is the mean signal intensity of white matter areas in T2-weighted images, SI(T1wC) is the signal intensity of contrast-enhanced T1-weighted images, and WM mean SI(T1wC) is the mean signal intensity of white matter areas in contrast-enhanced T1-weighted images.

[0010] In one embodiment of the present invention, unsupervised clustering is performed by using a median filter to eliminate the extreme voxel signal intensities of the contrast-enhanced image, and then using a fuzzy C-means clustering method to separate the contrast-enhanced image into cystic and non-cystic parts based on the signal intensity differences.

[0011] In one embodiment of the present invention, the results of the linear regression analysis are that the volume of the tumor, the average signal intensity of the tumor in the contrast-enhanced image, the average signal intensity of the cyst part in the contrast-enhanced image, the average signal intensity of the non-cyst part in the contrast-enhanced image, age, and the ratio of cysts to tumors among multiple characteristic parameters are positively correlated with the reduction in tumor volume after radiosurgery.

[0012] In one embodiment of the present invention, the operating method of the automatic analysis system for magnetic resonance imaging proposed in the present invention includes the following steps: obtaining multiple images of the subject's brain from a nuclear magnetic resonance imaging machine; deriving contrast-enhanced T1-weighted images and T2-weighted images from the multiple images, and preprocessing the multiple images; calculating the ratio of the T2-weighted image to the contrast-enhanced T1-weighted image to produce a contrast-enhanced image; performing unsupervised clustering on the region of interest of the contrast-enhanced image to distinguish cystic parts from non-cystic parts, and calculating multiple related feature parameters based on the clustering; analyzing the volume change of the brain tumor corresponding to the region of interest after radiosurgery; and performing linear regression analysis on the multiple feature parameters and the volume change of the tumor as a basis for prognostic assessment.

[0013] In one embodiment of the present invention, after pre-processing and homogeneity correction of multiple images, the T2-weighted image is aligned with the contrast-enhanced T1-weighted image, and brain segmentation is performed on both the T2-weighted image and the contrast-enhanced T1-weighted image, and gray matter regions, white matter regions, cerebrospinal fluid regions, bone regions, and soft tissue regions are generated respectively.

[0014] In one embodiment of the present invention, the ratio of the T2-weighted image to the contrast-enhanced T1-weighted image is calculated to generate a contrast-enhanced image that satisfies the following relationship:

[0015] SI(T2w / T1wC)=(SI(T2w) / WM mean SI(T2w)) / (SI(T1wC) / WM mean SI(T1wC)),

[0016] Wherein, SI(T2w / T1wC) is the signal intensity of contrast-enhanced images, SI(T2w) is the signal intensity of T2-weighted images, WM mean SI(T2w) is the mean signal intensity of white matter areas in T2-weighted images, SI(T1wC) is the signal intensity of contrast-enhanced T1-weighted images, and WM mean SI(T1wC) is the mean signal intensity of white matter areas in contrast-enhanced T1-weighted images.

[0017] In one embodiment of the present invention, unsupervised clustering is performed by using a median filter to eliminate the extreme voxel signal intensities of the contrast-enhanced image, and then using a fuzzy C-means clustering method to separate the contrast-enhanced image into cystic and non-cystic parts based on the signal intensity differences.

[0018] In one embodiment of the present invention, the results of the linear regression analysis are that the volume of the tumor, the average signal intensity of the tumor in the contrast-enhanced image, the average signal intensity of the cyst part in the contrast-enhanced image, the average signal intensity of the non-cyst part in the contrast-enhanced image, age, and the ratio of cysts to tumors among multiple characteristic parameters are positively correlated with the reduction in tumor volume after radiosurgery.

[0019] In summary, the technical solution of the present invention has obvious advantages and beneficial effects compared with the prior art. Through the technical solution of the present invention, a regression model is reliably established to evaluate the prognosis of tumor treatment.

[0020] The above description will be described in detail below with reference to implementation examples, and a further explanation of the technical solution of the present invention will be provided. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] To make the above and other objects, features, advantages and embodiments of the present invention more apparent, the following descriptions of the accompanying drawings are given:

[0022] Figure 1 is a block diagram of an automatic analysis system for magnetic resonance imaging according to an embodiment of the present invention; and

[0023] Figure 2 The flowchart is a method for operating an automatic analysis system for magnetic resonance imaging according to one embodiment of the present invention.

[0024] The description of the accompanying drawings is as follows:

[0025] 100: Automatic analysis system for magnetic resonance imaging

[0026] 110: Memory

[0027] 120: Processor

[0028] 130: Display

[0029] 190: MRI machine

[0030] 200: Run method

[0031] S201~S206:Steps DETAILED DESCRIPTION

[0032] To make the description of the present invention more detailed and complete, reference is made to the accompanying drawings and various embodiments described below. In the drawings, like numbers represent the same or similar elements. On the other hand, well-known elements and steps are not described in the embodiments to avoid unnecessary limitations on the present invention.

[0033] In the embodiments and claims, the description involving “connection” may generally refer to one element being indirectly coupled to another element through other elements, or one element being directly connected to another element without going through other elements.

[0034] In the embodiments and claims, the description of “connection” may generally refer to one component indirectly communicating with another component via other components through wired and / or wireless communication, or one component being physically connected to another component without going through other components.

[0035] In the embodiments and claims, unless the context specifically limits the use of the article, "a", "an" and "the" may refer to one or more items.

[0036] As used herein, "about," "approximately," or "substantially" is used to modify any quantity that may vary slightly, but such slight variation does not alter its essence. Unless otherwise specified in the embodiments, the error range of the value modified by "about," "approximately," or "substantially" is generally within 20%, preferably within 10%, and more preferably within 5%.

[0037] Figure 1 FIG. 1 is a block diagram of an automatic analysis system 100 for magnetic resonance imaging according to an embodiment of the present invention. Figure 1 As shown, the automatic analysis system 100 for magnetic resonance imaging includes a memory 110, a processor 120, and a display 130. For example, the memory 110 may be a hard disk, a flash memory, or other storage media, the processor 120 may be a central processing unit, and the display 130 may be a built-in display or an external screen.

[0038] In terms of architecture, the automatic MRI analysis system 100 is communicatively coupled to the MRI machine 190 , and the memory 110 and the display 130 are communicatively coupled to the processor 120 .

[0039] During use, the memory 110 stores at least one instruction, and the processor 120 is communicatively coupled to the memory. The processor 120 is configured to access and execute at least one instruction to: obtain multiple images of a subject's brain from an MRI machine 190; derive contrast-enhanced T1-weighted images and T2-weighted images from the multiple images and pre-process the multiple images; calculate the ratio of the T2-weighted image to the contrast-enhanced T1-weighted image to generate a contrast-enhanced image; perform unsupervised clustering on a region of interest (ROI) in the contrast-enhanced image to separate cystic and non-cystic regions, and calculate multiple related feature parameters accordingly; analyze tumor volume changes after radiosurgery in a brain tumor corresponding to the ROI; and perform linear regression analysis between the multiple feature parameters and the tumor volume changes to serve as a basis for prognostic assessment.

[0040] In one embodiment of the present invention, after preprocessing and homogeneity correction of multiple images, the T2-weighted image is aligned with the contrast-enhanced T1-weighted image, and brain segmentation is performed on both the T2-weighted image and the contrast-enhanced T1-weighted image, and gray matter regions, white matter regions, cerebrospinal fluid regions, bone regions, and soft tissue regions are generated respectively.

[0041] The cystic portion of a tumor (e.g., an acoustic neuroma) typically exhibits low signal intensity on contrast-enhanced T1-weighted images and high signal intensity on T2-weighted images. In contrast, the non-cystic portion typically exhibits high signal intensity on contrast-enhanced T1-weighted images and low signal intensity on T2-weighted images. Therefore, to enhance the contrast between the cystic and non-cystic portions of the tumor, thereby improving the performance of the subsequent fuzzy C-means clustering method, in one embodiment of the present invention, the processor 120 accesses and executes instructions to calculate the ratio of the T2-weighted image to the contrast-enhanced T1-weighted image to generate a contrast-enhanced image that satisfies the following relationship:

[0042] SI(T2w / T1wC)=(SI(T2w) / WM mean SI(T2w)) / (SI(T1wC) / WM mean SI(T1wC)),

[0043] Wherein, SI(T2w / T1wC) is the signal intensity of contrast-enhanced images, SI(T2w) is the signal intensity of T2-weighted images, WM mean SI(T2w) is the mean signal intensity of white matter areas in T2-weighted images, SI(T1wC) is the signal intensity of contrast-enhanced T1-weighted images, and WM mean SI(T1wC) is the mean signal intensity of white matter areas in contrast-enhanced T1-weighted images.

[0044] Continuing from the above, in one embodiment of the present invention, unsupervised clustering uses a median filter to eliminate extreme voxel signal intensities in the contrast-enhanced image. Fuzzy C-means clustering is then used to separate the contrast-enhanced image into cystic and non-cystic components based on signal intensity differences. Processor 120 then accesses and executes instructions to calculate multiple related feature parameters.

[0045] For example, a median filter can be implemented in Python. Fuzzy C-means clustering is an algorithm in which each data point (e.g., voxel) can belong to two or more clusters. Voxels in a region of interest (ROI) in contrast-enhanced images are classified into cystic and non-cystic components based on signal intensity differences. In practice, the ROI corresponding to the tumor location can be manually defined by a specialist or pre-defined by a computer.

[0046] For example, multiple feature parameters may include radiological feature quantification. Radiological features include the average signal intensity of the tumor, the average signal intensity of the cystic part, the average signal intensity of the non-cystic part, the ratio of the cyst to the tumor, and cyst shape characteristics (such as sphericity, flatness, and elongation). In practice, the processor 120 is used to access and execute instructions to calculate the average signal intensity of the tumor by averaging the signal intensity of the region of interest on the contrast-enhanced image. The average signal intensity of the cystic part and the average signal intensity of the non-cystic part are obtained by averaging the signal intensities of the cystic part and the non-cystic part of the tumor, respectively, and then segmenting them using the fuzzy C-means clustering method. The cyst ratio is defined as the ratio of the volume of the cystic part to the volume of the tumor. The processor 120 can execute the PyRadiomics software package to obtain the spherical characteristics of the cystic part, which include sphericity, flatness, and elongation.

[0047] After radiosurgery (e.g., gamma knife radiosurgery) is performed on a brain tumor corresponding to a region of interest (e.g., an acoustic neuroma), the MRI machine 190 can capture postoperative images. The processor 120 is used to access and execute instructions to measure the volume of the tumor based on the postoperative images, thereby analyzing the volume change of the tumor. In practice, an exponential fitting model is suitable for estimating the volume change of a tumor after radiosurgery. Therefore, the tumor response to radiosurgery is evaluated by the specific growth rate, which can be derived from the following formula:

[0048] SGR=ln(Vt / Vo) / t,

[0049] Where SGR is the specific growth rate, Vo is the tumor volume at the time of radiosurgery, Vt is the most recently measured tumor volume, and t is the time between the radiosurgery and the most recent one. In practice, the patient's gender, age, and radiation dose can also be integrated into these multiple characteristic parameters.

[0050] Regarding linear regression, in one embodiment of the present invention, linear regression can be univariate linear regression and / or multivariate regression, used to evaluate the relationship between clinical variables and radiological features after radiosurgery and SGR. The results of the linear regression analysis indicate that tumor volume, mean signal intensity of the tumor on contrast-enhanced images, mean signal intensity of the cystic portion on contrast-enhanced images, mean signal intensity of the non-cystic portion on contrast-enhanced images, age, and cyst ratio, among multiple characteristic parameters, are positively correlated with tumor volume reduction after radiosurgery. In other words, larger tumor volume, stronger mean signal intensity of the tumor on contrast-enhanced images, stronger mean signal intensity of the cystic portion on contrast-enhanced images, stronger mean signal intensity of the non-cystic portion on contrast-enhanced images, older age, and higher cyst ratio are associated with smaller tumor volume after radiosurgery. Thus, the display 130 can display the results of the linear regression analysis, effectively serving as a basis for prognostic assessment.

[0051] To further explain the operation method of the automatic analysis system 100 for magnetic resonance imaging, please refer to Figures 1 and 2 , Figure 2 FIG. 2 is a flow chart of an operating method 200 of an automatic analysis system 100 for magnetic resonance imaging according to an embodiment of the present invention. Figure 2 As shown, the operating method 200 includes steps S201 to S206 (it should be understood that the steps mentioned in this embodiment, except for those whose order is specifically described, can be adjusted in sequence according to actual needs, and can even be executed simultaneously or partially simultaneously).

[0052] In step S201, multiple images of the subject's brain are obtained from the magnetic resonance imaging machine 190. In step S202, contrast-enhanced T1-weighted images and T2-weighted images are obtained from the multiple images, and the multiple images are pre-processed. In step S203, the ratio of the T2-weighted image to the contrast-enhanced T1-weighted image is calculated to generate a contrast-enhanced image. In step S204, unsupervised clustering is performed on the region of interest of the contrast-enhanced image to separate the cyst part and the non-cyst part, and multiple related feature parameters are calculated accordingly. In step S205, after radiosurgery is performed on the brain tumor corresponding to the region of interest, the volume change of the tumor is analyzed. In step S206, a linear regression analysis is performed on the multiple feature parameters and the volume change of the tumor to serve as a basis for prognostic evaluation.

[0053] In one embodiment of step S202, after pre-processing and homogeneity correction of multiple images, the T2-weighted image is aligned with the contrast-enhanced T1-weighted image, and brain segmentation is performed on both the T2-weighted image and the contrast-enhanced T1-weighted image to generate gray matter regions, white matter regions, cerebrospinal fluid regions, bone regions, and soft tissue regions, respectively.

[0054] In one embodiment of step S203, calculating the ratio of the T2-weighted image to the contrast-enhanced T1-weighted image to generate the contrast-enhanced image satisfies the following relationship:

[0055] SI(T2w / T1wC)=(SI(T2w) / WM mean SI(T2w)) / (SI(T1wC) / WM mean SI(T1wC)),

[0056] Wherein, SI(T2w / T1wC) is the signal intensity of contrast-enhanced images, SI(T2w) is the signal intensity of T2-weighted images, WM mean SI(T2w) is the mean signal intensity of white matter areas in T2-weighted images, SI(T1wC) is the signal intensity of contrast-enhanced T1-weighted images, and WM mean SI(T1wC) is the mean signal intensity of white matter areas in contrast-enhanced T1-weighted images.

[0057] In one embodiment of step S204, unsupervised clustering is performed by using a median filter to eliminate extreme voxel signal intensities in the contrast-enhanced image, and then using a fuzzy C-means clustering method to separate the contrast-enhanced image into cystic and non-cystic parts based on signal intensity differences.

[0058] In one embodiment of step S206, the result of the linear regression analysis is that the volume of the tumor, the average signal intensity of the tumor in the contrast-enhanced image, the average signal intensity of the cyst part in the contrast-enhanced image, the average signal intensity of the non-cyst part in the contrast-enhanced image, age, and the ratio of cysts to tumors among the multiple characteristic parameters are positively correlated with the reduction in tumor volume after radiosurgery.

[0059] In summary, the technical solution of the present invention has obvious advantages and beneficial effects compared with the prior art. Through the technical solution of the present invention, a regression model is reliably established to evaluate the prognosis of tumor treatment.

[0060] Although the present invention has been disclosed in the form of embodiments as described above, this is not intended to limit the present invention. Any person skilled in the art may make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. An automatic analysis system for magnetic resonance imaging, characterized in that: Include: a memory storing at least one instruction; and a processor communicatively coupled to the memory, wherein the processor is configured to access and execute the at least one instruction to: Acquire multiple images of a brain of a subject from a magnetic resonance imaging machine; A contrast-enhanced T1-weighted image and a T2-weighted image are obtained from the image, and the image is preprocessed. After the preprocessing corrects the image for homogeneity, the T2-weighted image is aligned with the contrast-enhanced T1-weighted image, and brain segmentation is performed on both the T2-weighted image and the contrast-enhanced T1-weighted image to generate gray matter regions, white matter regions, cerebrospinal fluid regions, bone regions, and soft tissue regions, respectively. Calculating a ratio of the T2-weighted image to the contrast-enhanced T1-weighted image to generate a contrast-enhanced image, wherein the ratio of the T2-weighted image to the contrast-enhanced T1-weighted image to generate the contrast-enhanced image satisfies the following relationship: SI(T2w / T1wC)=(SI(T2w) / WM mean SI(T2w)) / (SI(T1wC) / WM mean SI(T1wC)), wherein SI(T2w / T1wC) is the signal intensity of the contrast-enhanced image, SI(T2w) is the signal intensity of the T2-weighted image, WM mean SI(T2w) is the mean signal intensity of the white matter region of the T2-weighted image, SI(T1wC) is the signal intensity of the contrast-enhanced T1-weighted image, and WM mean SI(T1wC) is the mean signal intensity of the white matter region of the contrast-enhanced T1-weighted image; performing unsupervised clustering on a region of interest of the contrast-enhanced image to separate a cystic portion and a non-cystic portion, and calculating a plurality of related feature parameters accordingly; analyzing a volume change of a brain tumor corresponding to the region of interest after radiosurgery is performed on the tumor; and A linear regression analysis is performed between the characteristic parameters and the volume change of the tumor to serve as a basis for prognostic evaluation.

2. The automatic analysis system for magnetic resonance imaging according to claim 1, wherein: The unsupervised clustering eliminates the extreme voxel signal intensities of the contrast-enhanced image through a median filter, and then separates the contrast-enhanced image into the cystic part and the non-cystic part according to the signal intensity difference through a fuzzy C-means clustering method.

3. The automatic analysis system for magnetic resonance imaging according to claim 2, wherein: The results of the linear regression analysis showed that the characteristic parameters, including the volume of the tumor, the average signal intensity of the tumor in the contrast-enhanced image, the average signal intensity of the cystic part in the contrast-enhanced image, the average signal intensity of the non-cystic part in the contrast-enhanced image, age, and the ratio of cysts to tumors, were positively correlated with the reduction in tumor volume after radiosurgery.

4. A method for operating an automatic analysis system for magnetic resonance imaging, characterized in that: The run method contains: Acquire multiple images of a brain of a subject from a magnetic resonance imaging machine; A contrast-enhanced T1-weighted image and a T2-weighted image are obtained from the image, and the image is preprocessed. After the preprocessing corrects the image for homogeneity, the T2-weighted image is aligned with the contrast-enhanced T1-weighted image, and brain segmentation is performed on both the T2-weighted image and the contrast-enhanced T1-weighted image to generate gray matter regions, white matter regions, cerebrospinal fluid regions, bone regions, and soft tissue regions, respectively. Calculating a ratio of the T2-weighted image to the contrast-enhanced T1-weighted image to generate a contrast-enhanced image, wherein the ratio of the T2-weighted image to the contrast-enhanced T1-weighted image to generate the contrast-enhanced image satisfies the following relationship: SI(T2w / T1wC)=(SI(T2w) / WM mean SI(T2w)) / (SI(T1wC) / WM mean SI(T1wC)), wherein SI(T2w / T1wC) is the signal intensity of the contrast-enhanced image, SI(T2w) is the signal intensity of the T2-weighted image, WM mean SI(T2w) is the mean signal intensity of the white matter region of the T2-weighted image, SI(T1wC) is the signal intensity of the contrast-enhanced T1-weighted image, and WM mean SI(T1wC) is the mean signal intensity of the white matter region of the contrast-enhanced T1-weighted image; performing unsupervised clustering on a region of interest of the contrast-enhanced image to separate a cystic portion and a non-cystic portion, and calculating a plurality of related feature parameters accordingly; analyzing a volume change of a brain tumor corresponding to the region of interest after radiosurgery is performed on the tumor; and A linear regression analysis is performed between the characteristic parameters and the volume change of the tumor to serve as a basis for prognostic evaluation.

5. The operating method according to claim 4, characterized in that: The unsupervised clustering eliminates the extreme voxel signal intensities of the contrast-enhanced image through a median filter, and then separates the contrast-enhanced image into the cystic part and the non-cystic part according to the signal intensity difference through a fuzzy C-means clustering method.

6. The operating method according to claim 5, characterized in that: The results of the linear regression analysis showed that the characteristic parameters, including the volume of the tumor, the average signal intensity of the tumor in the contrast-enhanced image, the average signal intensity of the cystic part in the contrast-enhanced image, the average signal intensity of the non-cystic part in the contrast-enhanced image, age, and the ratio of cysts to tumors, were positively correlated with the reduction in tumor volume after radiosurgery.

Images