KR102215056B1 - Medical Image De-Identification System, Method and Computer Readable Recording Medium - Google Patents

Abstract

Disclosed is a medical image de-identification system, method, and computer-readable recording medium. A medical image de-identification system according to an embodiment of the present invention includes a medical image receiving unit that receives a medical image including an image that can identify a facial region of a patient, and a facial region that separates the facial region from the medical image. Separation unit, a request receiving unit that receives a request to provide data for at least one of the eyes, nose, and mouth included in the facial area from the client, and a de-identification unit that performs de-identification on the rest of the facial area excluding the requested area Includes.

Description

Medical Image De-Identification System, Method and Computer Readable Recording Medium}

The present invention relates to a medical image de-identification system, method, and computer-readable recording medium, and more specifically, to perform irreversible de-identification of a body part that can identify an individual, thereby making it impossible to reconstruct personal information. It relates to a medical image de-identification system, method, and computer-readable recording medium.

For medical images such as CT, MRI, ultrasound, and endoscope, personal information may be included in the image file, and in the case of high-resolution CT and MRI images including faces, the appearance of the patient's face can be known through some processing.

FIG. 1 exemplarily shows a 3D image that can be restored from a medical image. Referring to FIG. 1, it can be seen that a patient's face can be restored from a medical image. Although the medical image may be provided as a 2D image, even if it is provided as a 2D image, a 3D image may be generated using a plurality of 2D images, and thus a problem of personal information leakage may occur.

Accordingly, there is a need for a medical image management method that can prevent the leakage of patient personal information.
The background technology of the present invention is disclosed in Korean Patent Application Publication No. 10-2012-0035299 (published on April 16, 2012, title of the invention: an image protection processing device for privacy protection, and an image security system and method using the same). have.

An object of the present invention is to provide a medical image de-identification system, a method, and a computer-readable recording medium capable of preventing the leakage of personal information of a patient.

A medical image de-identification system according to an embodiment of the present invention includes a medical image receiving unit that receives a medical image including an image that can identify a facial region of a patient, and a facial region that separates the facial region from the medical image. Separation unit, a request receiving unit that receives a request to provide data for at least one of the eyes, nose, and mouth included in the facial area from the client, and a de-identification unit that performs de-identification on the rest of the facial area excluding the requested area Includes.

In addition, the de-identification unit may remove the personal identification text included in the medical image or replace it with another text.

In addition, the de-identification performed by the de-identification unit is an irreversible data conversion for the eyes, nose, and mouth, and the de-identification unit performs masking or blurring on a target region to be de-identified. Can be done.

In addition, the facial region separating unit may separate the facial region using machine learning results of the eyes, nose, and mouth.

Meanwhile, in the medical image de-identification method according to an embodiment of the present invention, in a medical image de-identification method using a medical image de-identification system, a medical image including an image capable of identifying a facial region of a patient is generated. Receiving, separating the facial area from the medical image, receiving a request to provide data for at least one of eyes, nose, and mouth included in the facial area from a client, and the rest of the face excluding the requested area And performing de-identification on the region.

In addition, in performing the de-identification, the personal identification text included in the medical image may be removed or replaced with another text.

In addition, in the step of performing the de-identification, irreversible data conversion for the eyes, nose, and mouth may be performed, and in the step of performing the de-identification, masking or Blurring can be performed.

Also, in the step of separating the facial area, the facial area may be separated by using machine learning results of the eyes, nose, and mouth.

In addition, a computer-readable recording medium in which a program for performing the medical image de-identification method according to the present invention is recorded may be provided.

The present invention can provide a medical image de-identification system, a method, and a computer-readable recording medium capable of preventing the leakage of personal information of a patient.

1 is a diagram illustrating an exemplary 3D image reconstructable from a medical image.
2 is a diagram schematically showing the configuration of a medical image de-identification system according to an embodiment of the present invention.
3 is a diagram illustrating a medical image that is an object of de-identification.
4 is a diagram illustrating an exemplary result of de-identification of a medical image.
5 is a diagram schematically showing an embodiment of a de-identification method.
6 is a flowchart schematically illustrating a method of de-identifying a medical image according to an embodiment of the present invention.

Advantages and features of the present invention, and a method of achieving them will be apparent with reference to embodiments described in detail together with the accompanying drawings. However, the present invention is not limited to the embodiments presented below, but may be implemented in various different forms, and it should be understood that it includes all transformations, equivalents, or substitutes included in the spirit and scope of the present invention. do. The embodiments presented below are provided to complete the disclosure of the present invention, and to fully inform a person of ordinary skill in the art to which the present invention belongs. In describing the present invention, when it is determined that a detailed description of a related known technology may obscure the subject matter of the present invention, a detailed description thereof will be omitted.

The terms used in the present application are only used to describe specific embodiments, and are not intended to limit the present invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In the present application, terms such as "comprises" or "have" are intended to designate the presence of features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, but one or more other features. It is to be understood that the presence or addition of elements or numbers, steps, actions, components, parts, or combinations thereof, does not preclude in advance. Terms such as first and second may be used to describe various elements, but the elements should not be limited by the terms. These terms are used only for the purpose of distinguishing one component from another component.

2 is a diagram schematically illustrating a configuration of a medical image de-identification system according to an embodiment of the present invention.

2, a medical image de-identification system 100 according to an embodiment of the present invention includes a medical image receiving unit 110, a facial region separation unit 120, a request receiving unit 130, and a de-identifying unit ( 140).

The medical image receiving unit 110 receives a medical image including an image capable of identifying a facial region of a patient. The medical image may be an image such as Computed Tomography (CT) or Magnetic Resonance Imaging (MRI), and may be a plurality of 2D images or 3D images.

The medical image de-identification system and method according to the present invention can be obtained from the medical image in order to solve the problem that the patient's personal information is unnecessarily disclosed when the medical image of the patient is provided to the outside for the purpose of research. It aims to remove possible personal information.

Accordingly, the medical image received by the medical image receiving unit 110 is an image capable of identifying the patient's facial area, and the shape of the patient's face, such as eyes, nose, and mouth, is an image capable of specifying an individual. Can be understood as

The medical image may further include text information indicating personal information of the patient as well as an image of the patient's body part. For example, the medical image may include information such as the patient's name, contact information, and diagnosis name in text form.

The image corresponding to the medical image and the text including the personal information can be understood as a de-identification target by the medical image de-identification system and method according to the present invention.

The facial region separation unit 120 separates the patient's facial region from the medical image. The facial area may be understood as a facial area including eyes, nose, mouth, and ears of the patient. In order to separate the facial region from the medical image, the facial region separating unit 120 may use machine learning results of the human eye, nose, mouth, and ears. For example, the facial region separating unit 120 determines the positions of the eyes, nose, and mouth from the CT or MRI images of the head of a plurality of patients. Can determine the location of.

In this case, the object of the machine learning may be a 2D image or a 3D image. When the medical image is a 2D image, the positions of eyes, nose, mouth, and ears may be learned from multiple 2D images. Alternatively, when the medical image is a 3D image, learning to determine a facial region including the patient's eyes, nose, mouth, and ears may be performed.

Accordingly, when a 2D image is learned and the learning result is used, a plurality of images including eyes, nose, mouth, and ears may be selected by the facial region separation unit 120.

The request receiving unit 130 receives a request to provide data for at least one of eyes, nose, and mouth included in the facial area from a client. The medical image may be used for purposes such as research, and the client may be a person, an institution, or another system that wishes to use the medical image.

The purpose of the study is various and the body part subject to the study may be various, and the client may request a medical image of at least one of the eyes, nose, and mouth. In this case, if the patient's medical image is provided as it is without any separate processing, the patient can be identified through the 3D medical image, and even if the 2D medical image is provided, a 3D image can be created by combining it, so patient information will be exposed. I can.

In order to prevent unintended patient information from being exposed, the de-identification unit 140 de-identifies the rest of the facial area except for the portion requested by the client.

For example, when the client requests data on the patient's nose, the de-identification unit 140 performs de-identification on eyes, mouths, ears, etc. excluding the nose. The de-identification is irreversible. When the client receives a medical image in which the nose has been de-identified, it is impossible to restore the patient's nose from the corresponding medical image.

The de-identification unit 140 may convert the medical image into a 2D image to perform de-identification on a plurality of images including the nose, and directly de-identify the nose region in the 3D medical image. You can perform anger. Meanwhile, the de-identification unit 140 may convert the medical image into a plurality of 2D images after de-identifying the 3D medical image. That is, it can be understood that the medical image provided to the client is in 2D form.

The de-identification may be masking or blurring of the target region, for example, masking all data corresponding to the de-identification target region with '0', or image processing to blur the target region Can be As another example, a mosaic may be applied to the target area.

Meanwhile, in another embodiment of the present invention, the de-identification unit 140 may remove the personal identification text included in the medical image or replace it with another text. As described above, data corresponding to the patient's personal information can be viewed as a de-identification target according to the present invention, and the personal identification text can be viewed as the patient's personal information.

Accordingly, when the medical image includes text that can identify the individual patient, the de-identification unit 140 may perform de-identification by deleting or replacing the text with another text.

3 is a diagram illustrating a medical image that is an object of de-identification.

Referring to FIG. 3, FIG. 3(a) shows an image of any one of a plurality of 2D medical images, and FIG. 3(b) is an image in which a skin area is separately divided among the medical images shown in FIG. 3(a). .

As described with reference to FIG. 2, the facial region separation unit 120 separates the patient's facial region from the medical image. Since the skin region shown in FIG. 3B includes a region corresponding to the patient's eyes, nose, mouth, and ears, it can be understood as including the facial region.

Since the skin area shown in FIG. 3(b) contains information that can identify a patient, it is subject to de-identification. In the medical image of FIG. 3(a), the skin area corresponding to the skin area of FIG. 3(b) Parts other than the area are not subject to de-identification. This is because the image inside the skin area is difficult to be used to identify the patient.

The facial region separating unit 120 may utilize the result of performing machine learning to separate the skin region from the medical image as illustrated in FIG. 3A. For example, it is possible to learn by receiving a plurality of raw data tagged with the skin region from a medical image, and the facial region separating unit 120 uses the learning result to obtain skin from a newly input medical image. Areas can be separated.

Alternatively, as described with reference to FIG. 2, the eyes, nose, mouth, eyes, nose, mouth, etc. are newly inputted from medical images using the result of receiving and learning a plurality of original data tagged with the eyes, nose, mouth, and ears. You can determine the location of your ears, etc.

Meanwhile, although only 2D medical images are shown in FIG. 3, the present invention is not necessarily limited to a system or method for a two-dimensional image, and a 3D medical image is also used for the facial area or skin of the patient by using a machine learning result of a similar method. Areas can be separated.

4 is a diagram illustrating an exemplary result of de-identification of a medical image.

Referring to FIG. 4, from medical images in which de-identification has not been sequentially performed to medical images in which de-identification has been performed for the entire facial region from FIGS. 4A to 4C are shown. Although the medical image shown in FIG. 4 is a 3D image, de-identification may be performed directly on a 3D medical image, or a 3D image of FIG. 4 may be obtained from a 2D image on which de-identification has been performed.

Referring to FIG. 4(a), it can be seen that the patient's eyes, nose, and mouth are expressed to the extent that an individual can be identified. In addition, it can be seen from FIG. 4(b) that de-identification is performed on the patient's eyes and mouth, so that an individual cannot be identified. As described with reference to FIG. 2, the request receiving unit 130 receives a request to provide data for a specific part from a client. Referring to FIG. 4(b), it is assumed that a request for providing data for the nose from the client is received. I can understand.

Meanwhile, in FIG. 4(c), it can be understood that de-identification has been performed on all of the patient's eyes, nose, and mouth, and it can be understood that the client has not requested data on the eyes, nose, and mouth. However, even if de-identification is performed on all of the patient's eyes, nose, and mouth as shown in Fig. 4(c), image damage does not occur in the subcutaneous area including the brain and blood vessels. It cannot be specified, but there is no problem in using the necessary information.

In addition, a medical image that has been de-identified according to a client's request may be a 3D image as illustrated in FIG. 4, but the information provided to the client may be a 2D image as illustrated in FIG. 3. Accordingly, the client can use the provided 2D image for a required purpose, but even if the 3D image is generated by using the 2D image, it is not possible to obtain information sufficient to identify the individual patient.

5 is a diagram schematically showing an embodiment of a de-identification method.

As described with reference to FIG. 2, the de-identification unit 140 may perform de-identification of a medical image, and the de-identification may be performed by masking or blurring the target region to be de-identified ( blurring), etc.

FIG. 5 exemplarily shows a de-identification method using a convolution filter, where a source pixel corresponds to an image to be de-identified, and a convolution kernel corresponds to a matrix applied to the source pixel.

The source pixel and the central element of the convolution kernel are multiplied with each other, and the calculated result, and the sum of the products of the neighboring elements of the source pixel and the elements at a corresponding position of the convolution kernel may be determined as a new pixel value.

In the example of FIG. 5, the product of the source pixel and the central element of the convolution kernel is 0, but the sum of the products of the remaining elements is -8, so the new pixel value is determined to be -8.

Meanwhile, the size of the matrix corresponding to the source pixel and the convolution kernel may be determined in various ways, and is not necessarily limited to 3x3 as shown in FIG. 5. Likewise, since the weights of each element of the convolution kernel are also not limited to the values shown in FIG. 5, the degree of de-identification can be adjusted by adjusting the kernel size and the kernel weight.

In the de-identification process according to the present invention, when the above-described convolution is performed on a pixel to be de-identified, all values are designated as 0 for the boundary area to which the square matrix cannot be applied, or by duplicating the values designated in the vicinity How to do it can be applied.

When Kernel Convolution is applied in this way, it can be determined that irreversible de-identification has occurred. In the de-identified image, only the approximate shape of the object can be estimated, and it is impossible to convert it into original data.

On the other hand, in order to secure additional irreversibility for data on which a group of irreversible de-identification has been performed, hash mapping or the like can be applied. The hash mapping can be applied based on the signal intensity of Voxel in the unidentified area.

6 is a flowchart schematically illustrating a method of de-identifying a medical image according to an embodiment of the present invention.

Referring to FIG. 6, a method for de-identifying a medical image according to an embodiment of the present invention includes a medical image receiving step (S10), a facial region separation step (S20), a request receiving step (S30), and a de-identification step. It includes (S40).

The medical image de-identification method according to the present invention can use the medical image de-identification system described with reference to the preceding drawings, and in the medical image receiving step (S10), an image that can identify the patient's facial area is included. Receive medical images. The medical image may be an image such as Computed Tomography (CT) or Magnetic Resonance Imaging (MRI), and may be a plurality of 2D images or 3D images.

The medical image de-identification system and method according to the present invention can be obtained from the medical image in order to solve the problem that the patient's personal information is unnecessarily disclosed when the medical image of the patient is provided to the outside for the purpose of research. It aims to remove possible personal information.

Accordingly, the medical image received in the medical image receiving step (S10) is an image capable of identifying the patient's facial area, and the shape of the patient's face such as eyes, nose, and mouth is such that an individual can be identified. It can be understood through video.

The medical image may further include text information indicating personal information of the patient as well as an image of the patient's body part. For example, the medical image may include information such as the patient's name, contact information, and diagnosis name in text form.

The image corresponding to the medical image and the text including the personal information can be understood as a de-identification target by the medical image de-identification system and method according to the present invention.

In the step of separating the facial area (S20), the facial area is separated from the medical image. The facial area may be understood as a facial area including eyes, nose, mouth, and ears of the patient. In order to separate the facial area from the medical image, in the facial area separation step (S20), machine learning results for the human eye, nose, mouth, and ears may be used. For example, in the facial area separation step (S20), the eyes, nose, and mouth are newly inputted through learning to determine the positions of eyes, nose, and mouth from CT or MRI images of the head of a plurality of patients. Can determine the location of.

In this case, the object of the machine learning may be a 2D image or a 3D image. When the medical image is a 2D image, the positions of eyes, nose, mouth, and ears may be learned from multiple 2D images. Alternatively, when the medical image is a 3D image, learning to determine a facial region including the patient's eyes, nose, mouth, and ears may be performed.

Therefore, when a 2D image is learned and the learning result is used, a plurality of images including eyes, nose, mouth, and ears may be selected in the facial region separation step S20.

In the request receiving step S30, a request for providing data for at least one of eyes, nose, and mouth included in the facial area is received from a client. The medical image may be used for purposes such as research, and the client may be a person, an institution, or another system that wishes to use the medical image.

The purpose of the study is various and the body part subject to the study may be various, and the client may request a medical image of at least one of the eyes, nose, and mouth. In this case, if the patient's medical image is provided as it is without any separate processing, the patient can be identified through the 3D medical image, and even if the 2D medical image is provided, a 3D image can be created by combining it, so patient information will be exposed. I can.

In the de-identification step (S40), in order to prevent unintended patient information from being exposed, de-identification is performed on the rest of the facial area except for the portion requested by the client.

For example, when the client requests data on the patient's nose, in the de-identification step (S40), de-identification is performed on the eyes, mouth, and ears excluding the nose. The de-identification is irreversible. When the client receives a medical image in which the nose has been de-identified, it is impossible to restore the patient's nose from the corresponding medical image.

In the de-identification step (S40), de-identification may be performed on a plurality of images including the nose by converting the medical image into a 2D image, and the ratio of the nose region directly in the 3D medical image Identification can be performed. Meanwhile, in the de-identification step S40, after de-identification of the 3D medical image is performed, the corresponding medical image may be converted into a plurality of 2D images. That is, it can be understood that the medical image provided to the client is in 2D form.

The de-identification may be masking or blurring of the target region, for example, masking all data corresponding to the de-identification target region with '0', or image processing to blur the target region Can be As another example, a mosaic may be applied to the target area.

Meanwhile, in another embodiment of the present invention, in the de-identification step S40, the personal identification text included in the medical image may be removed or replaced with another text. As described above, data corresponding to the patient's personal information can be viewed as a de-identification target according to the present invention, and the personal identification text can be viewed as the patient's personal information.

Therefore, in the de-identification step (S40), when the medical image contains text that can specify the individual patient, de-identification may be performed by deleting or replacing the text with another text.

Meanwhile, the present invention can be implemented as a computer-readable code on a computer-readable recording medium. The computer-readable recording medium includes all types of recording devices storing data that can be read by a computer system. Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, floppy disk, and optical data storage devices.

Further, the computer-readable recording medium is distributed over a computer system connected by a network, so that the computer-readable code can be stored and executed in a distributed manner. In addition, functional programs, codes, and code segments for implementing the present invention can be easily inferred by programmers in the technical field to which the present invention belongs.

If there is no explicit order or contradictory description of the steps constituting the method according to the present invention, the steps may be performed in a suitable order. The present invention is not necessarily limited according to the order of description of the steps.

The use of all examples or illustrative terms (for example, etc.) in the present invention is merely for describing the present invention in detail, and the scope of the present invention is limited by the above examples or illustrative terms unless limited by the claims. It is not limited. In addition, a person skilled in the art may recognize that various modifications, combinations, and changes may be configured according to design conditions and factors within the scope of the appended claims or their equivalents.

Therefore, the spirit of the present invention is limited to the above-described embodiments and should not be defined, and not only the claims to be described later, but also all ranges equivalent to or equivalently changed from the scope of the present invention. It will be said to belong to the category.

100: medical image de-identification system
110: medical image receiving unit
120: facial area separation unit
130: request receiver
140: de-identification unit

Claims (13)

An image of a skin area that is photographed using a medical device and represents a cross-section of the patient's head, capable of identifying the patient's facial area outside the cross-section of the head, and an image of the subcutaneous area on the inside of the facial area inside the cross-section of the head A medical image receiving unit for receiving a 2D cross-sectional medical image including;
Using the learning result of determining the position of the eyes, nose, and mouth from the medical images of the head of a plurality of patients, the positions of the eyes, nose, and mouth related to the shape of the patient's face are obtained from the medical image, A face region separating unit separating the skin region and the subcutaneous region using the obtained position;
A request receiving unit for receiving a request to provide data on at least one of eyes, nose, and mouth included in the skin area from a client; And
A de-identification unit for de-identifying the rest of the skin area except for the requested area;
Including,
The 2D cross-sectional medical image includes a cut surface obtained by cutting the medical image in a second direction perpendicular to a first direction facing the patient's face,
The image of the subcutaneous region includes at least one image of the patient's brain and blood vessels, and is not subject to de-identification by the de-identification unit. The method of claim 1,
The 2D cross-sectional medical image is a medical image that is divided and converted to represent the cut surface of the head from a 3D image representing the head of a patient photographed using the medical device. The method of claim 1,
The de-identification unit removes the personal identification text included in the medical image or replaces it with another text. The method of claim 1,
The de-identification performed by the de-identification unit is an irreversible data conversion for the eyes, nose, and mouth. The method of claim 1,
The de-identification unit is a medical image de-identification system that performs masking or blurring on a region to be de-identified. The method of claim 1,
The facial region separating unit is a medical image de-identification system for separating the skin region using a machine learning result of the eyes, nose, and mouth. In the medical image de-identification method using a medical image de-identification system,
An image of a skin area that is photographed using a medical device and represents a cross-section of the patient's head, capable of identifying the patient's facial area outside the cross-section of the head, and an image of the subcutaneous area on the inside of the facial area inside the cross-section of the head Receiving a 2D cross-sectional medical image comprising a;
Using the learning result of determining the position of the eyes, nose, and mouth from the medical images of the head of a plurality of patients, the positions of the eyes, nose, and mouth related to the shape of the patient's face are obtained from the medical image, Separating the skin area and the subcutaneous area using the acquired position;
Receiving a request to provide data on at least one of eyes, nose, and mouth included in the skin area from a client; And
Performing de-identification on the remaining skin regions except for the requested region;
Including,
The 2D cross-sectional medical image includes a cut surface obtained by cutting the medical image in a second direction perpendicular to a first direction facing the patient's face,
The image of the subcutaneous region includes at least one image of the patient's brain and blood vessels, and is not subject to the de-identification. The method of claim 7,
The 2D cross-sectional medical image is a medical image divided and converted to represent the cut surface of the head from a 3D image representing the head of a patient photographed using the medical device. The method of claim 7,
In the step of performing the de-identification, the personal identification text included in the medical image is removed or replaced with another text. The method of claim 7,
In the step of performing the de-identification, a medical image de-identification method for performing irreversible data conversion for the eyes, nose, and mouth. The method of claim 7,
In the step of performing the de-identification, a medical image de-identification method of performing masking or blurring on a portion to be de-identified. The method of claim 7,
In the step of separating the skin region and the subcutaneous region, a medical image de-identification method of separating the skin region using machine learning results of the eyes, nose, and mouth. A computer-readable recording medium on which a program for performing the method according to any one of claims 7 to 12 is recorded.

Images