JP7408467B2 - Medical image processing device and medical image processing method - Google Patents

Description

Embodiments disclosed in this specification and the drawings relate to a medical image processing apparatus and a medical image processing method.

Conventionally, techniques for extracting specific anatomical structures from medical images obtained by medical image diagnostic equipment such as X-ray CT (Computed Tomography) equipment, ultrasound diagnostic equipment, and Magnetic Resonance Imaging (MRI) equipment have been known. It is being This technique is also called segmentation.

Japanese Patent Application Publication No. 2014-50457

One of the problems that the embodiments disclosed in this specification and drawings aim to solve is to improve the accuracy of extracting anatomical structures from medical images. However, the problems to be solved by the embodiments disclosed in this specification and the drawings are not limited to the above problems. Problems corresponding to each effect of each configuration shown in the embodiments described later can also be positioned as other problems.

The medical image processing apparatus according to the embodiment includes an acquisition section, a setting section, a classification section, and an extraction section. The acquisition unit acquires a medical image. The setting unit randomly sets sampling points in a region including a specific anatomical structure in the medical image. The classification unit classifies sampling points corresponding to the anatomical structure from among the sampling points. The extraction unit extracts the anatomical structure from the medical image using the classified sampling points.

FIG. 1 is a diagram showing an example of the configuration of a medical image processing apparatus according to the first embodiment. FIG. 2 is a diagram showing an example of extraction of each valve of the aorta, which is a comparative example of the first embodiment. FIG. 3 is a diagram for explaining an image generation method using Fourier transform related to a comparative example of the first embodiment. FIG. 4 is a diagram illustrating an example of heart valve extraction performed by the medical image processing apparatus according to the first embodiment. FIG. 5 is a flowchart showing the processing procedure of the processing performed by each processing function included in the processing circuit of the medical image processing apparatus according to the present embodiment. FIG. 6 is a diagram for explaining the effects obtained by the medical image processing apparatus according to the first embodiment. FIG. 7 is a diagram for explaining the effects obtained by the medical image processing apparatus according to the first embodiment. FIG. 8 is a diagram illustrating an example of heart valve extraction performed by the medical image processing apparatus according to the first modification of the first embodiment. FIG. 9 is a diagram illustrating an example of heart valve extraction performed by the medical image processing apparatus according to the second modification of the first embodiment. FIG. 10 is a diagram illustrating an example of the configuration of a medical image processing apparatus according to the second embodiment. FIG. 11 is a diagram illustrating an example of setting sampling points performed by the medical image processing apparatus according to the second embodiment. FIG. 12 is a diagram illustrating an example of setting sampling points performed by a medical image processing apparatus according to a modification of the second embodiment.

Embodiments of a medical image processing apparatus and a medical image processing method will be described in detail below with reference to the drawings.

(First embodiment)
FIG. 1 is a diagram showing an example of the configuration of a medical image processing apparatus according to the first embodiment.

For example, as shown in FIG. 1, a medical image processing apparatus 100 according to the present embodiment is connected to a medical image diagnostic apparatus 20 and a medical image storage apparatus 30 via a network 10 so as to be able to communicate with each other.

The medical image diagnostic apparatus 20 acquires medical images of a subject (patient, etc.) used for image diagnosis and the like. Specifically, the medical image diagnostic apparatus 20 generates a two-dimensional image or a three-dimensional image (also called volume data) of the subject as a medical image. For example, the medical image diagnostic device 20 is an X-ray CT device, an ultrasound diagnostic device, an MRI device, or the like.

The medical image storage device 30 acquires a medical image from the medical image diagnostic device 20 via the network 10, and stores the acquired medical image in a storage circuit within the device itself. For example, the medical image storage device 30 is realized by computer equipment such as a server or a workstation.

The medical image processing apparatus 100 acquires medical images from the medical image diagnostic apparatus 20 or the medical image storage apparatus 30 via the network 10, and performs various image processing on the acquired medical images. For example, the medical image processing apparatus 100 is realized by computer equipment such as a server, a workstation, a personal computer, or a tablet terminal.

Specifically, the medical image processing apparatus 100 includes a network (NetWork: NW) interface 110, a storage circuit 120, an input interface 130, a display 140, and a processing circuit 150.

The NW interface 110 controls transmission and communication of various data sent and received between the medical image processing apparatus 100 and other devices connected via the network 10. Specifically, the NW interface 110 is connected to the processing circuit 150 and outputs the medical image received from the medical image diagnostic apparatus 20 or the medical image storage apparatus 30 to the processing circuit 150. For example, the NW interface 110 is realized by a network card, network adapter, NIC (Network Interface Controller), or the like.

The storage circuit 120 stores various data, various programs, and the like. Specifically, the storage circuit 120 is connected to the processing circuit 150, and stores input medical images or transfers the stored medical images to the processing circuit according to commands sent from the processing circuit 150. Output to 150. For example, the memory circuit 120 is realized by a semiconductor memory element such as a RAM (Random Access Memory) or a flash memory, a hard disk, an optical disk, or the like.

The input interface 130 receives various instructions and input operations for various information from the operator. Specifically, the input interface 130 is connected to the processing circuit 150 , converts an input operation received from an operator into an electrical signal, and outputs the electrical signal to the processing circuit 150 . For example, the input interface 130 may include a trackball, a switch button, a mouse, a keyboard, a touchpad that performs input operations by touching the operation surface, a touchscreen that integrates a display screen and a touchpad, and a non-control device that uses an optical sensor. This is realized by a touch input interface, a voice input interface, etc. Note that in this specification, the input interface 130 is not limited to one that includes physical operation parts such as a mouse and a keyboard. For example, examples of the input interface 130 include an electrical signal processing circuit that receives an electrical signal corresponding to an input operation from an external input device provided separately from the device and outputs this electrical signal to a control circuit.

The display 140 displays various information and data. Specifically, the display 140 is connected to the processing circuit 150 and displays various information and data output from the processing circuit 150. For example, the display 140 is realized by a liquid crystal monitor, a CRT (cathode ray tube) monitor, a touch panel, or the like.

The processing circuit 150 controls the components of the medical image processing apparatus 100 in response to input operations received from the operator via the input interface 130. For example, the processing circuit 150 causes the storage circuit 120 to store the medical image output from the NW interface 110. Further, for example, the processing circuit 150 reads a medical image from the storage circuit 120 and displays it on the display 140.

The overall configuration of the medical image processing apparatus 100 according to the present embodiment has been described above. Based on the above-described configuration, the medical image processing apparatus 100 according to the present embodiment has a function of extracting a specific anatomical structure from a medical image obtained by the medical image diagnosis apparatus 20.

For example, when diagnosing or formulating a treatment plan for aortic valve valvular disease, a doctor or technician measures an index related to the shape of the aortic valve. The shape-related index mentioned here is, for example, the length of the free edge of each valve (right coronary cusp, left coronary cusp, non-coronary cusp) included in the aortic valve, the area of the valve leaflet, etc. When such an index is measured, a medical image processing apparatus generally performs a process of extracting each valve of the aorta from the medical image.

FIG. 2 is a diagram showing an example of extraction of each valve of the aorta, which is a comparative example of the first embodiment.

For example, as shown in FIG. 2, in a general method, sampling points are set at equal intervals for each grid point formed by pixels or voxels of a medical image (such as a clinical image). Thereafter, each sampling point is classified as whether it is a valve or a valve other than a valve, and if it is a valve, it is classified as a right coronary leaflet, a left coronary leaflet, or a valve without a crown leaflet. Then, each valve of the aorta is extracted by converting the classified sampling point group into surface data.

Here, with the above-described method, the accuracy of extracting anatomical structures may be lower depending on the type of medical image. Specifically, when a medical image generated by an image generation method that emphasizes the periodicity of frequency components included in the image is used, the accuracy of extracting anatomical structures may be lowered.

For example, in an X-ray CT device, an X-ray source that irradiates a subject with X-rays and a detector that detects the X-rays that have passed through the subject rotate on a circular orbit centered on the subject. An image is generated by moving and performing Fourier transform on the distribution of X-rays detected at each rotation angle. In such an image generation method using Fourier transform, an image is constructed by a combination of a plurality of frequency components.

FIG. 3 is a diagram for explaining an image generation method using Fourier transform related to a comparative example of the first embodiment.

For example, as shown in FIG. 3, in an image generation method using Fourier transform, the waveform of a signal indicating the distribution of X-rays is divided into multiple frequency components (4 .0×sinθ, 0.5×sin2θ, 2.0×sin3θ, 1.0×sin4θ, etc.), and an image is generated based on the magnitude of the amplitude of each frequency component. In this case, the waveform of the signal indicating the distribution of X-rays can be expanded into many waveforms, but in reality, in order to shorten processing time, only frequency components with large amplitudes are used to generate images, and Frequency components with small values are truncated.

In this way, in the image generation method using Fourier transform, since an image is constructed by a combination of a plurality of frequency components, the periodicity of the frequency components included in the image may be excessively emphasized. Therefore, for example, if noise is mixed into a signal indicating X-rays detected by a detector, the periodicity of the frequency component representing the noise may be emphasized, and as a result, periodic An error will occur. This error may reduce the accuracy of extracting anatomical structures from medical images.

For this reason, the medical image processing apparatus 100 according to the present embodiment is configured to improve the accuracy of extracting anatomical structures from medical images.

The medical image processing apparatus 100 having such a configuration will be described in detail below. Note that an example will be described below in which the specific anatomical structure extracted from the medical image is a heart valve. Here, the heart valve may be a valve included in the aortic valve, or may be a valve included in the mitral valve, tricuspid valve, or pulmonary valve.

Returning to FIG. 1, specifically, in this embodiment, the processing circuit 150 includes an acquisition function 151, a specific function 152, a setting function 153, a classification function 154, and an extraction function 155. Here, the acquisition function 151 is an example of an acquisition unit. Further, the specific function 152 is an example of a specific section. Furthermore, the setting function 153 is an example of a setting section. Furthermore, the classification function 154 is an example of a classification section. Further, the extraction function 155 is an example of an extraction unit.

The acquisition function 151 acquires medical images. Specifically, the acquisition function 151 acquires a medical image from the medical image diagnostic apparatus 20 or the medical image storage apparatus 30 via the NW interface 110. The acquisition function 151 then stores the acquired medical image in the storage circuit 120.

The identification function 152 identifies a region including a heart valve on the medical image acquired by the acquisition function 151. Specifically, the identification function 152 reads out the medical image acquired by the acquisition function 151 from the storage circuit 120, and identifies a region including the heart valve on the read medical image. Here, various known region extraction methods can be used to identify the region including the heart valve on the medical image.

The setting function 153 randomly sets sampling points in the area including the heart valve in the medical image acquired by the acquisition function 151. Specifically, the setting function 153 randomly sets sampling points for the region including the heart valve specified on the medical image by the specifying function 152.

The classification function 154 classifies sampling points corresponding to heart valves from among the sampling points set by the setting function 153. Specifically, for each sampling point set for a region including a heart valve by the setting function 153, the classification function 154 selects a pixel at the position where the sampling point is set in the medical image acquired by the acquisition function 151. The sampling points corresponding to heart valves are classified by determining whether or not they correspond to heart valves.

The extraction function 155 extracts heart valves from the medical image using the sampling points corresponding to the heart valves classified by the classification function 154. For example, the extraction function 155 extracts heart valves from the medical image by converting the sampling point group classified by the classification function 154 into surface data. The extraction function 155 then displays the extracted heart valves on the display 140. At this time, for example, the extraction function 155 measures an index related to the shape of the heart valve (such as the length of the free edge of each valve or the area of the valve leaflet) based on the extraction result of the heart valve 41, and displays the measured index. 140 may be further displayed.

FIG. 4 is a diagram illustrating an example of heart valve extraction performed by the medical image processing apparatus 100 according to the first embodiment.

For example, as shown in FIG. 4A, in this embodiment, the acquisition function 151 acquires a medical image 42 in which a heart valve 41 is imaged. For example, the medical image 42 is a CT image obtained by an X-ray CT device, an ultrasound image obtained by an ultrasound diagnostic device, an MR image obtained by an MRI device, or the like. Note that, in the example shown in FIG. 4, among the voxels included in the medical image 42, only the voxel 44 representing the heart valve 41 is shown.

Further, the specifying function 152 specifies a region 43 including the heart valve 41 on the medical image 42 obtained by the obtaining function 151. For example, the identification function 152 roughly identifies a region 43 that includes the heart valve 41 on the medical image 42 . In this case, the region 43 that includes the heart valve 41 includes not only the region of the heart valve 41 but also a part of the region around the heart valve 41 among the regions other than the heart valve 41 .

Then, the setting function 153 randomly sets sampling points for the region 43 including the heart valve 41 specified on the medical image 42 by the specifying function 152. Here, the setting function 153 sets sampling points at equal intervals to each grid point formed by the voxels of the medical image 42, as shown in FIG. Instead of setting, sampling points are randomly set as shown in FIG. 4(c). For example, in the example shown in FIG. 4, sampling points are indicated by ● and ○.

Thereafter, the classification function 154 determines, for each sampling point set by the setting function 153, whether or not the voxel 44 at the position where the sampling point is set corresponds to the heart valve 41. Classify the corresponding sampling points. For example, in the example shown in FIG. 4, sampling points classified as corresponding to the heart valve 41 are indicated by ○, and sampling points classified as not corresponding to the heart valve are indicated by ●.

Then, the extraction function 155 extracts the heart valve 41 from the medical image 42 using the sampling points classified as corresponding to the heart valve 41 by the classification function 154.

Each processing function that the processing circuit 150 has has been described above. Here, for example, the processing circuit 150 is realized by a processor. In that case, each of the processing functions described above is stored in the storage circuit 120 in the form of a computer-executable program. Then, the processing circuit 150 reads and executes each program stored in the storage circuit 120, thereby realizing a function corresponding to each program. In other words, the processing circuit 150 has each processing function shown in FIG. 1 in a state where each program is read.

Note that the processing circuit 150 may be configured by combining a plurality of independent processors, and each processor may implement each processing function by executing a program. Further, each processing function of the processing circuit 150 may be realized by being appropriately distributed or integrated into a single or multiple processing circuits. Moreover, each processing function that the processing circuit 150 has may be realized by a mixture of hardware such as a circuit and software. Further, although an example has been described here in which programs corresponding to each processing function are stored in a single storage circuit 120, the embodiment is not limited to this. For example, a configuration may be adopted in which programs corresponding to each processing function are stored in a distributed manner in a plurality of storage circuits, and the processing circuit 150 reads each program from each storage circuit and executes it.

FIG. 5 is a flowchart showing a processing procedure of processing performed by each processing function included in the processing circuit 150 of the medical image processing apparatus 100 according to the present embodiment.

For example, as shown in FIG. 5, in this embodiment, when the acquisition function 151 receives an instruction to start processing from the operator via the input interface 130 (step S101, Yes), 20 or the medical image storage device 30 (step S102). This processing is realized, for example, by the processing circuit 150 calling a predetermined program corresponding to the acquisition function 151 from the storage circuit 120 and executing it.

Subsequently, the specifying function 152 specifies a region including the heart valve on the medical image obtained by the obtaining function 151 (step S103). This processing is realized, for example, by the processing circuit 150 calling a predetermined program corresponding to the specific function 152 from the storage circuit 120 and executing it.

Subsequently, the setting function 153 randomly sets sampling points for the region including the heart valve specified on the medical image by the specifying function 152 (step S104). This processing is realized, for example, by the processing circuit 150 calling a predetermined program corresponding to the setting function 153 from the storage circuit 120 and executing it.

Subsequently, the classification function 154 classifies sampling points corresponding to heart valves from among the sampling points set by the setting function 153 (step S105). This processing is realized, for example, by the processing circuit 150 calling a predetermined program corresponding to the classification function 154 from the storage circuit 120 and executing it.

Next, the extraction function 155 extracts heart valves from the medical image using the sampling points corresponding to the heart valves classified by the classification function 154 (step S106), and displays the extracted heart valves on the display 140. (Step S107). This processing is realized, for example, by the processing circuit 150 calling a predetermined program corresponding to the extraction function 155 from the storage circuit 120 and executing it.

According to the above-described configuration, the accuracy of extracting heart valves from medical images can be improved, as described below.

6 and 7 are diagrams for explaining the effects obtained by the medical image processing apparatus 100 according to the first embodiment.

For example, medical images such as CT images, ultrasound images, and MR images are generally generated using an image generation method that uses Fourier transform, so the periodicity of frequency components representing noise may be emphasized. obtain. As a result, periodic errors will occur within the image, which may reduce the accuracy with which heart valves are extracted from medical images.

For example, as shown on the left side of FIG. 6, when the signal 61 on which the medical image is based is normal (no noise mixed in), only the frequency component 62 representing the signal is expanded by Fourier expansion. However, for example, as shown on the right side of FIG. 6, if noise is mixed into the signal 63 that is the basis of the medical image (the waveform with noise mixed in is shown by a dashed line), the frequency that would normally be expanded In addition to component 64, a noise frequency component 65 appears. This causes periodic errors in the medical image.

In such a case, for example, if sampling points are set at equal intervals on the medical image as shown on the left side of FIG. 7, the sampling points are arranged periodically, so the frequency of the sampling points and It is possible that the frequency of the noise matches. In that case, the error in classifying the sampling points becomes large.

On the other hand, in the first embodiment, as shown on the right side of FIG. 7, the sampling points are randomly set for the medical image, so the sampling points are arranged aperiodically. As a result, the frequency of the sampling point and the frequency of the noise almost never match, so that errors in classification of the sampling point can be reduced. As a result, according to the first embodiment, the accuracy of extracting heart valves from medical images can be improved.

Although the first embodiment has been described above, the above-described embodiment can also be implemented by appropriately changing a part of the configuration of the medical image processing apparatus 100. Therefore, a modification of the first embodiment will be described below.

(First modification of the first embodiment)
For example, as a first modification, the degree of sparseness of randomly set sampling points may be changed within a region including a heart valve in a medical image.

Specifically, the setting function 153 sets the sampling points densely or coarsely depending on whether the anatomical structures (two heart valves are shown as an example) in the medical image acquired by the acquisition function 151 are crowded. .

FIG. 8 is a diagram illustrating an example of heart valve extraction performed by the medical image processing apparatus 100 according to the first modification of the first embodiment.

For example, as shown in FIG. 8A, in this modification, the acquisition function 151 acquires a medical image 82 in which two heart valves 81 are imaged. Note that, in the example shown in FIG. 8, among the voxels included in the medical image 82, only voxels 88 representing two heart valves 81 are shown.

Further, the specifying function 152 specifies a region 83 including two heart valves 81 on the medical image 82 obtained by the obtaining function 151. For example, the identification function 152 roughly identifies a region 83 that includes two heart valves 81 on the medical image 82 . In this case, the region 83 that includes the two heart valves 81 includes, in addition to the region of the heart valve 81, a part of the region surrounding the heart valve 81 among the regions other than the heart valve 81.

Then, the setting function 153 randomly sets sampling points for the region 83 including the heart valve 81 specified on the medical image 82 by the specifying function 152. Here, the setting function 153 sets sampling points at equal intervals to each grid point formed by the voxels of the medical image 42, as shown in FIG. Instead of setting, sampling points are randomly set as shown in FIG. 8(c).

At this time, in this modification, the setting function 153 determines whether or not the region 83 including the heart valve 81 includes a region where heart valves are denser than other regions within the region 83. do. Then, when the area includes a region where heart valves are more densely packed than other regions, the setting function 153 sets sampling points more densely in the region than in other regions. That is, the setting function 153 increases the number of sampling points arranged per unit area for regions where it is difficult to extract heart valves, compared to other regions.

For example, the setting function 153 determines whether the region 83 including the heart valves 81 includes a junction where two heart valves 81 join as a dense region. In the vicinity of the junction, the two heart valves 81 may overlap or come into contact with each other, so it is generally difficult to extract each heart valve separately. Therefore, when the region 83 including the heart valve 81 includes a joint, the setting function 153 compares the region including the joint with other regions in the region 83 including the heart valve 81. Set the sampling points densely. For example, as shown in FIG. 8C, the setting function 153 roughly sets sampling points in the region 83 that includes the heart valve 81 and does not include the junction, and sets sampling points roughly in the region that does not include the junction. sets the sampling points densely.

After that, similarly to the first embodiment, the classification function 154 determines, for each sampling point set by the setting function 153, whether the voxel 84 at the position where the sampling point is set corresponds to the heart valve 81. By determining this, the sampling points corresponding to the heart valves 81 are classified. Further, the extraction function 155 extracts the heart valve 81 from the medical image 82 using the sampling points classified as corresponding to the heart valve 81 by the classification function 154.

According to the first modification described above, the accuracy of extracting a heart valve can be improved compared to the case where sampling points are roughly set over the entire region including the heart valve in a medical image. Further, compared to the case where sampling points are set densely over the entire region including the heart valve in a medical image, the processing time for extracting the heart valve can be shortened.

(Second modification of the first embodiment)
For example, as a second modified example, by setting sampling points for a region including a heart valve in a medical image and classifying the sampling points multiple times, a heart valve can be extracted based on multiple classification results. You can also do this.

Specifically, the setting function 153 derives a plurality of sampling results by performing a process of randomly setting sampling points for a region including a heart valve in the medical image acquired by the acquisition function 151 multiple times. . Further, the classification function 154 derives a plurality of classification results by classifying sampling points corresponding to heart valves for each of the plurality of sampling results derived by the setting function 153. Further, the extraction function 155 extracts heart valves from the medical image based on the plurality of classification results derived by the classification function 154.

FIG. 9 is a diagram illustrating an example of heart valve extraction performed by the medical image processing apparatus 100 according to the second modification of the first embodiment.

For example, as shown in FIG. 9A, in this modification, the acquisition function 151 acquires a medical image 92 in which a heart valve 91 is imaged. Note that, in the example shown in FIG. 9, among the voxels included in the medical image 92, only the voxel 94 representing the heart valve 91 is shown.

Further, the specifying function 152 specifies a region 93 including the heart valve 91 on the medical image 92 obtained by the obtaining function 151. For example, the identification function 152 roughly identifies a region 93 that includes the heart valve 91 on the medical image 92 . In this case, the region 93 that includes the heart valve 91 includes, in addition to the region of the heart valve 91, a part of the region around the heart valve 91 among the regions other than the heart valve 91.

Then, the setting function 153 randomly sets sampling points for the region 93 including the heart valve 91 specified on the medical image 92 by the specifying function 152.

At this time, in this modification, the setting function 153 performs a process of randomly setting sampling points for a region 93 including the heart valve 91 in the medical image 42, as shown in (b1) to (b3) in FIG. By performing multiple times, multiple sampling results are derived.

Thereafter, the classification function 154 determines, for each sampling point set by the setting function 153, whether or not the voxel 94 at the position where the sampling point is set corresponds to the heart valve 91. Classify the corresponding sampling points.

At this time, in this modification, the classification function 154 selects sampling points corresponding to the heart valve 91 for each of the plurality of sampling results derived by the setting function 153, as shown in (b1) to (b3) in FIG. By classifying, multiple classification results are derived.

In this modification, the extraction function 155 extracts the heart valve 91 from the medical image 92 based on the plurality of classification results derived by the classification function 154. Here, the extraction function 155 determines the final classification result based on a plurality of classification results, as shown in FIG. 9(c). For example, the extraction function 155 determines the final classification result by selecting from among the plurality of classification results the classification result that shows the shape most similar to the standard shape of a heart valve. Alternatively, for example, the extraction function 155 may determine the final classification result by calculating an average distribution from a plurality of classification results. Then, the extraction function 155 extracts the heart valve 91 from the medical image 92 using the determined final classification result.

According to the second modification described above, by extracting the shape of the valve from a plurality of classification results, it is possible to further improve the accuracy of extracting the heart valve.

(Second embodiment)
Note that in the first embodiment and the modified example described above, an example was described in which sampling points are directly set for a medical image acquired from the medical image diagnostic apparatus 20 or the medical image storage apparatus 30. It is not limited to this. For example, when a diagnosis or the like is performed using a medical image, post-processing such as noise removal processing may be performed on the medical image. In that case, the settings of the sampling points may be adjusted depending on the state of the medical image after the noise removal process is performed. Below, such an example will be described as a second embodiment. In addition, in the second embodiment, the points different from the first embodiment will be mainly explained, and the same reference numerals will be given to the same components as in the first embodiment, and detailed explanation will be omitted.

FIG. 10 is a diagram illustrating an example of the configuration of a medical image processing apparatus according to the second embodiment.

For example, as shown in FIG. 10, in this embodiment, the processing circuit 250 of the medical image processing apparatus 200 includes an acquisition function 151, a specific function 252, a setting function 253, a classification function 154, an extraction function 155, It has a noise removal function 256. Here, the specific function 252 is another example of the specific section. Further, the setting function 253 is another example of the setting section. Further, the noise removal function 256 is an example of a noise removal unit.

In this embodiment, the noise removal function 256 performs noise removal processing on the medical image acquired by the acquisition function 151. Specifically, the noise removal function 256 reads the medical image acquired by the acquisition function 151 from the storage circuit 120, and performs noise removal processing on the read medical image. Here, various known noise removal methods can be used to perform noise removal processing on the medical image. Then, the noise removal function 256 causes the storage circuit 120 to store the medical image on which the noise removal process has been performed.

Further, in this embodiment, the identification function 252 identifies a region including a heart valve on the medical image acquired by the noise removal function 256. Specifically, the specifying function 252 reads out the medical image that has been subjected to noise removal processing by the noise removal function 256 from the storage circuit 120, and specifies a region including the heart valve on the read medical image. Here, various known region extraction methods can be used to identify the region including the heart valve on the medical image.

Further, in this embodiment, the setting function 253 randomly sets sampling points for the region including the heart valve specified on the medical image by the specifying function 252.

At this time, in the present embodiment, the setting function 253 performs noise removal processing by the noise removal function 256 and converts the medical image into a plurality of signals having different frequencies, and selects a predetermined number of signals having large amplitudes from among the plurality of signals. The number of randomly set sampling points is adjusted so that the number of sampling points arranged at intervals corresponding to the period of the signal is smaller than the number of sampling points arranged at intervals corresponding to other periods.

FIG. 11 is a diagram illustrating an example of setting sampling points performed by the medical image processing apparatus 200 according to the second embodiment.

For example, as shown in FIG. 11, in this embodiment, the setting function 253 converts the medical image after noise removal processing into a plurality of signals with different frequencies, and among the plurality of converted signals, A predetermined number of signals are identified in descending order of period. For example, the setting function 253 specifies three signals: a signal with period A ₁ , a signal with period A ₂ , and a signal with period A ₃ . Thereafter, the setting function 253 configures the region including the heart valve so that the number of sampling points arranged at intervals corresponding to the period of each identified signal is less than the number of sampling points arranged at intervals corresponding to other periods. Determine the number of sampling points set for. Then, the setting function 253 randomly sets the determined number of sampling points for the region including the heart valve.

Thereafter, similarly to the first embodiment, the classification function 154 classifies sampling points corresponding to heart valves from among the sampling points set by the setting function 253. Additionally, an extraction function 155 extracts heart valves from the medical image acquired by the acquisition function 151 using the classification sampling points by the classification function 154 .

According to the second embodiment described above, even if a signal of a specific period is emphasized by noise removal processing performed on a medical image, the frequency of such a signal and the frequency of the sampling point match. Errors in classification of sampling points can be reduced. As a result, according to the second embodiment, the accuracy of extracting heart valves from medical images can be improved.

(Modified example of second embodiment)
In addition, in the second embodiment described above, an example was explained in which the setting of sampling points is adjusted according to the state of the medical image after the noise removal process. The settings of the sampling points may be adjusted accordingly.

Specifically, the setting function 253 converts the medical image that has been subjected to noise removal processing by the noise removal function 256 into a plurality of signals having different frequencies, and selects one of the plurality of signals according to the method of the noise removal processing. The number of sampling points that are randomly set such that the number of sampling points that are placed at intervals that correspond to the period of the signal to be emphasized is less than the number of sampling points that are placed at intervals that correspond to other periods. Adjust.

For example, in this modification, the noise removal function 256 performs noise removal processing by deep learning using a convolutional neural network.

FIG. 12 is a diagram illustrating an example of setting sampling points performed by the medical image processing apparatus 200 according to a modification of the second embodiment.

For example, as shown in Fig. 12, in noise removal processing using deep learning using a convolutional neural network, 4 x 4 voxels are condensed into 2 x 2 voxels by pooling multiple times on a medical image. The 2×2 voxels are further condensed into one voxel. In this process, since pooling is performed for each 4×4 voxel, signals with a period corresponding to an interval of four voxels are emphasized.

Therefore, in this case, the setting function 253 configures the region including the heart valve so that the number of sampling points arranged at intervals of four voxels is smaller than the number of sampling points arranged at other intervals. Determine the number of sampling points to be set. Then, the setting function 253 randomly sets the determined number of sampling points for the region including the heart valve.

According to the above-mentioned modification, even if a signal with a specific period is emphasized depending on the noise removal processing method, the frequency of such a signal and the frequency of the sampling point almost never match, so the sampling Errors in point classification can be reduced. As a result, according to the second embodiment, the accuracy of extracting heart valves from medical images can be further improved.

Note that in the above-mentioned modification, an example has been described in which the noise removal process is performed by deep learning using a convolutional neural network, but the method of the noise removal process is not limited to this. For example, even if noise removal processing is performed using a method other than deep learning using a convolutional neural network, if the noise removal processing emphasizes a signal with a specific period, the above-mentioned modification example can be applied. , it is possible to further improve the accuracy of extracting heart valves from medical images.

Further, in the second embodiment and the modified example described above, an example in which noise removal processing is performed has been described, but the post-processing performed on a medical image is not limited to this. For example, even if post-processing other than noise removal processing is performed, if the post-processing emphasizes a signal of a specific period, the above-mentioned modification can be applied to extract heart valves from a medical image. It is possible to further improve accuracy.

Furthermore, in the first embodiment, the second embodiment, and each modification example described above, when the medical image processing apparatus is provided with a specific function for identifying a region including a heart valve on a medical image acquired by the acquisition function, Although an example has been described, the embodiment is not limited to this. For example, a medical image processing device does not necessarily have to have specific functions. In that case, the setting function randomly sets sampling points for the entire area in the medical image acquired by the acquisition function. In this case, the processing time involved in extracting the heart valve will be longer than in the case where the specific function is provided, but the effect of improving the accuracy of extracting the heart valve can be obtained.

Note that in the above-mentioned embodiments and modifications, an example was explained in which the specific anatomical structure extracted from a medical image is a heart valve; however, the method described in the above-mentioned embodiments and modifications The present invention can be similarly applied to the case of extracting anatomical structures from medical images.

Further, the configuration of the medical image processing apparatus described in the embodiment and modification described above can also be applied to a client server system. In this case, the server device is equipped with the above-mentioned acquisition function, specific function, setting function, classification function, extraction function, and noise removal function, and can extract a specific anatomical structure from a medical image in response to a request sent from the client device. and sends the result to the client device.

Furthermore, the configuration of the medical image processing apparatus described in the embodiment and modification described above can also be applied to the medical image diagnostic apparatus 20. In this case, an information processing device such as a console device included in the medical image diagnostic apparatus 20 is provided with the above-described acquisition function, specific function, setting function, classification function, extraction function, and noise removal function. In this case, for example, each time the medical image diagnostic apparatus 20 acquires a medical image of the subject, it may use each processing function to extract a specific anatomical structure from the medical image in real time.

Furthermore, in the above-described embodiments and modified examples, the acquisition unit, identification unit, setting unit, classification unit, extraction unit, and noise removal unit in this specification are respectively referred to as the acquisition function, identification function, setting function, and classification function of the processing circuit. Although an example has been described in which the present invention is implemented using a function, an extraction function, and a noise removal function, the embodiments are not limited to this. For example, the acquisition unit, identification unit, setting unit, classification unit, extraction unit, and noise removal unit in this specification are the acquisition function, identification function, setting function, classification function, extraction function, and noise removal function described in the embodiment. Alternatively, the same function may be realized by only hardware, only software, or a mixture of hardware and software.

Furthermore, the term "processor" used in the description of the above-described embodiments refers to, for example, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or an Application Specific Integrated Circuit (ASIC). Circuits such as programmable logic devices (e.g., Simple Programmable Logic Devices (SPLDs), Complex Programmable Logic Devices (CPLDs), and Field Programmable Gate Arrays (FPGAs)) means. Here, instead of storing the program in the memory circuit, the program may be directly incorporated into the circuit of the processor. In this case, the processor realizes its functions by reading and executing a program built into the circuit. Furthermore, each processor of this embodiment is not limited to being configured as a single circuit for each processor, but may also be configured as a single processor by combining multiple independent circuits to realize its functions. good.

Here, the program executed by the processor is provided by being pre-installed in a ROM (Read Only Memory), a storage circuit, or the like. This program is a file in a format that can be installed or executable on these devices, such as CD (Compact Disk)-ROM, FD (Flexible Disk), CD-R (Recordable), DVD (Digital Versatile Disk), etc. The information may be provided recorded on a computer readable non-transitory storage medium. Further, this program may be stored on a computer connected to a network such as the Internet, and may be provided or distributed by being downloaded via the network. For example, this program is composed of modules including each of the processing functions described above. In actual hardware, a CPU reads a program from a storage medium such as a ROM and executes it, so that each module is loaded onto the main storage device and generated on the main storage device.

Furthermore, in the embodiments and modifications described above, each component of each device illustrated is functionally conceptual, and does not necessarily need to be physically configured as illustrated. In other words, the specific form of distributing or integrating each device is not limited to what is shown in the diagram, and all or part of the devices may be functionally or physically distributed or integrated in arbitrary units depending on various loads and usage conditions. Can be integrated and configured. Further, each processing function performed by each device may be realized in whole or in part by a CPU and a program that is analyzed and executed by the CPU, or may be realized as hardware using wired logic.

Furthermore, among the processes described in the embodiments and modifications described above, all or part of the processes described as being performed automatically can be performed manually, or may be performed manually. All or part of the described processing can also be performed automatically using known methods. In addition, information including processing procedures, control procedures, specific names, and various data and parameters shown in the above documents and drawings may be changed arbitrarily, unless otherwise specified.

According to at least one embodiment described above, the accuracy of extracting anatomical structures from medical images can be improved.

Although several embodiments have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, substitutions, and changes can be made without departing from the gist of the invention. These embodiments and their modifications are included within the scope and gist of the invention as well as within the scope of the invention described in the claims and its equivalents.

100,200 Medical image processing device 150,250 Processing circuit 151 Acquisition function 152,252 Specific function 153,253 Setting function 154 Classification function 155 Extraction function 256 Noise removal function

Claims (14)

an acquisition unit that acquires medical images;
By randomly setting sampling points for a region including a specific anatomical structure in the medical image, a periodic error that occurs when noise is mixed in the medical image is reduced. a setting section for setting sampling points arranged in a specific manner ;
a classification unit that classifies sampling points corresponding to the anatomical structure from among the sampling points;
A medical image processing apparatus, comprising: an extraction unit that extracts the anatomical structure from the medical image using classified sampling points. further comprising a noise removal unit that performs noise removal processing on the medical image,
The setting unit converts the medical image subjected to the noise removal process into a plurality of signals having different frequencies, and is arranged at intervals corresponding to a period of a predetermined number of signals having large amplitudes among the plurality of signals. adjusting the number of randomly set sampling points such that the number of sampling points is less than the number of sampling points arranged at intervals corresponding to other periods;
The medical image processing device according to claim 1 . further comprising a noise removal unit that performs noise removal processing on the medical image,
The setting unit converts the medical image subjected to the noise removal processing into a plurality of signals having different frequencies, and among the plurality of signals, the setting unit converts the medical image subjected to the noise removal processing to correspond to the period of the signal to be emphasized according to the method of the noise removal processing. adjusting the number of randomly set sampling points such that the number of sampling points arranged at intervals corresponding to other periods is less than the number of sampling points arranged at intervals corresponding to other periods;
The medical image processing device according to claim 1 . an acquisition unit that acquires medical images;
a noise removal unit that performs noise removal processing on the medical image;
a setting unit that randomly sets sampling points for a region including a specific anatomical structure in the medical image;
a classification unit that classifies sampling points corresponding to the anatomical structure from among the sampling points;
an extraction unit that extracts the anatomical structure from the medical image using the classified sampling points;
Equipped with
The setting unit converts the medical image subjected to the noise removal process into a plurality of signals having different frequencies, and is arranged at intervals corresponding to a period of a predetermined number of signals having large amplitudes among the plurality of signals. adjusting the number of randomly set sampling points such that the number of sampling points is less than the number of sampling points arranged at intervals corresponding to other periods;
Medical image processing device. an acquisition unit that acquires medical images;
a noise removal unit that performs noise removal processing on the medical image;
a setting unit that randomly sets sampling points for a region including a specific anatomical structure in the medical image;
a classification unit that classifies sampling points corresponding to the anatomical structure from among the sampling points;
an extraction unit that extracts the anatomical structure from the medical image using the classified sampling points;
Equipped with
The setting unit converts the medical image subjected to the noise removal processing into a plurality of signals having different frequencies, and among the plurality of signals, the setting unit converts the medical image subjected to the noise removal processing to correspond to the period of the signal to be emphasized according to the method of the noise removal processing. adjusting the number of randomly set sampling points such that the number of sampling points arranged at intervals corresponding to other periods is less than the number of sampling points arranged at intervals corresponding to other periods;
Medical image processing device. further comprising a specifying unit that specifies a region including the anatomical structure on the medical image;
The setting unit randomly sets sampling points for a region including the anatomical structure identified on the medical image.
The medical image processing device according to any one of claims 1 to 5 . the anatomical structure is a heart valve;
The medical image processing device according to any one of claims 1 to 6 . The classification unit determines, for each sampling point set for a region including the anatomical structure, whether a pixel at a position in the medical image at which the sampling point is set corresponds to the anatomical structure. classifying sampling points corresponding to the anatomical structure;
The medical image processing device according to any one of claims 1 to 7 . The setting unit roughly sets the sampling points in a region where the anatomical structures are not dense, and sets the sampling points densely in a region where the anatomical structures are dense.
The medical image processing device according to any one of claims 1 to 8 . The setting unit derives a plurality of sampling results by performing a process of randomly setting sampling points for a region including the anatomical structure multiple times,
The classification unit derives a plurality of classification results by classifying sampling points corresponding to the anatomical structure for each of the plurality of sampling results,
The extraction unit extracts the anatomical structure from the medical image based on the plurality of classification results.
A medical image processing apparatus according to any one of claims 1 to 9 . The medical image is an ultrasound image obtained by an ultrasound diagnostic device,
The medical image processing device according to any one of claims 1 to 10 . Acquire medical images,
By randomly setting sampling points for a region including a specific anatomical structure in the medical image, a periodic error that occurs when noise is mixed in the medical image is reduced. Set the sampling points located according to the
classifying sampling points corresponding to the anatomical structure from among the sampling points;
extracting the anatomical structure from the medical image using the classified sampling points;
A medical image processing method comprising: Acquire medical images,
Performing noise removal processing on the medical image,
Randomly setting sampling points for a region including a specific anatomical structure in the medical image,
classifying sampling points corresponding to the anatomical structure from among the sampling points;
extracting the anatomical structure from the medical image using the classified sampling points;
including that
Setting the sampling points involves converting the medical image subjected to the noise removal process into a plurality of signals having different frequencies, and setting an interval corresponding to the cycle of a predetermined number of signals with large amplitudes among the plurality of signals. adjusting the number of randomly set sampling points such that the number of sampling points arranged at intervals is less than the number of sampling points arranged at intervals corresponding to other periods;
Medical image processing method. Acquire medical images,
Performing noise removal processing on the medical image,
Randomly setting sampling points for a region including a specific anatomical structure in the medical image,
classifying sampling points corresponding to the anatomical structure from among the sampling points;
extracting the anatomical structure from the medical image using the classified sampling points;
including that
Setting the sampling points involves converting the medical image subjected to the noise removal processing into a plurality of signals having different frequencies, and selecting a signal that is emphasized according to the method of the noise removal processing among the plurality of signals. The number of randomly set sampling points is adjusted so that the number of sampling points arranged at intervals corresponding to the period of is less than the number of sampling points arranged at intervals corresponding to other periods. include,
Medical image processing method.

Images