JP6867117B2 - Medical image processing method and medical image processing device - Google Patents

Description

The present invention relates to a medical image processing method and a medical image processing apparatus.

Image diagnosis and image analysis occupy an important position in various fields of medicine. For example, in the field of ophthalmology, various imaging devices such as a slit lamp microscope, a fundus camera, a scanning laser ophthalmoscope (SLO), an optical coherence tomography (OCT), a laser speckle flowography (LSFG), and a surgical microscope are used. To. There are also imaging devices that can be used for multiple imaging techniques. For example, imaging techniques performed by a fundus camera include infrared observation, color imaging, fluorescein fluorescence imaging, indocyanine green fluorescence imaging, spontaneous fluorescence imaging, and red reflex imaging. In addition, imaging methods performed by OCT include morphological imaging such as B scan and volume scanning, angiography, blood flow measurement, and polarized imaging. Further, many ophthalmic devices (inspection devices) other than the photographing device are also provided with an infrared moving image photographing function for observing the eye to be inspected.

In addition to the field of ophthalmology, X-ray imaging equipment, X-ray computed tomography (CT) equipment, magnetic resonance imaging (MRI) equipment, positron emission tomography (PET) equipment, single photon emission tomography (SPECT) equipment , Various imaging devices such as endoscopes are used.

The image acquired by the imaging device is used for screening and analysis for lesion identification. Further, in follow-up observation and preoperative and postoperative observation, comparative observation and comparative analysis of a plurality of images acquired at different timings are performed.

In addition, applications of technologies such as artificial intelligence and non-linear signal processing to image processing are being promoted. For example, techniques for converting the resolution of an image are known.

Japanese Unexamined Patent Publication No. 8-263649 Japanese Unexamined Patent Publication No. 2006-350498 Japanese Unexamined Patent Publication No. 2010-520521 Japanese Unexamined Patent Publication No. 2014-225917 Japanese Unexamined Patent Publication No. 2015-50585

The development of medical imaging technology is very fast. For example, the scan speed and resolution of OCT have made remarkable progress in the last decade or so. On the other hand, some diseases, such as glaucoma, require decades of follow-up. In such a case, the quality (resolution, etc.) of the plurality of images to be compared may be significantly different, and comparative observation and comparative analysis may not be performed appropriately. Similarly, the same problem arises when comparing a low-resolution image obtained by a simple imaging device in an emergency or a home visit with a high-resolution image obtained by a high-performance imaging device.

In addition, when only a simple imaging device can be used, such as in an emergency or disaster, it is necessary to make a diagnosis by referring to a low-resolution image (only), but improving the accuracy of such a diagnosis is not possible. It is difficult with conventional technology.

In addition, the resolution may decrease due to the influence of the communication environment and the like. For example, in telemedicine, live images and captured images are transmitted to remote locations, so a decrease in resolution limits the recognition and judgment of doctors.

An object of the present invention is to solve a problem caused by a low resolution of a medical image.

One aspect of the medical image processing method according to the exemplary embodiment is a method of processing a medical image using a computer. The computer includes an artificial intelligence engine that has learned multiple combinations of low resolution images and high resolution images. The computer receives an image of the eye to be inspected having a predetermined resolution, converts this image into a first image having a first resolution higher than the predetermined resolution by an artificial intelligence engine, and executes a predetermined process based on the first image. To do. Artificial intelligence engines include ophthalmic artificial intelligence engines with an ophthalmic knowledge base that includes feature information that represents the anatomical features of the eye. The artificial intelligence engine for ophthalmology converts the image of the eye to be inspected into the first image at least based on the feature information.
One aspect of the medical image processing apparatus according to the exemplary embodiment includes an image receiving unit, an image conversion unit, and a processing unit. The image receiving unit receives an image of the eye to be inspected having a predetermined resolution. The image conversion unit includes an artificial intelligence engine that has learned a plurality of combinations of a low-resolution image and a high-resolution image, and converts an image received by the image reception unit into a first image having a first resolution higher than a predetermined resolution. Convert. The processing unit executes a predetermined process based on the first image generated by the image conversion unit. Artificial intelligence engines include ophthalmic artificial intelligence engines with an ophthalmic knowledge base that includes feature information that represents the anatomical features of the eye. The artificial intelligence engine for ophthalmology converts the image of the eye to be inspected into the first image at least based on the feature information.

According to the exemplary embodiment, the problem caused by the low resolution of the medical image can be solved.

It is a flowchart which shows the exemplary medical image processing method. It is a flowchart which shows the exemplary medical image processing method. It is a flowchart which shows the exemplary medical image processing method. It is a flowchart which shows the exemplary medical image processing method. It is a flowchart which shows the exemplary medical image processing method. It is a flowchart which shows the exemplary medical image processing method. It is the schematic which shows the structure of the exemplary medical image processing apparatus.

An exemplary embodiment of the present invention will be described in detail with reference to the drawings. An exemplary medical image processing method can be realized by an exemplary medical image processing apparatus. The medical image processing device may include, for example, two or more devices (for example, one or more computers, one or more storage devices, etc.) capable of communicating with each other. Alternatively, the medical image processing device may be a single device (eg, a computer equipped with a storage device).

The hardware and software for realizing the medical image processing method are not limited to the medical image processing devices exemplified below, and may include a combination of any hardware and any software that contributes to the realization. .. As a typical example, a medical image processor may include hardware and software that act as an artificial intelligence engine.

The medical image processed according to the embodiment may include an image obtained by photographing the eye to be inspected using an arbitrary modality (imaging device, imaging method). For example, the medical image processed according to the embodiment may include an image acquired by photographing the eye to be inspected by an ophthalmologic imaging device such as a fundus camera, SLO, or OCT. Further, the medical image processed according to the embodiment may include an image created by the medical image processing method (the medical image processing apparatus). For example, the medical image processed according to the embodiment may include an image created by processing the image acquired by the ophthalmologic imaging apparatus by the medical image processing apparatus. Further, the medical image processed by the embodiment may include an image created by another medical image processing device. For example, the medical image processed according to the embodiment may include an image created by processing an image acquired by an ophthalmologic imaging apparatus by another medical image processing apparatus.

The medical image is an image of a predetermined part of the subject. The medical image may be a still image or a moving image. The image of the predetermined portion may include at least an image representing at least a part of the predetermined portion, and may further include an image representing at least a part of another portion (for example, a portion in the vicinity of the predetermined portion). For example, when the predetermined site is the retina (or the tissue constituting the retina), the medical image may include only the image of the retina, a choroid, a sclera, a vitreous body, a lamina cribrosa, etc. located in the vicinity of the retina. The image of may be further included. Further, the medical image may be an image showing the structure (morphology) of a predetermined part or an image showing the function thereof. Examples of functional images include images showing blood flow dynamics (blood flow volume, blood flow velocity, etc.) such as OCT blood flow images, fMRI images, and ultrasonic Doppler images.

Hereinafter, an example in which the medical image processing method is applied to an ophthalmology will be mainly described, but the clinical department to which this can be applied is not limited to the ophthalmology, and can be applied to any other clinical department. For example, typical examples of medical images to which the embodiment can be applied include X-ray images, CT images, MRI images, PET images, SPECT images, endoscopic images, gastrocamera images, colon camera images, ultrasonic images, and the like. There is.

<Example of medical image processing method>
An example of applying the medical image processing method to ophthalmology will be described. An example of a medical image processing apparatus for realizing a medical image processing method will be described later.

The processing related to the medical image processing method is executed by a computer. The computer may include an artificial intelligence engine. The computer includes one or more processors. The processor is, for example, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), an ASIC (Application Specific Integrated Circuit), a programmable logic device (for example, a SPILD (Field Programmable Gate Array)) and the like may be included. The processor realizes a desired function by reading and executing a program stored in a storage device (storage circuit), for example.

The processor can control the storage device and the display device. The storage device is configured to be able to store programs, data, information, and the like used by the medical image processing method, and may be arranged inside the computer or outside the computer. The same applies to the arrangement of output devices.

An output device is a device that outputs information, and typical examples thereof include a display device, an audio output device, a printing device, a communication device having a data transmission function, and a data writer that records information on a recording medium. is there. At least one such example is used as an output device.

An example of a medical image processing method is shown in FIG. Subject registration can be performed before the series of processes described below, during the execution of the series of processes, or after the series of processes. The subject means a person who has taken a medical image processed by a medical image processing method. The examinee may include not only a person who has undergone a medical examination but also a person who has not undergone a medical examination. Those who have already consulted a medical institution include, for example, those who have received a definitive diagnosis by a doctor and those who have received a diagnosis by a doctor but have not yet reached a definitive diagnosis. Those who have not undergone a medical institution include, for example, those who have undergone a medical examination and those who have undergone a screening.

In addition, a step of acquiring medical information about the subject may be included before the series of processes described below, during the execution of the series of processes, or after the series of processes. Medical information includes, for example, information that can be acquired, created, and used in medical practice such as medical examination, diagnosis, and treatment. Typical examples of medical information include electronic medical records, personal health records (PHR), surgical records, and examination data.

The processing that can be additionally executed in the medical image processing method is not limited to the registration of the subject and the acquisition of medical information, and the desired processing other than these can be additionally executed at a desired timing.

Some examples of typical medical image processing methods will be described. The following examples can be applied respectively. Alternatively, any two or more of the following examples can be combined. It is also possible to modify any of the steps in the example below.

<First example of medical image processing method>
A first example of the medical image processing method will be described with reference to FIG. In this example, the basic medical image processing method according to the embodiment will be described.

(S1: Accept medical images)
First, the computer receives an image of a predetermined part of the subject having a predetermined resolution. The predetermined resolution may be a relatively low resolution. For example, this low-resolution image may be an image acquired by a low-resolution imaging device used in the past, an image obtained by a simple imaging device in an emergency or a home visit, or a remote image in telemedicine or the like. It may be an image transmitted from the ground.

The computer receives input of a medical image, for example, via a communication line or a recording medium. For example, when the computer includes a communication device, the computer can acquire a medical image stored in an image archiving device (PACS, cloud image server, etc.) through a communication line. Alternatively, if the computer includes a data reader (such as a drive device), the computer can read the medical image recorded on the recording medium. The configuration for receiving medical images (or medical information, etc.) is not limited to these.

(S2: Convert resolution)
Next, the computer converts the low-resolution image received in step S1 into a first image having a first resolution higher than that resolution. That is, the computer increases the resolution of the medical image.

The first resolution can be set arbitrarily. For example, the first resolution may be set as a default value, or may be set according to the subsequent processing. As an example of the latter, when the comparison of medical images is performed later, the resolution of the (high resolution) image to be compared with the low resolution image can be set to the first resolution.

Resolution conversion can be done by an artificial intelligence engine. The artificial intelligence engine may be, for example, one or more combinations of the processes described in any of the above cited references, one or more combinations of the processes described in other publicly known documents, or a combination thereof. The resolution is converted by executing any combination of two or more.

Examples of artificial intelligence technologies that can be used in artificial intelligence engines include neural networks, deep learning, support vector machines, Bayesian networks, correlation rule learning, reinforcement learning, expression learning, data mining, natural language processing, and inference. Further, the resolution may be converted by referring to a knowledge base (medical knowledge base, ophthalmic knowledge base, etc.).

As a typical example, a plurality of combinations (learning samples) of a low-resolution image and a high-resolution image can be learned, and inference based on the combination (learning sample) can be performed to perform resolution conversion. For example, a low-resolution image and a high-resolution image can be continuously acquired using a fundus camera or OCT, and the correlation, association, and law of the combination of the low-resolution image and the high-resolution image can be learned.

It is also possible to acquire only a high-resolution image, reduce the resolution to create a low-resolution image, and learn a combination thereof.

In addition, only low-resolution images are acquired, a plurality of high-resolution images are created from the images, and an appropriate one of these high-resolution images is selected by a doctor or an artificial intelligence engine to obtain a low-resolution image and a high-resolution image. The combination may be determined.

As other typical examples, super-resolution technology including processing to increase the number of pixels, super-resolution technology including lifting of high-frequency components, reconstruction-type super-resolution technology combining these, and estimation of lost high-frequency components. Super-resolution technology represented by super-resolution technology, multi-frame super-resolution technology, in-frame super-resolution technology, super-resolution technology including non-linear signal processing, and the like may be applied. Further, a known up-conversion technique may be applied.

(S3: Execute a predetermined process)
The computer executes a predetermined process based on the first image whose resolution has been increased in step S2.

The predetermined process may be an arbitrary process. As a typical example of a predetermined process, a process for displaying two or more images of a predetermined part of a subject side by side, a process for analyzing two or more images, and two or more analysis data obtained by the process. The process for displaying the images side by side, the process for obtaining the time-series change of the characteristics of a predetermined site from two or more analysis data, and the process for inferring the disease suffering by the subject and its candidate will be described later. ..

The low-resolution image received in step S1, the high-resolution image (first image) generated in step S2, the information obtained by the predetermined processing, and the like are stored in the storage device and / or output by the output device. Will be done.

<Second example of medical image processing method>
A second example of the medical image processing method will be described with reference to FIG. In this example, a medical image processing method for displaying N images of a predetermined part of a subject side by side will be described (N is an integer of 2 or more).

(S11: Accept medical images)
First, in the same manner as in step S1 in the first example, the computer receives an image (low resolution image) of a predetermined portion of the subject having a predetermined resolution.

(S12: Convert to the first image with the first resolution)
Next, in the same manner as in step S2 in the first example, the computer generates a first image having a first resolution higher than a predetermined resolution by converting the resolution of the image acquired in step S11.

(S13: Acquire the second to Nth images of the second resolution)
The computer acquires a second image representing the same portion as the first image, which has a second resolution higher than the resolution of the low resolution image acquired in step S11.

More generally, the computer acquires one or more images representing the same parts as the first image, each having a higher resolution than the resolution of the low resolution image acquired in step S11. The number of images acquired here is arbitrary. Further, when two or more images are acquired here, the resolutions of these images may be the same or different. That is, the second resolution may be a single resolution higher than the predetermined resolution, or may be two or more resolutions higher than the predetermined resolution.

The second resolution may be any resolution higher than the resolution of the low resolution image acquired in step S11. For example, the second resolution may be equal to the first resolution. Alternatively, the second resolution may be approximately equal to the first resolution. The term "nearly" equal to the first resolution and the second resolution means, for example, to allow a difference in resolution to the extent that observation and analysis can be performed with the same accuracy and / or accuracy, or the first resolution and the second resolution. If they are different, it means that observation and analysis can be performed with sufficient accuracy and / or accuracy even at the lower resolution.

The timing of executing the process of this step is not limited to the timing, and may be arbitrary. For example, by accepting a plurality of images in step S11 and applying the high resolution of step S12 to at least a part of these images, a high resolution image group can be acquired.

Further, the process of this step includes a process of accepting an image acquired by a high-resolution photographing device, a process of accepting an image whose resolution has been increased in the past, and / or a process of reading such an image from a storage device. It may be included. The timing of performing these processes may be before step S11, during execution of step S11, between step S11, step S11 and step S12, during execution of step S12, or after step S12.

(S14: Display a plurality of images side by side in the order according to the shooting date)
The computer sets at least a part (plurality of images) of the first image obtained in step S12 and the N images including the second to Nth images obtained in step S13 according to their shooting dates. Display them on the display device in order. At this time, it is possible to display the shooting dates of a plurality of images.

The shooting date of the image is the date (and time, etc.) at which the image was acquired by photographing the subject using the modalidi. The shooting date is, for example, transmitted and received together with the image in a state associated with the image. As a typical example, when Digital Imaging and Communications in Medicine (DICOM) is used as a medical image format and communication protocol, the shooting date can be recorded in the tag information in the DICOM file. In addition, information indicating the shooting date can be embedded in the background area of the image or the like.

The computer, for example, obtains the shooting dates of each of the plurality of images, assigns them a time-series order by comparing the acquired shooting dates, and arranges the plurality of images according to the time-series order. be able to. When two or more images having the same shooting date exist, the time series order can be determined by referring to, for example, the shooting time.

By displaying a plurality of images arranged in chronological order in this way, the user can easily grasp the time-series changes in the structure and function of a predetermined portion of the subject. Further, when only a part of N images is displayed, the displayed images can be specified by using an operation device. In addition, a function for enlarging any one of the plurality of images can be provided.

<Third example of medical image processing method>
A third example of the medical image processing method will be described with reference to FIG. In this example, a medical image processing method for acquiring N analysis data of a predetermined part of a subject, and a medical image processing method for displaying the analysis data side by side will be described (N is 2 or more). Let it be an integer of).

(S21: Accepts medical images)
First, in the same manner as in step S1 in the first example, the computer receives an image (low resolution image) of a predetermined portion of the subject having a predetermined resolution.

(S22: Convert to the first image with the first resolution)
Next, in the same manner as in step S2 in the first example, the computer generates a first image having a first resolution higher than a predetermined resolution by converting the resolution of the image acquired in step S21.

(S23: Acquire the second to Nth images of the second resolution)
The computer acquires a second image representing the same portion as the first image, which has a second resolution higher than the resolution of the low resolution image acquired in step S21. More generally, the computer acquires one or more images representing the same parts as the first image, each having a higher resolution than the resolution of the low resolution image acquired in step S21. The number of images acquired here is arbitrary.

The second resolution may be similar to that in the second example described above. Further, the processing of this step and the execution timing thereof may be the same as those in the second example.

(S24: Analyzing the 1st to Nth images)
The computer performs a predetermined analysis process on each of the first image obtained in step S22 and the second to Nth images obtained in step S23. Typical examples of the analysis process are papillary three-dimensional shape analysis, retinal nerve fiber layer (RNFL) thickness analysis, retinal pigment epithelial (RPE) layer thickness analysis, and segment analysis (intravisual cells) for images acquired by fundus OCT. Contact / outer segment (IS / OS) lines, cone outer segment tip (COST), outer nuclear layer, external limiting membrane, choroid), retinal analysis, etc. As a result, the first analysis data based on the first image, the second analysis data based on the second image, ..., The Nth analysis data based on the Nth image can be obtained.

(S25: Display multiple analysis data side by side in the order according to the shooting date)
The computer displays at least a part (plurality of analysis data) of the N analysis data generated in step S24 on the display device in an order according to the shooting date of the corresponding plurality of images. At this time, it is possible to display the shooting dates of a plurality of images. The shooting date and the processing related thereto may be the same as in the second example described above.

By displaying a plurality of analysis data arranged in chronological order in this way, the user can easily grasp the time-series change of the analysis data of the predetermined portion of the subject. Further, when only a part of N analysis data is displayed, it can be configured so that the displayed analysis data can be specified by using the operation device. In addition, a function for enlarging any one of the plurality of analysis data can be provided.

<Fourth example of medical image processing method>
A fourth example of the medical image processing method will be described with reference to FIG. In this example, a medical image processing method for acquiring N analysis data of a predetermined part of a subject, and further, a medical image for generating information representing a time-series change in the characteristics of a predetermined part from the analysis data. The image processing method will be described (N is an integer of 2 or more).

(S31: Accept medical images)
First, in the same manner as in step S1 in the first example, the computer receives an image (low resolution image) of a predetermined portion of the subject having a predetermined resolution.

(S32: Convert to the first image with the first resolution)
Next, in the same manner as in step S2 in the first example, the computer generates a first image having a first resolution higher than a predetermined resolution by converting the resolution of the image acquired in step S31.

(S33: Acquire the second to Nth images of the second resolution)
The computer acquires a second image representing the same portion as the first image, which has a second resolution higher than the resolution of the low resolution image acquired in step S31. More generally, the computer acquires one or more images representing the same parts as the first image, each having a higher resolution than the resolution of the low resolution image acquired in step S31. The number of images acquired here is arbitrary.

The second resolution may be similar to that in the second example described above. Further, the processing of this step and the execution timing thereof may be the same as those in the second example.

(S34: Analyzing the 1st to Nth images)
The computer performs a predetermined analysis process on each of the first image obtained in step S32 and the second to Nth images obtained in step S33. This analysis process may be similar to that in the third example described above.

(S35: Generates information representing time-series changes in characteristics)
The computer changes the characteristics of the predetermined part of the subject over time based on at least a part (plurality of analysis data) of the N analysis data generated in step S24 and the shooting dates of these plurality of images. Generates information that represents. Typical examples of this information include color maps and graphs showing time-series changes in the three-dimensional shape of the teat, and color maps and graphs showing time-series changes in RNFL thickness. The shooting date and the processing related thereto may be the same as in the second example described above. In addition, the computer can output the generated information to the output device.

By generating such information, the user can easily grasp the time-series changes in the characteristics (structure, function, distribution, value, etc.) of the predetermined portion of the subject. Further, when information based on only a part of N analysis data is generated, it can be configured so that the analysis data used for generating the information can be specified by using the operation device. That is, the period and interval of time series change can be changed arbitrarily.

<Fifth example of medical image processing method>
A fifth example of the medical image processing method will be described with reference to FIG. In this example, a medical image processing method for making inferences about a disease suffering from a subject or a candidate thereof will be described. Here, the disease affecting the subject may include a disease for which a definitive diagnosis has already been made. In addition, the candidate for the disease affecting the subject may include a disease for which a definitive diagnosis has not yet been made and a disease for which a medical examination, a screening, a screening, etc. are to be performed.

(S41: Accepts medical images)
First, in the same manner as in step S1 in the first example, the computer receives an image (low resolution image) of a predetermined portion of the subject having a predetermined resolution.

(S42: Convert resolution)
Next, in the same manner as in step S2 in the first example, the computer generates a first image having a first resolution higher than a predetermined resolution by converting the resolution of the image acquired in step S41.

(S43: Acquire medical information)
The computer obtains medical information about the subject. The medical information may include any information that can be acquired, created, and used in medical practice, such as electronic medical records, personal health records (PHR), surgical records, and examination data.

The computer can access the medical information database via, for example, a communication line or a recording medium, and acquire medical information of the subject. Alternatively, the computer can also generate medical information. For example, the above-mentioned analysis data can be used as medical information. The timing of executing the process of this step is not limited to the timing, and may be arbitrary.

(S44: Make inferences about the disease)
Based on the first image obtained in step S42 and the medical information obtained in step S43, the computer makes an inference about the disease suffering by the subject or a candidate thereof. Inference is performed by an artificial intelligence engine.

Inference can be, for example, determining whether or not a person has a predetermined disease, calculating the probability of having a predetermined disease, identifying the degree of progression of the predetermined disease, specifying the type of necessary test, and the like. May include. Learning according to the content of inference is performed in advance for the artificial intelligence engine. By making inferences about such diseases, diagnostic support becomes possible.

<Sixth example of medical image processing method>
A sixth example of the medical image processing method will be described with reference to FIG. In this example, application to the field of ophthalmology will be described.

The computer of this example may include an artificial intelligence engine (artificial intelligence engine for ophthalmology) that has been trained in ophthalmology. The artificial intelligence engine for ophthalmology includes a knowledge base about ophthalmology. The ophthalmic knowledge base may include any specialized books or treatises on ophthalmic diseases, specialized books or treatises on eye structure (anatomy), specialized books or treatises on eye images or examinations, and the like.

(S51: Accepts the image of the eye to be inspected)
First, in the same manner as in step S1 in the first example, the computer receives an image (low resolution image) of a predetermined portion of the eye to be inspected having a predetermined resolution.

(S52: Determine whether it is a left eye image or a right eye image)
The computer determines whether the image acquired in step S51 is a left eye image or a right eye image.

This determination is performed by an ophthalmic artificial intelligence engine or other processor. This determination is made, for example, with reference to the anatomical features of the eye. Typically, is the image acquired in step S51 a left eye image based on the position of the optic disc, the position of the macula, the relative position of the optic disc and the macula, the running of blood vessels, the running of nerve fibers, etc. It is possible to determine whether it is a right eye image. Alternatively, for example, in the tag information in the DICOM file, information indicating whether the image is a left-eye image or a right-eye image may be recorded.

The timing of executing this left-right discrimination may be before or after increasing the resolution of the image. It is also possible to execute the determination in parallel with increasing the resolution.

(S53: Convert resolution)
Next, in the same manner as in step S2 in the first example, the computer generates a first image having a first resolution higher than a predetermined resolution by converting the resolution of the image acquired in step S51.

This process is performed, for example, by an ophthalmic artificial intelligence engine. For example, if the ophthalmic knowledge base contains characteristic information representing the anatomical features of the eye (position and morphology of the optic nerve head, position and morphology of the macula, running and morphology of blood vessels, running and morphology of nerve fibers, etc.), for ophthalmology The artificial intelligence engine can convert the resolution at least based on the feature information.

The feature information may include left eye information representing the anatomical features of the left eye and right eye information representing the anatomical features of the right eye. In this case, the left-right discrimination (step S52) can be executed before the resolution is increased, and the feature information can be selectively used according to the determination result. Specifically, when it is determined that the image acquired in step S51 is a left eye image, the artificial intelligence engine for ophthalmology generates the first image by increasing the resolution at least based on the left eye information. can do. On the other hand, when the image acquired in step S51 is determined to be the right eye image, the artificial intelligence engine for ophthalmology may generate the first image by increasing the resolution at least based on the right eye information. it can.

(S54: Execute a predetermined process)
The computer executes a predetermined process based on the first image whose resolution has been increased in step S2. The predetermined process may be an arbitrary process as described above.

The low-resolution image received in step S51, the discrimination result obtained in step S52, the high-resolution image (first image) generated in step S53, the information obtained by a predetermined process, and the like are stored in the storage device. , And / or output by the output device.

<Example of medical image processing device>
An example of a medical image processing apparatus for realizing the above-mentioned medical image processing method will be described. An example of the configuration of such a device is shown in FIG.

The medical image processing device 1 includes a control unit 10, a storage unit 20, a data processing unit 30, and a communication unit 40. The user interface (UI) 100 may or may not be included in the medical image processing apparatus 1. The elements included in the medical image processing device 1 are configured as a single device or two or more devices. For example, the medical image processing device 1 includes a single computer equipped with all of these elements.

As an example of the case where the medical image processing device 1 includes two or more devices, a computer including a control unit 10, a computer including a storage unit 20, and a computer including a data processing unit 30 can be individually provided. .. Alternatively, the medical image processing device 1 includes a computer including any two of the control unit 10, the storage unit 20, and the data processing unit 30, and a computer including the other one. Modes of communication between different computers may include wired and / or wireless communications, including leased and / or public lines, local area networks (LANs), wide area networks (WANs), short-range communications. And at least one of the Internet may be included.

<Control unit 10>
The control unit 10 executes various controls. For example, the control unit 10 executes control of each element of the medical image processing device 1 and coordinated control of two or more elements. In addition, the control unit 10 can control an external device of the medical image processing device 1. For example, when the user interface 100 is not included in the medical image processing device 1, the control unit 10 can control the user interface 100. The control unit 10 includes a processor.

<Output control unit 11>
The output control unit 11 controls the output device described above. In this example, the output control unit 11 can execute step S3 of FIG. 1, step S14 of FIG. 2, step S25 of FIG. 3, step S54 of FIG. 6, and the like.

The output device includes, for example, at least one of a display device, an audio output device, a printing device, a communication device, and a data writer. That is, the output control unit 11 is configured to execute at least one of the following controls: control of the display device for displaying information; control of the voice output device for outputting voice information; Control of a printing device for printing information on paper; control of a communication device for transmitting information to an external device; control of a data writer for recording information on a recording medium.

In the example shown in FIG. 2, a display unit 101 is provided as a display device. Further, the communication unit 40 can be used as a communication device. Although not shown, one or more of an audio output device, a printing device, and a data writer may be provided. In addition, other output devices may be provided.

<Memory unit 20>
The storage unit 20 stores various types of data. As an example of the data stored in the storage unit 20, there is subject information such as the subject name and the subject ID. The storage unit 20 includes, for example, at least one of a semiconductor storage device, a magnetic storage device, an optical storage device, and a photomagnetic storage device.

<Knowledge Base 21>
The knowledge base 21 is stored in the storage unit 20. The knowledge base 21 may include a medical knowledge base such as an ophthalmic knowledge base, and a knowledge base for natural language analysis and image analysis. In this example, as described above, at least a part of the knowledge base 21 can be created by learning a plurality of combinations (learning samples) of a low-resolution image and a high-resolution image.

<Communication unit 40>
The communication unit 40 performs a process of transmitting data to another device (external device 200) and a process of receiving data from the external device 200. The communication unit 40 includes a known communication device according to the communication method with the external device 200.

The external device 200 may include a computer (server, database, etc.) or a medical device (inspection device, imaging device, etc.).

When the external device 200 is an imaging device, an image archiving device, or the like, the communication unit 40 can function as an image receiving unit that receives a medical image provided by the external device 200. In this example, the communication unit 40 is controlled by, for example, the control unit 10, step S1 in FIG. 1, step S11 in FIG. 2, step S21 in FIG. 3, step S31 in FIG. 4, and step S41 in FIG. Step S51 of FIG. 6 and the like can be executed. Further, it is also possible to execute step S13 of FIG. 2, step 23 of FIG. 3, step S33 of FIG. 4, and the like.

When the external device 200 is a device that creates or stores the above-mentioned medical information, the communication unit 40 can function as a medical information acquisition unit that acquires medical information from the external device 200. In this example, the communication unit 40 can execute step S43 of FIG. 5 or the like under the control of the control unit 10, for example.

<Data processing unit 30>
The data processing unit 30 executes various data processing. In this example, the data processing unit 30 can perform image resolution conversion, left-eye image / right-eye image discrimination, image analysis, inference, and the like. The data processing unit 30 includes a processor and a computer program. The data processing unit 30 may include an artificial intelligence engine (artificial intelligence engine for ophthalmology, etc.). The knowledge base 21 is used by an artificial intelligence engine.

The image resolution conversion corresponds to step S2 in FIG. 1, step S12 in FIG. 2, step S22 in FIG. 3, step S32 in FIG. 4, step S42 in FIG. 5, step S53 in FIG. 6, and the like. The image resolution conversion is executed by the image conversion unit 31. The image conversion unit 31 typically includes an artificial intelligence engine, but it may not be included. The operation of the image conversion unit 31 is realized by the cooperation between the hardware (processor, storage device, etc.) for resolution conversion and the computer program.

The discrimination between the left eye image and the right eye image corresponds to step S52 and the like in FIG. The discrimination between the left eye image and the right eye image is executed by the left / right discriminating unit 32. The left / right discriminating unit 32 may or may not include an artificial intelligence engine. The operation of the left / right discriminating unit 32 is realized by the cooperation between the hardware (processor, storage device, etc.) for discriminating the left eye image / right eye image and the computer program.

The image analysis corresponds to step S24 in FIG. 3, step S34 in FIG. 4, and the like. The image analysis is executed by the image analysis unit 33. The image analysis unit 33 may or may not include an artificial intelligence engine. The operation of the image analysis unit 33 is realized by the cooperation between the hardware (processor, storage device, etc.) for image analysis and the computer program.

The inference corresponds to step S44 and the like in FIG. The inference is executed by the inference unit 34. The inference unit 34 typically includes an artificial intelligence engine, but it may not be included. The operation of the inference unit 34 is realized by the cooperation between the hardware for inference (processor, storage device, etc.) and the computer program.

<User interface 100>
The user interface 100 includes a display unit 101 and an operation unit 102. The display unit 101 includes a display device such as a flat panel display. The operation unit 102 includes operation devices such as a mouse, a keyboard, a trackpad, buttons, keys, a joystick, and an operation panel.

The display unit 101 and the operation unit 102 do not need to be configured as separate devices. For example, it is possible to use a device such as a touch panel in which a display function and an operation function are integrated. In that case, the operation unit 102 includes the touch panel and a computer program. The operation content for the operation unit 102 is input to the control unit 10 as an electric signal. Further, the graphical user interface (GUI) displayed on the display unit 101 and the operation unit 102 may be used to perform operations and information input.

<Operation example of medical image processing device>
Some examples of operations that the medical image processing apparatus 1 can perform will be described. The following operation examples can be applied respectively. Alternatively, any two or more of the following operation examples can be combined. It is also possible to modify any of the steps in the following operation example.

<First operation example>
The first operation example will be described. First, the communication unit 40 receives an image of a predetermined portion of the subject having a predetermined resolution (for example, step S1 in FIG. 1).

Next, the image conversion unit 31 (for example, an artificial intelligence engine (for ophthalmology)) converts the image received by the communication unit 40 into a first image having a first resolution higher than a predetermined resolution (for example, FIG. Step S2 of 1.

Subsequently, the data processing unit 30 (image analysis unit 33, inference unit 34, etc.) and / or the control unit 10 (output control unit 11 and the like) perform predetermined processing based on the first image generated by the image conversion unit 31. (For example, step S3 in FIG. 1).

<Second operation example>
A second operation example will be described. First, the communication unit 40 receives an image of a predetermined portion of the subject having a predetermined resolution (for example, step S11 in FIG. 2).

Next, the image conversion unit 31 (for example, an artificial intelligence engine (for ophthalmology)) converts the image received by the communication unit 40 into a first image having a first resolution higher than a predetermined resolution (for example, FIG. Step S12 of 2.

The communication unit 40 (and / or the data processing unit 30 or the like) acquires a second image of a predetermined portion having a second resolution higher than the predetermined resolution (for example, step S13 in FIG. 2). Alternatively, the communication unit 40 (and / or the data processing unit 30 or the like) acquires the second to Nth images of the predetermined portion having a second resolution higher than the predetermined resolution (for example, step S13 in FIG. 2).

The output control unit 11 arranges the first image having the first resolution and the second image (or the second to Nth images) having the second resolution in the order according to their shooting dates on the display unit 101. Display (for example, step S14 in FIG. 2). At this time, a plurality of images can be selected and displayed from the first to Nth images.

<Third operation example>
A third operation example will be described. First, the communication unit 40 receives an image of a predetermined portion of the subject having a predetermined resolution (for example, step S21 in FIG. 3).

Next, the image conversion unit 31 (for example, an artificial intelligence engine (for ophthalmology)) converts the image received by the communication unit 40 into a first image having a first resolution higher than a predetermined resolution (for example, FIG. Step 3 S22).

The communication unit 40 (and / or the data processing unit 30 or the like) acquires a second image (or a second to Nth image) of a predetermined portion having a second resolution higher than the predetermined resolution (for example, the step of FIG. 3). S23).

The image analysis unit 33 performs a predetermined analysis process on the first image to generate the first analysis data, and performs an analysis process on the second image to generate the second analysis data (for example, the step of FIG. 3). S24). Alternatively, the image analysis unit 33 performs a predetermined analysis process on each of the first to Nth images to generate a plurality of analysis data (for example, step S24 in FIG. 3).

The output control unit 11 displays the first analysis data and the second analysis data (or the first to Nth analysis data) on the display unit 101 in an order according to the shooting date of the images (for example, FIG. 3). Step S25). At this time, a plurality of analysis data can be selected and displayed from the first to Nth analysis data.

<Fourth operation example>
A fourth operation example will be described. First, the communication unit 40 receives an image of a predetermined portion of the subject having a predetermined resolution (for example, step S31 in FIG. 4).

Next, the image conversion unit 31 (for example, an artificial intelligence engine (for ophthalmology)) converts the image received by the communication unit 40 into a first image having a first resolution higher than a predetermined resolution (for example, FIG. Step S32 of step 4).

The communication unit 40 (and / or the data processing unit 30 or the like) acquires the second to Nth images of a predetermined portion having a second resolution higher than the predetermined resolution (for example, step S33 in FIG. 4).

The image analysis unit 33 performs a predetermined analysis process on each of the first to Nth images to generate the first to Nth analysis data (for example, step S34 in FIG. 4).

The image analysis unit 33 further generates information representing a time-series change in the characteristics of a predetermined part of the subject based on a part or all of the first to Nth analysis data (for example, step 35 in FIG. 4). ). The output control unit 11 is, for example, an output device (display unit) so as to output one or more of the generated information, any of the first to Nth images, and any one or more of the first to Nth analysis data. 101 etc.) can be controlled.

<Fifth operation example>
A fifth operation example will be described. First, the communication unit 40 receives an image of a predetermined portion of the subject having a predetermined resolution (for example, step S41 in FIG. 5).

Next, the image conversion unit 31 (for example, an artificial intelligence engine (for ophthalmology)) converts the image received by the communication unit 40 into a first image having a first resolution higher than a predetermined resolution (for example, FIG. Step S42 of step 5).

The communication unit 40 acquires medical information from the external device 200 (for example, step S43 in FIG. 5). In addition to or instead of this, the data processing unit 30 can generate medical information (eg, step S43 in FIG. 5).

The inference unit 34 makes an inference regarding the disease suffering from the subject or a candidate thereof based on the first image and medical information (for example, step S54 in FIG. 5). It is also possible to make an inference based on the first image without using medical information.

The output control unit 11 can control the output device (display unit 101, etc.) so as to output one or more of the first image and medical information as a result of inference, for example.

<Sixth operation example>
A sixth operation example will be described. First, the communication unit 40 receives an image of a predetermined portion of the eye to be inspected having a predetermined resolution (for example, step S51 in FIG. 6).

Next, the left-right discriminating unit 32 discriminates whether the received image is a left-eye image or a right-eye image (for example, step S52 in FIG. 6). The left / right discrimination process may be executed after the process by the image conversion unit 31.

The image conversion unit 31 (for example, an artificial intelligence engine (for ophthalmology)) converts the image received by the communication unit 40 into a first image having a first resolution higher than a predetermined resolution (for example, the step of FIG. 6). S53).

Here, when the knowledge base 21 (ophthalmology knowledge base) includes feature information representing anatomical features of the eye, the image conversion unit 31 (artificial intelligence engine for ophthalmology) performs resolution conversion based on at least the feature information. This makes it possible to generate the first image.

Further, when the feature information includes left eye information representing the anatomical features of the left eye and right eye information representing the anatomical features of the right eye, and the processing by the left / right discriminating unit 32 has already been executed. , The image conversion unit 31 (artificial intelligence engine for ophthalmology) generates a first image based on at least left eye information for the left eye image, and at least based on the right eye information for the right eye image. One image can be generated.

Subsequently, the data processing unit 30 (image analysis unit 33, inference unit 34, etc.) and / or the control unit 10 (output control unit 11 and the like) perform predetermined processing based on the first image generated by the image conversion unit 31. (For example, step S54 in FIG. 6).

<Action / effect>
The actions and effects of the embodiments exemplified above will be described. The actions and effects described below are exemplary, and any two or more of these can be arbitrarily combined, or any one or more of these can be combined with known art.

The medical image processing method of the embodiment is a method of processing a medical image using a computer. The computer receives an image of a predetermined part of the subject having a predetermined resolution, converts this image into a first image having a first resolution higher than the predetermined resolution, and performs a predetermined process based on the first image. Execute.

According to such a medical image processing method, when long-term follow-up is required such as glaucoma, it is possible to increase the resolution of a low-resolution image acquired in the past, for example. As a result, the problems of comparative observation and comparative analysis caused by the difference in quality (resolution) of a plurality of images to be compared are solved.

In addition, since it is possible to increase the resolution of a low-resolution image obtained by a simple photographing device in an emergency or a home visit, it is preferable to compare it with a high-resolution image obtained by a high-performance photographing device. be able to.

In addition, even when only a simple imaging device can be used, such as in an emergency or disaster, the low-resolution image can be made high-resolution and used for diagnosis, so it is possible to improve the accuracy of diagnosis. Become.

In addition, even if the resolution is lowered due to the influence of the communication environment in telemedicine, etc., the resolution of live images and captured images can be increased, so that the recognition and judgment of doctors are less restricted than in the past. ..

As is clear from some of these operational examples, according to the exemplary embodiments, the higher resolution of the medical image causes medical use, such as any of the above problems, problems derived from it, and related problems. It is possible to solve problems caused by image resolution.

In embodiments, the computer may include an artificial intelligence engine. In this case, an artificial intelligence engine can be used to convert the accepted image into a first image. That is, the resolution can be increased by the artificial intelligence engine.

By using such an artificial intelligence engine, it is possible to perform a process of increasing the resolution of a low-resolution image with higher accuracy and higher accuracy.

In the embodiment, the computer acquires a second image of a predetermined portion of a subject having a second resolution higher than a predetermined resolution, and displays the first image and the second image side by side on a display means. May be configured to perform. Here, the display means may be included in the computer or may be provided outside the computer.

According to such an embodiment, the first image and the second image having a resolution higher than a predetermined resolution can be comparatively observed.

In the embodiment, the computer causes the display means to display a plurality of images of a predetermined portion having a resolution higher than a predetermined resolution, including the first image and the second image, in an order according to their shooting dates. Can be done.

According to such an embodiment, since a plurality of images having a resolution higher than a predetermined resolution can be displayed side by side in the order in which they were taken, it is possible to easily grasp the time-series change of the predetermined portion. It is possible.

In the embodiment, the computer acquires a second image of a predetermined portion having a second resolution higher than a predetermined resolution, further performs a predetermined analysis process on the first image to generate the first analysis data, and The second image can be analyzed to generate the second analysis data.

According to such an embodiment, the results of analyzing the first image and the second image having a resolution higher than a predetermined resolution can be compared, so that the comparison can be made with higher accuracy and higher accuracy than when the resolution is not increased. Analysis is possible.

In the embodiment, the computer performs analysis processing on each of a plurality of images of a predetermined portion having a resolution higher than a predetermined resolution, including the first image and the second image, to generate a plurality of analysis data, and further. A plurality of analysis data can be arranged and displayed on the display means in an order according to the shooting date of the plurality of images.

According to such an embodiment, the results of analyzing a plurality of images having a resolution higher than a predetermined resolution can be displayed side by side in the shooting order of the images, so that the time-series change of the analysis data can be easily grasped. It is possible.

In the embodiment, the computer performs analysis processing on each of a plurality of images of a predetermined portion having a resolution higher than a predetermined resolution, including the first image and the second image, to generate a plurality of analysis data, and further. Based on the shooting dates of the plurality of images and the plurality of analysis data, it is possible to generate information representing the time-series changes in the characteristics of the predetermined portion.

According to such an embodiment, information representing a time-series change in the characteristics of a predetermined part (that is, a time-series change in analysis data) is generated based on the result of analyzing a plurality of images having a resolution higher than a predetermined resolution. Therefore, it is possible to provide diagnostic materials with higher accuracy and higher accuracy than before.

In embodiments, the artificial intelligence engine can make inferences about the disease or candidate thereof affecting the subject based on the first image.

According to such an embodiment, by combining the use of a high-resolution image and the use of an artificial intelligence engine, a diagnosis such as screening of a predetermined disease, presence / absence of a disease, degree of disease progression, selection of a test type, etc. It will be possible to provide support with higher accuracy and accuracy than before.

In an embodiment, if the computer can obtain medical information about the subject, the artificial intelligence engine can perform inferences based on the first image and the medical information.

According to such an embodiment, the medical information of the subject can be referred to in addition to the high-resolution image, so that the accuracy and accuracy of the diagnostic support can be further improved.

In the embodiment, the image of the predetermined portion may be an image of the eye to be inspected. In ophthalmology, various types of images are used for diagnosis and screening. Therefore, it can be said that ophthalmology is one of the clinical departments to which the application of the embodiment is particularly effective. Needless to say, it is also possible to apply the embodiment to clinical departments other than ophthalmology.

In embodiments, the computer may include an ophthalmic artificial intelligence engine with an ophthalmic knowledge base. In this case, the artificial intelligence engine for ophthalmology can convert the image of the eye to be inspected into the first image.

By using such an artificial intelligence engine for ophthalmology, it is possible to perform a process of increasing the resolution of the image of the eye to be inspected with higher accuracy and higher accuracy.

In embodiments, the ophthalmic knowledge base may include feature information that represents the anatomical features of the eye. In this case, the ophthalmic artificial intelligence engine can generate the first image based on at least the feature information.

According to such an embodiment, the image of the eye to be inspected can be made high resolution by referring to the anatomical features of the eye, so that the accuracy and accuracy of the process of making the image of the eye to be inspected high resolution can be further improved. be able to.

In embodiments, the feature information may include left eye information representing the anatomical features of the left eye and right eye information representing the anatomical features of the right eye. In this case, a computer (an ophthalmic artificial intelligence engine or other processor) can determine whether the image of the eye to be inspected is a left eye image or a right eye image. Further, when the image of the eye to be inspected is determined to be the left eye image, the ophthalmic artificial intelligence engine can generate the first image based on at least the left eye information. On the other hand, when the image of the eye to be inspected is determined to be the right eye image, the ophthalmic artificial intelligence engine can generate the first image based on at least the right eye information.

According to such an embodiment, when the image of the eye to be inspected is the image of the left eye, the anatomical features of the left eye are used to perform high resolution, and the image of the eye to be inspected is the image of the right eye. In some cases, the anatomical features of the right eye can be used to achieve higher resolution. Therefore, it is possible to further improve the accuracy and accuracy of the process for increasing the resolution of the image to be inspected.

The medical image processing apparatus for realizing one or more of the above-exemplified medical image processing methods includes, for example, an image receiving unit, an image conversion unit, and a processing unit. The image receiving unit receives an image of a predetermined part of the subject having a predetermined resolution. In the example shown in FIG. 7, the communication unit 40 functions as an image reception unit. The image conversion unit converts the received image into a first image having a first resolution higher than a predetermined resolution. In the example shown in FIG. 7, the image conversion unit 31 functions as an image conversion unit. The processing unit executes a predetermined process based on the first image generated by the image conversion unit. In the example shown in FIG. 7, the data processing unit 30 and / or the control unit 10 and the like function as processing units.

According to such a medical image processing apparatus, it is possible to increase the resolution of a low-resolution image acquired in the past, for example, when long-term follow-up is required such as glaucoma. As a result, the problems of comparative observation and comparative analysis caused by the difference in quality (resolution) of a plurality of images to be compared are solved.

In addition, since it is possible to increase the resolution of a low-resolution image obtained by a simple photographing device in an emergency or a home visit, it is preferable to compare it with a high-resolution image obtained by a high-performance photographing device. be able to.

In addition, even when only a simple imaging device can be used, such as in an emergency or disaster, the low-resolution image can be made high-resolution and used for diagnosis, so it is possible to improve the accuracy of diagnosis. Become.

In addition, even if the resolution is lowered due to the influence of the communication environment in telemedicine, etc., the resolution of live images and captured images can be increased, so that the recognition and judgment of doctors are less restricted than in the past. ..

As is clear from some of these operational examples, according to an exemplary medical image processing device, due to the higher resolution of medical images, any of the above-mentioned problems, problems derived from them, problems related to them, etc. , It is possible to solve the problems caused by the resolution of medical images.

In the embodiment, the image of the predetermined portion may be an image of the eye to be inspected. In this case, the image converter may include an ophthalmic artificial intelligence engine with an ophthalmic knowledge base. This ophthalmic artificial intelligence engine is configured to convert the image of the eye to be inspected into a first image.

According to such a medical image processing apparatus, by using an artificial intelligence engine for ophthalmology, it is possible to perform a process of increasing the resolution of an image of an eye to be inspected with higher accuracy and higher accuracy.

In embodiments, the ophthalmic knowledge base may include feature information that represents the anatomical features of the eye. In this case, the ophthalmic artificial intelligence engine can generate the first image based on at least the feature information.

According to such a medical image processing device, the image of the eye to be inspected can be made high resolution by referring to the anatomical features of the eye, so that the accuracy and accuracy of the process of making the image of the eye to be inspected high resolution are further improved. Can be planned.

In embodiments, the feature information may include left eye information representing the anatomical features of the left eye and right eye information representing the anatomical features of the right eye. In this case, the medical image processing apparatus may include a discriminating unit that determines whether the image of the eye to be inspected is a left eye image or a right eye image. In the example shown in FIG. 7, the left / right discriminating unit 32 functions as a discriminating unit. When the image of the eye to be inspected is determined to be the left eye image, the ophthalmic artificial intelligence engine can generate the first image based on at least the left eye information. On the other hand, when the image of the eye to be inspected is determined to be the right eye image, the ophthalmic artificial intelligence engine can generate the first image based on at least the right eye information.

According to such an embodiment, it is possible to perform high resolution by utilizing the anatomical features corresponding to the eye to be examined depicted in the image, and therefore, the process of increasing the resolution of the image of the eye to be inspected. It is possible to further improve the accuracy and accuracy.

In the embodiment, the processing unit may include a display control unit. The display control unit causes the display means to display the second image of the predetermined portion having a second resolution higher than the predetermined resolution acquired in advance and the first image side by side. In the example shown in FIG. 7, the output control unit 11 functions as a display control unit.

According to such an embodiment, the first image and the second image having a resolution higher than a predetermined resolution can be comparatively observed.

In the embodiment, the processing unit may include an image analysis unit. The image analysis unit performs a predetermined analysis process on the first image to generate the first analysis data, and analyzes the second image of a predetermined portion having a second resolution higher than the predetermined resolution acquired in advance. The second analysis data is generated. In the example shown in FIG. 7, the image analysis unit 33 functions as an image analysis unit.

According to such an embodiment, the results of analyzing the first image and the second image having a resolution higher than a predetermined resolution can be compared, so that the comparison can be made with higher accuracy and higher accuracy than when the resolution is not increased. Analysis is possible.

In the embodiment, the processing unit may include an inference unit. The inference unit makes inferences about the disease suffering from the subject or a candidate thereof based on the first image. In the example shown in FIG. 7, the inference unit 34 functions as an inference unit.

According to such an embodiment, by combining the use of high-resolution images and inference processing, diagnostic support such as screening for a predetermined disease, presence / absence of a disease, degree of disease progression, and selection of a test type can be provided. It is possible to provide with higher accuracy and higher accuracy than before.

The steps included in the medical image processing method according to the embodiment and the elements (configuration, operation, etc.) included in the medical image processing apparatus are not limited to the above examples.

The embodiments described above are merely examples of the present invention. A person who intends to carry out the present invention can arbitrarily make modifications (omission, substitution, addition, etc.) within the scope of the gist of the present invention.

1 Medical image processing device 10 Control unit 11 Output control unit 20 Storage unit 21 Knowledge base 30 Data processing unit 31 Image conversion unit 32 Left / right discrimination unit 33 Image analysis unit 34 Reasoning unit 40 Communication unit 100 User interface 101 Display unit 102 Operation unit 200 External device

Claims (4)

A method of processing medical images using a computer,
The computer
Includes an artificial intelligence engine that has learned multiple combinations of low-resolution and high-resolution images
Accepts an image of the eye to be inspected with a predetermined resolution,
The artificial intelligence engine converts the image into a first image having a first resolution higher than the predetermined resolution.
Performing a predetermined process based on the first image,
The artificial intelligence engine includes an ophthalmic artificial intelligence engine having an ophthalmic knowledge base including feature information representing anatomical features of the eye.
A medical image processing method , wherein the artificial intelligence engine for ophthalmology converts an image of the eye to be inspected into the first image based on at least the feature information. The feature information includes left eye information representing the anatomical features of the left eye and right eye information representing the anatomical features of the right eye.
The computer determines whether the image of the eye to be inspected is a left eye image or a right eye image.
When the artificial intelligence engine for ophthalmology is determined to be the left eye image, the first image is generated based on at least the left eye information, and when it is determined to be the right eye image, the first image is generated. The medical image processing method according to claim 1 , wherein the first image is generated based on at least the right eye information. The anatomical features include any of the location of the optic disc, the morphology of the optic disc, the location of the macula, the morphology of the macula, the running of blood vessels, the morphology of blood vessels, the running of nerve fibers, and the morphology of nerve fibers. The medical image processing method according to claim 1 or 2 , wherein the medical image processing method is characterized. An image reception unit that accepts images of the eye to be inspected with a predetermined resolution,
An image that includes an artificial intelligence engine that has learned a plurality of combinations of a low-resolution image and a high-resolution image, and converts the image received by the image receiving unit into a first image having a first resolution higher than the predetermined resolution. Conversion part and
A processing unit that executes a predetermined process based on the first image is provided .
The artificial intelligence engine includes an ophthalmic artificial intelligence engine having an ophthalmic knowledge base including feature information representing anatomical features of the eye.
The ophthalmic artificial intelligence engine converts the image of the eye to be inspected into the first image based on at least the feature information.
A medical image processing device characterized by the fact that.

Images