TWI730777B - Medical image synthesis method - Google Patents

Abstract

The present invention is a medical image synthesis method, which includes inputting first model information and second model information, and extracting the tissue density value of the second model information, and defining the part greater than the preset value of tissue density as immutable tissue information; At least three reference feature points are generated on the immutable tissue information of the first model information, and three corresponding feature points corresponding to the reference feature points are searched for in the immutable tissue information of the second model information; finally, the three reference features of the first model information The point is superimposed with the three corresponding feature points of the second model information with a reference to fuse the first model information and the second model information to generate synthetic model information. The invention can quickly synthesize different types of medical images, can reduce repeated comparisons of multiple images, and can improve diagnosis efficiency and accuracy.

Description

Medical image synthesis method

The present invention relates to a technology for image data processing or generation, in particular to a synthetic method that can effectively integrate medical images.

Computed tomography (CT) uses countless X-rays to penetrate the human body to obtain multiple CT images of the inside of the human body, and then a computer synthesizes the multiple CT images into three-dimensional tomographic images.

Magnetic Resonance Imaging (MRI) is also used to show the state of the patient's body. The principle of magnetic resonance imaging is to use a strong magnetic field to cause the nuclei to release electromagnetic waves. Because different substances have different hydrogen atoms content, different substances will produce different Resonance phenomenon, and convert it into an image display.

At present, both CT imaging and MRI are widely used in medical examinations. Although both can non-destructively observe the static structure and dynamic function of the body to obtain the characteristics of organs, tissues, and nerves, CT imaging and MRI The imaging clarity presented by different tissues is different. In detail, CT images are more advantageous for detecting and imaging hard tissues, such as bones. Therefore, CT images are often used to diagnose diseases such as skulls and calcified lesions; MRI is for the detection of soft tissues such as brain, internal organs, and blood vessels. Imaging has advantages and is often used in the diagnosis of soft tissue diseases such as astrocytomas, ganglia, gliomas, arteriovenous malformations and hematomas in the brain.

Because CT images and MRI have different imaging definitions for different tissues, usually doctors still use the human eye to repeatedly compare the information of the two when making a diagnosis. Certain data integrity has been lost and the diagnosis will be affected. Precision and efficiency.

Although there are existing ideas to merge the two information, it cannot be achieved simply because in reality it is necessary to face the mismatch in pixels or cutting numbers between CT images and MRI, and both CT images and MRI will be taken by the patient. Posture, and improve the difficulty of synthesis. Therefore, how to integrate the two information completely has become an important topic in the field of biomedicine.

In view of this, the present invention addresses the shortcomings of the above-mentioned conventional technologies and proposes a medical image synthesis method to effectively overcome the above-mentioned problems.

The main purpose of the present invention is to provide a medical image synthesis method, which can fuse different types of medical images into one detection image, which can reduce the repeated comparison of two images by doctors, and effectively improve the diagnosis efficiency and diagnosis accuracy.

Another object of the present invention is to provide a medical image synthesis method, which uses a special image synthesis method to make the medical image synthesis fast and accurate.

Another object of the present invention is to provide a medical image synthesis method that can convert different types of medical images into the same coordinates, reduce the shortcomings of incomplete synthesis, and improve synthesis efficiency.

To achieve the above objective, the present invention provides a medical image synthesis method. The steps include first inputting a first model information and a second model information. Then, the tissue density values of the first model information and the second model information are retrieved, and the part of the first model information and the second model information that is greater than the preset value of the tissue density is defined as the immutable tissue information. At least three reference feature points are generated on the immutable organization information of the first model information. Search for three corresponding feature points corresponding to the reference feature point in the immutable tissue information of the second model information. Finally, the three reference feature points of the first model information and the three corresponding feature points of the second model information are superimposed on a reference basis, and the first model information and the second model information are merged to generate a composite model information.

In this embodiment, the step of generating at least three reference feature points on the immutable tissue information of the first model information further includes defining the first generated reference feature point as the first central positioning point. Then, the first model information is rotated using the first central positioning point as a reference, the remaining two reference feature points are sequentially generated, and the relative distance and relative angle of the two reference feature points corresponding to the first central positioning point are stored.

In this embodiment, in the step of searching for the immutable tissue information of the second model information, the three corresponding feature points corresponding to the reference feature point further includes making the second model information generate the projection angle of the reference feature point according to the first one Rotate so that the second model information and the first model information present the same angle. Then search the second model information for the corresponding feature point that is the same as the first generated reference feature point in the first model information, and define the corresponding feature point as the second central positioning point. Finally, the second central positioning point is used as a reference, and the second model information is rotated according to the relative distance and relative angle to search for two corresponding feature points in the second model information.

In this embodiment, the step of fusing the first model information and the second model information to generate the synthesized model information includes comparing the information density of the first model information and the second model information, and replacing the density information with a higher information density Low-density parts to generate synthetic model information.

In this embodiment, the immutable tissue information can be bone tissue information or bronchial tissue information.

Detailed descriptions are given below by specific embodiments, so that it will be easier to understand the purpose, technical content, features, and effects of the present invention.

This embodiment is a technology for fusing different types of medical images into one detection image, which can reduce the physician's repeated comparison of two images, and can effectively improve diagnosis efficiency and diagnosis accuracy.

In order to better understand how to achieve the above effects, and then explain the implementation of the present invention, please first refer to the first figure to illustrate the device used in the medical image synthesis method of this embodiment. The medical image synthesis device 1 includes a first scanner 10 , A second scanner 12 is respectively connected to a processor 14, and the processor 14 is also connected to a display 16. The first scanner 10 may be a computerized tomography (CT) scanner, a magnetic resonance imaging (Magnetic Resonance Imaging, MRI) scanner, a Positron Emission Tomography (Positron Emission Tomography) scanner, or a single photon emission tomography. The (Single-photon Emission Computed Tomography) scanner is used to scan the patient's body to generate the first model information. The second scanner 12 can also be a computed tomography scanner, a magnetic resonance imaging scanner, a positron emission tomography scanner, or a single photon emission tomography scanner for scanning the human body to generate the second model information. The above-mentioned first scanner 10 and second scanner 12 are not limited to a computed tomography scanner, a magnetic resonance imaging scanner, a positron emission tomography scanner or a single photon emission tomography scanner, but for clear description The technical feature of this case is that the first scanner 10 is defined as a magnetic resonance imaging scanner, and the second scanner 12 is a computed tomography scanner in this embodiment.

The processor 14 can be a computer with arithmetic function to receive the first model information and the second model information generated by the first scanner 10 and the second scanner 12, and the processor 14 can process and fuse the first model information and the first model information. 2. Model information to generate a composite model information. The display 16 may be an image display, and the display 16 receives the control of the processor 14 to display composite model information.

After the description of the system architecture of this embodiment is complete, it will be explained how the medical image synthesis method of this embodiment is combined with the medical image synthesis device 1 to process medical images. Please refer to the first figure, the second figure, and the third figure A to the third figure D. The steps of the medical image synthesis method of this embodiment include: first enter step S10 and refer to the third figure A, the first scanner 10 scans the human body After generating a first model information 20, the second scanner 12 scans the human body to generate a second model information 30, which is transmitted to the processor 14, and the first model information 20 and the second model information 30 are input to the processor 14, and After the processor 14 receives the first model information 20 and the second model information 30, it can adjust the coordinate information of the first model information 20 and the second model information 30 so that the first model information 20 and the second model information 30 present the same world Coordinate information. In this embodiment, the first model information 20 is three-dimensional model information formed by a combination of complex magnetic resonance images (MRI); the second model information 30 is three-dimensional model information formed by a combination of computer tomography (CT).

Next, in step S12, the processor 14 captures the tissue density values of the first model information 20 and the second model information 30. In this embodiment, the tissue density value is in Heinz units, and the processor 14 uses the first model information 20 and the second model information 20 The tissue density value of the model information 30 determines whether the tissue is immutable tissue information or variable tissue information. For example, the immutable tissue information may be bone tissue information or bronchial tissue information, and the variable tissue information may be heart, lung, or Softer tissues such as the stomach. In this embodiment, the part greater than the preset value of tissue density is defined as immutable tissue information, and the preset value of tissue density may be 300 Hounsfield unit (HU).

As shown in step S14 and referring to the third figure B, the processor 14 then generates at least three reference feature points 22, 22', 22" on the immutable tissue information of the first model information 20, and of course the reference feature point 22 is The number selection includes but is not limited to three. When the user wants to increase the accuracy and reliability of the synthesis, the user can increase the number of reference feature points as appropriate. In this embodiment, the processor 14 may first determine the first model The amount of immutable organization information in the information 20 or the second model information 30 is selected to generate at least three reference feature points 22, 22', 22" for the first model information 20 with less immutable organization information. In detail, and with reference to the third A and third B diagrams, this embodiment exemplifies that the first model information 20 is a three-dimensional model information composed of a complex magnetic resonance image (MRI), and the second model information 30 is a computer tomography. Three-dimensional model information composed of scans (CT); the nature of MRI images is that it is easier to present soft tissues, that is, it is easy to present variable tissues. Invariable tissues such as bones are lost in many places, and there will be many places in MRI images. Fragmented surface appearance; the nature of CT images can make immutable tissues such as bones appear clearer and correct. In the 3D model, immutable tissues can present a smooth surface appearance; it can be seen that the first model of this embodiment The information 20 is the three-dimensional model information formed by a combination of multiple magnetic resonance images (MRI), and the second model information 30 is the three-dimensional model information formed by the combination of computer tomography (CT). Therefore, the first model information 20 can be rendered immutable The organization information is less than the immutable organization information that can be presented in the second model information 30. At this time, the processor 14 can select the first model information 20 with less immutable organization as the reference, which can ensure that the The three reference feature points 22, 22', 22" selected on the immutable organization of the model information 20 can be found in the corresponding positions of the second model information 30, and can be executed more efficiently.

Please continue to refer to step S14 to describe in detail the steps of the processor 14 generating at least three reference feature points 22, 22', 22" on the immutable tissue information of the first model information 20. The processor 14 can automatically randomly select the first model The three reference feature points 22, 22', 22" are arbitrarily generated on the immutable organization of the information 20, but the processor 14 will generate the first reference feature point 22 when generating the three reference feature points 22, 22', 22" , Is defined as the first central positioning point to facilitate obtaining the angles and positions of the remaining two reference feature points 22', 22" relative to the reference feature point 22. Please refer to the third figure B for coordination. When the first reference feature point 22 is generated, a projection line 24 will be generated at the same time to hit the reference feature point 22 of the immutable tissue. At this time, the processor 14 can print according to the projection line 24. The three-dimensional coordinates (X, Y, Z) of the reference feature point 22 are used to obtain the current projection angle according to the three-dimensional coordinates, that is, the current visual angle of the reference feature point 22 relative to the entire first model information 20. Then fix the first model information 20 with the first central positioning point, rotate the first model information 20 based on the first central positioning point, and generate the remaining two reference feature points 22', 22" in sequence, and then generate the remaining two reference feature points At the same time as 22' and 22", the processor 14 also stores and records the distance and relative angle of the two reference feature points 22', 22" with respect to the reference feature point 22, so as to facilitate the subsequent search for corresponding feature points in the second model information 30 More accurate and faster. In this embodiment, the first generated reference feature point 22 is a point on the rib, and then the first model information 20 is rotated based on the reference feature point 22, so that the viewing angle falls on the right shoulder to generate the reference Feature point 22', and then still use the reference feature point 22 to rotate the first model information 20 to the rear spine to select the reference feature point 22". Of course, three reference feature points 22, 22 can also be defined on the same bone. 22', 22", can effectively improve the benchmark.

In addition to the aforementioned processor 14 that can automatically generate the three reference feature points 22, 22', 22" at random, of course, users can also select the reference feature points 22, 22', 22" by themselves. The input device of the device 14 (not shown in the figure), such as operating a mouse to click the reference feature points 22, 22', 22" on the immutable tissue of the first model information 20, and the user clicks the reference feature point 22 , 22', 22", the processor 14 can only display the immutable tissues to facilitate the user to select the reference feature points 22, 22', 22". The rest is the method of clicking the reference feature points 22, 22', 22" It is the same as the three reference feature points 22, 22', and 22" that the processor 14 generates randomly and automatically, so it will not be repeated here.

Then go to step S16 and refer to the third B and third C pictures, the processor 14 searches the immutable tissue information of the second model information 30, and the three corresponding feature points 32 corresponding to the reference feature points 22, 22', 22" , 32', 32". In detail, when the processor 14 searches for the immutable tissue information of the second model information 30, the processor 14 will respond to the three corresponding feature points 32, 32', 32" corresponding to the reference feature points 22, 22', 22" The second model information 30 is controlled to rotate the second model information 30 according to the projection angle of the first generated reference feature point 22, so that the second model information 30 and the first model information 20 present the same visual angle. Then the processor 14 searches the second model information 30 for the corresponding feature point 32 that is the same as the first generated reference feature point 22 in the first model information 20, and defines the corresponding feature point 32 as the second central positioning point; processing; When searching for the second central positioning point, the device 14 captures the first image of the reference feature point 22 generated first in the first model information 20, and compares the same in the second information model 30 according to the first image. The area of the first image is defined as the corresponding feature point 32 corresponding to the first generated reference feature point 22.

After the processor 14 finds the second central positioning point in the second information model 30, it uses the second central positioning point as a reference, and rotates according to the relative distance and relative angle recorded by the previously generated reference feature points 22, 22', 22" The second model information 30 is used to search for the remaining two corresponding feature points 32', 32" in the second model information 30. Similarly, the processor 14 will also compare the image of the two corresponding feature points 32', 32" with the first Image comparison of the two reference feature points 22', 22" in the model information 20 to ensure that the two corresponding feature points 32', 32" found are correct. By generating the reference feature points 22, 22', 22" With the recorded relative distance and relative angle, the method of searching the corresponding feature point through the relative distance and relative angle can increase the search speed and accuracy, and make the synthesis speed fast.

Finally, go to step S18 and refer to the third B to third D, the processor 14 corresponds the three reference feature points 22, 22', 22" of the first model information 20 to the three reference feature points 32 of the second model information 30 , 32', and 32" are superimposed on a basis, and the first model information 20 and the second model information 30 are combined to generate a composite model information 40. The processor 14 controls the display 16 to display the composite model information 40 on the display 16. . During the fusion process, the processor 14 compares the number of information densities of the first model information 20 and the second model information 30. In this embodiment, the information density may be the pixel density, and the processor 14 compares the information density After the quantity, the part with high information density will be replaced with low information density to generate synthetic model information 40. In detail, assuming that the information density of an area in the second model information 30 is 100×100×100, and the portion of the first model information 20 relative to the area is 50×50×50, the processor 14 can The area with higher information density in the second model information 30 replaces the area with significantly lower information density in the first model information 20. The high information density replaces the low information density. The many-to-one replacement mode can use average, median, and internal Interpolation, the interpolation method can be polynomial interpolation, Lagrange Interpolation, Bilinear Interpolation, Radial Basis Function Interpolation, etc. There are methods to decide, and to ensure the truthfulness of each piece of data. However, if the information density is lower to replace the higher information density, it is a one-to-many replacement mode. At this time, it is necessary to generate virtual values through interpolation and other methods. Relatively speaking, the generated information is not accurate.

Although the foregoing embodiment of superimposing the image models generated by the magnetic resonance imaging scanner and the computed tomography scanner is described, the present invention is of course not limited to the foregoing embodiment, as long as the present invention is a superposition of any two image models. , Can use the technology of the present invention. For example, the technology of the present invention can also superimpose the image model of a CT scanner and a positron emission tomography scanner, or combine a magnetic resonance imaging scanner with a scanner or a single photon emission tomography scanner. The image model is superimposed. Through the method of the present invention, different types of medical images can be quickly fused into one detection image, which can reduce the physician's repeated comparison of two image models, and can effectively improve diagnosis efficiency and diagnosis accuracy.

In summary, the present invention can quickly fuse different types of medical images into one detection image through a special synthesis method, which can reduce the physician's repeated comparison of two images, and can effectively improve diagnosis efficiency and diagnosis accuracy. The present invention can further convert different types of medical images into the same coordinates, reduce the shortcomings of incomplete synthesis, and improve the synthesis efficiency.

Only the above are only preferred embodiments of the present invention and are not used to limit the scope of implementation of the present invention. Therefore, all equivalent changes or modifications made in accordance with the characteristics and spirit of the application scope of the present invention should be included in the patent application scope of the present invention.

1: Medical image synthesis device 10: The first scanner 12: second scanner 14: processor 16: display 20: First model information 22: Reference feature points 22’: Reference feature point 22": Reference feature point 24: Projection line 30: Second model information 32: Corresponding feature points 32’: Corresponding feature points 32”: Corresponding feature points 40: Synthetic model information S10~S18: steps

The first figure is a block diagram of the system applied in this embodiment. The second figure is a flowchart of the method of this embodiment. The third A to the third D are schematic diagrams of the continuous composite image of this embodiment.

S10~S18: steps

Claims (9)

A medical image synthesis method includes the following steps: inputting a first model information and a second model information; capturing the first model information and the second model information tissue density value, the first model information and the second model information In the second model information, the part greater than the default value of tissue density is defined as immutable tissue information; generating at least three benchmark feature points on the immutable tissue information of the first model information, including: the first generated benchmark The feature point is defined as the first central location point; and the first model information is rotated based on the first central location point, the other two reference feature points are sequentially generated, and the two reference feature points are stored corresponding to the first center The relative distance and relative angle of the positioning point; the three corresponding feature points corresponding to the reference feature point in the immutable tissue information of the second model information; and the three reference feature points of the first model information and the first model information The three corresponding feature points of the second model information are superimposed with a reference, and the first model information and the second model information are merged to generate a composite model information. The medical image synthesis method according to claim 1, wherein in the immutable tissue information of the second model information, the step of three corresponding feature points corresponding to the reference feature point includes: making the second model information Rotate according to the projection angle of the first generation of the reference feature point, so that the second model information and the first model information present the same angle; the second model information is searched for and the first model information is generated first The corresponding feature point whose reference feature point is the same, and the corresponding feature point is defined as the second central positioning point; and Taking the second central positioning point as a reference, and rotating the second model information according to the relative distance and the relative angle to search for the two corresponding feature points in the second model information. The medical image synthesis method according to claim 2, wherein the step of searching for the corresponding feature point that is the same as the reference feature point first generated in the first model information in the second model information is to extract Take the first image of the reference feature point generated first in the first model information to compare the same area of the first image in the second information model based on the first image to define the area as The corresponding feature point corresponding to the reference feature point generated first. The medical image synthesis method according to claim 1, wherein the step of fusing the first model information and the second model information to generate the synthesized model information includes comparing the first model information with the second model information In order to generate the composite model information, the high information density replaces the low information density part of the information density. The medical image synthesis method according to claim 1, wherein the first model information is three-dimensional model information formed by a combination of complex magnetic resonance imaging (MRI). The medical image synthesis method according to claim 1, wherein the second model information is 3D model information combined by computerized tomography (CT). The medical image synthesis method according to claim 1, wherein the tissue density is preset to 300 Hounsfield unit (HU). The medical image synthesis method according to claim 1, wherein the immutable tissue information is bone tissue information or bronchial tissue information. The medical image synthesis method according to claim 1, wherein the first model information and the second model information have the same world coordinate information.

Images