JP5817582B2 - Brain function disease diagnosis support apparatus and brain function disease diagnosis support method - Google Patents

Description

  The present invention relates to a brain function disease diagnosis support apparatus and a brain function disease diagnosis support method for supporting diagnosis of a brain function disease.

  In recent years, aging has progressed due to the advancement of medical care and the like, and the number of patients with cerebral function diseases has increased significantly with aging. Among brain function disorders, an increase in the number of patients with dementia has been recognized. For example, the International Conference on Prevention of Dementia in 2007 (sponsored by the Alzheimer's Association in Washington DC, USA) ), If the number of Alzheimer-type dementia patients continues to increase as they are now, it is reported that the number of patients in the world may exceed 100 million in 2050.

  Dementia includes Alzheimer-type dementia and senile dementia such as cerebrovascular dementia. Alzheimer-type dementia and senile dementia use different drugs and treatments. Different. However, in the early stage of the disease, it is difficult to distinguish both. In particular, for Alzheimer-type dementia, new drugs that delay the progression of the disease have been developed, and the development of an accurate discrimination method in the early stage of the disease is desired for treatment.

  In Alzheimer-type dementia, visuospatial perception disorder in which the positional relationship, size, and perspective of an object cannot be grasped is recognized at an early stage (for example, Non-Patent Document 1). As Alzheimer's dementia progresses further, misunderstandings of people who are not able to correctly recognize the situation in which they are placed, such as people, their surroundings, time, and location, disapproval of persons, disapproval of faces, and misidentification of persons, Symptoms that cannot be identified are observed.

  Conventionally, in order to diagnose a disease at an early stage, a revised Hasegawa simplified intelligence evaluation scale is adopted in which a doctor asks a question and a subject answers to determine whether or not the patient has Alzheimer-type dementia.

  Further, Alzheimer's dementia can be diagnosed based on whether or not there is a visual space perception disorder. Therefore, it is expected to develop a technique for detecting the presence or absence of visuospatial perception through an eye movement test, which is a non-invasive test and places less burden on the subject.

  Therefore, an apparatus has been developed that detects the eye movement and head movement of the subject, derives the movement of the subject's line of sight, and displays the locus of the movement of the line of sight (for example, Patent Document 1). In addition, a technique for diagnosing Alzheimer-type dementia by measuring the size of a subject's pupil at the time of miosis or mydriasis has been developed (for example, Patent Document 2).

Japanese Examined Patent Publication No. 6-53107 JP 2002-34920 A

Clinical and pathological conditions of dementia Mikio Tokaibayashi (Clinical Neurology, Vol. 48, No. 7, 467-475, 2008)

  However, although the technique of Patent Document 1 described above can display the trajectory of the subject's line of sight, it cannot derive where the subject is gazing. Moreover, since it aims at the display about the locus | trajectory of a test subject's eyes | visual_axis, only the technique of patent document 1 WHEREIN: Whether a test subject suffers from a brain functional disease, and when suffering, Alzheimer type dementia Or autism, a type of brain function disease, cannot be diagnosed. Therefore, even if a diagnosis is performed using the technique of Patent Document 1, there is a possibility that a difference in the diagnosis results may occur depending on the experience and ability of the doctor.

  In view of such a problem, the present invention derives a point (gaze point) where the subject is gazing with a simple configuration and analyzes the gaze point, thereby possibly suffering from a brain function disease. It is an object of the present invention to provide a brain function disease diagnosis support apparatus and a brain function disease diagnosis support method that can quantitatively derive sex.

In order to solve the above-described problem, the brain function disease diagnosis support apparatus of the present invention includes an image display unit that displays an image to be viewed by a subject, and an image displayed on the display surface of the image display unit in a plurality of blocks. A block dividing unit that divides, an infrared irradiating unit that irradiates infrared rays toward the subject, and two imaging units that generate at least two pieces of image data having horizontal parallax by imaging infrared reflected light reflected by the subject. Based on the two image data generated by the imaging unit, a pupil calculation unit that calculates the position of the pupil of the subject, and gaze information that indicates the line of sight of the subject based on the position of the pupil calculated by the pupil calculation unit Based on the line-of-sight calculation unit, the line-of-sight information calculated by the line-of-sight calculation unit, a gaze point deriving unit for deriving whether or not there is a subject's gaze point for each of a plurality of blocks, and a gaze point deriving unit Results and information storage section for storing for each block, performs a predetermined calculation based on the result information storage unit has accumulated, and a computation section for determining the possibility whether there is a subject's brain function disorders The image displayed by the image display unit in the block is associated with each of the plurality of blocks, and the calculation unit performs weighting corresponding to the type of disease to be diagnosed based on the content of the image for each block and performs a predetermined process. It is characterized by performing an operation .

  The predetermined operation may be any one selected from the group of a neural network, a naive Bayes, a Bayesian network, and a decision tree method.

In order to solve the above problem, the brain function disease diagnosis support method of the present invention using an image display unit that displays an image for the subject to visually recognize, the image display unit displays an image for the subject to visually recognize Then, the image displayed on the display surface of the image display unit is divided into a plurality of blocks, the image displayed by the image display unit in the block is associated with each of the plurality of blocks, and infrared rays are irradiated toward the subject, at least the subject Infrared reflected light reflected in the step is imaged to generate two image data having horizontal parallax, the position of the pupil of the subject is calculated based on the two generated image data, and based on the calculated pupil position And calculating gaze information indicating the gaze of the subject, deriving whether there is a gaze point of the subject for each of a plurality of blocks based on the computed gaze information, and obtaining the derived result Accumulating for each lock, with reference to the accumulated result, by weighting corresponding to the type of disease to be diagnosed based on the contents of the image of each block, characterized the TURMERIC line a predetermined operation.

  According to the present invention, it is possible to derive the possibility of suffering from a brain function disease quantitatively by deriving the position of the gaze point of the subject with a simple configuration and analyzing the gaze point. It becomes.

It is a figure for demonstrating the arrangement | positioning relationship between a brain function disease diagnosis assistance apparatus and a test subject. It is a block diagram for demonstrating the structure of a brain function disease diagnosis assistance apparatus. It is a figure for demonstrating the image for a test | inspection displayed on the image display part. It is a figure for demonstrating the process by a block division part. It is a figure for demonstrating the cornea reflection image calculation process by a pupil calculation part, and the gaze calculation process by a gaze calculation part. It is a flowchart for demonstrating the flow of a process of the brain function disease diagnosis support method using a brain function disease diagnosis support apparatus. It is a figure for demonstrating the relationship between a block and a neural network. It is a flowchart for demonstrating the flow of a process of the brain function disease diagnosis support method concerning a modification. It is a figure for demonstrating a gaze block pattern. It is a figure for demonstrating the setting of the conditional probability to each block.

  Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. The dimensions, materials, and other specific numerical values shown in the embodiments are merely examples for facilitating the understanding of the invention, and do not limit the present invention unless otherwise specified. In the present specification and drawings, elements having substantially the same function and configuration are denoted by the same reference numerals, and redundant description is omitted, and elements not directly related to the present invention are not illustrated. To do.

(Brain function disease diagnosis support apparatus 100)
FIG. 1 is a diagram for explaining the positional relationship between the brain function disease diagnosis support apparatus 100 and the subject 10, and FIG. 2 is a block diagram for explaining the configuration of the brain function disease diagnosis support apparatus 100.

  As shown in FIG. 1, the subject 10 is positioned on the display surface 114 a side of the image display unit 114 constituting the brain function disease diagnosis support apparatus 100, and two imaging units provided in the vicinity of the image display unit 114. The subject 10 is imaged using 118a, 118b (only the imaging unit 118a is shown in FIG. 1) to generate image data (hereinafter referred to as diagnostic image data). The brain function disease diagnosis support apparatus 100 derives line-of-sight information indicating the line of sight of the subject 10 based on the diagnostic image data generated by the imaging units 118a and 118b.

  Among brain function disorders, Alzheimer-type dementia is characterized by visuospatial perceptual impairment that makes it impossible to grasp the positional relationship, size, and perspective of an object. On the other hand, among brain function disorders, autism is characterized in that it often does not see the eyes of other people who face each other in early childhood.

  Therefore, the brain function disease diagnosis support apparatus 100 according to the present embodiment determines the position on the image (test image) displayed by the image display unit 114 where the subject 10 is gazing (gaze point). The possibility of suffering from a brain function disorder such as Alzheimer type dementia or autism is determined by statistical processing. Hereinafter, a specific configuration of the brain function disease diagnosis support apparatus 100 will be described.

  As shown in FIG. 2, the brain function disease diagnosis support apparatus 100 includes an I / F unit 110, a storage unit 112, an image display unit 114, infrared irradiation units 116a and 116b, imaging units 118a and 118b, The control unit 120 includes an audio output unit 122 and a diagnostic display unit 124.

  The I / F unit 110 is connected to, for example, a server that manages an electronic medical record of a medical institution via a communication network 102 such as the Internet or a LAN (Local Area Network), and transmits and receives data.

  The storage unit 112 includes a storage medium such as a random access memory (RAM), a hard disk drive (HDD), a flash memory, and a nonvolatile RAM, and stores various pieces of information. In the present embodiment, the storage unit 112 stores in advance image data (hereinafter referred to as an inspection image) data to be visually recognized by the subject 10 and audio data for prompting the subject 10 to visually recognize the image.

  The image display unit 114 includes a liquid crystal display, an organic EL (Electro Luminescence) display, and the like, and displays an inspection image based on the inspection image data stored in the storage unit 112. The inspection image will be described in detail later.

  The infrared irradiation unit 116a is provided in the vicinity of the imaging unit 118a, and the infrared irradiation unit 116b is provided in the vicinity of the imaging unit 118b, and both irradiate the subject 10 with infrared rays. The infrared irradiation unit 116a is configured by an LED (Light Emitting Diode) or the like, and irradiates infrared light having a center wavelength of 780 nm to less than 900 nm (here, 850 nm). The infrared irradiation unit 116b is configured by an LED or the like, and irradiates infrared rays having a center wavelength of 900 nm to 970 nm (here, 950 nm). The infrared irradiation units 116a and 116b irradiate infrared rays so as to be parallel to the optical axes of the lenses of the imaging units 118a and 118b. The diagnostic image data generated when the imaging units 118a and 118b capture the reflected light when the infrared rays irradiated by the infrared irradiation units 116a and 116b are reflected by the subject 10 will be described in detail later.

  The imaging units 118a and 118b image infrared reflected light reflected by the subject 10 and generate diagnostic image data having horizontal parallax, respectively. More specifically, an optical filter that transmits only infrared rays having a center wavelength of 780 nm to less than 900 nm (here, 850 nm) is provided in front of the lens in the imaging unit 118a (subject 10 side). In addition, an optical filter that transmits only infrared light having a center wavelength of 900 nm to 970 nm (here, 950 nm) is provided in front of the lens in the imaging unit 118b (subject 10 side).

  The central control unit 120 is composed of a semiconductor integrated circuit including a CPU (Central Processing Unit), reads programs and parameters for operating the CPU itself from the ROM, and cooperates with the RAM as a work area and other electronic circuits. Thus, the entire brain function disease diagnosis support apparatus 100 is managed and controlled. In addition, the central control unit 120 functions as an image selection unit 150, a block division unit 152, a pupil calculation unit 154, a gaze calculation unit 156, a gaze point derivation unit 158, an information storage unit 160, and a calculation unit 162.

  The image selection unit 150 selects one test image data in response to an operation input by an examiner (such as a doctor) from the plurality of test image data stored in the storage unit 112, and when the selection is completed, An inspection image based on the predetermined inspection image data is displayed on the image display unit 114.

  FIG. 3 is a diagram for explaining the inspection image displayed on the image display unit 114. As described above, visuospatial perception disorder is recognized in Alzheimer type dementia. Therefore, when diagnosing whether there is a possibility of Alzheimer type dementia for the subject 10, for example, an image selection unit In response to an operation input by the inspector 150, the image display unit 114 displays the inspection image 130a including the scenery and the structure showing the distant view and the near view shown in FIG. In the inspection image 130a shown in FIG. 3A, only the two structures of the foreground structure 132a and the foreground structure 132b are shown. Three or more (for example, 5) structures may be shown in stages from the farthest to the farthest structure.

  In addition, as described above, since autism often does not look at the eyes of other people who face each other in early childhood, when the subject 10 is diagnosed as to whether or not there is a possibility of autism For example, the image selection unit 150 causes the image display unit 114 to display an inspection image 130b indicating a human face 132c, which is selected in accordance with an operation input by the inspector, as shown in FIG.

  The block dividing unit 152 divides an image (inspection image) displayed on the display surface 114a of the image display unit 114 into a plurality of blocks. FIG. 4 is a diagram for explaining processing by the block division unit 152. As shown in FIG. 4, the block dividing unit 152 converts the inspection image displayed on the display surface 114a of the image display unit 114 into the vertical direction (V-axis direction in FIG. 4) and the horizontal direction (in FIG. 4, In the (H-axis direction), the blocks are divided so as to be substantially equal in length. The block dividing unit 152 sets a unique address for each block (for example, H1 to H16 in the H-axis direction, V1 to V12 in the V-axis direction), and addresses the contents of the image displayed in the block. And stored in the storage unit 112. For example, V7 to V11 of address H11, V7 to V11 of H12, V7 to V11 of H13, V7 to V11 of H14, and V7 to V11 of H15 in the block are associated with the foreground structure 132a, and V2 to V3 of H11. , H12 V2-V3 are associated with distant view structure 132b.

  Further, the block dividing unit 152 does not necessarily divide the block into substantially the same size in all the blocks. For example, as shown in FIG. 4B, the block is a human face 132c in the inspection image. And other parts may have different sizes, or the parts of a human face (for example, eyes and mouth) and other parts may have different sizes.

  The pupil calculation unit 154 calculates the position of the pupil of the subject 10 (the center position of the pupil) based on the two diagnostic image data generated by the imaging units 118a and 118b. Here, since the diagnostic image data generated by the imaging unit 118a and the diagnostic image data generated by the imaging unit 118b have horizontal parallax, the pupil calculation unit 154 determines the position (coordinates) of the imaging unit 118a and a predetermined position. From the diagnostic image data generated by the imaging unit 118a, the predetermined position (coordinates) of the imaging unit 118b, and the diagnostic image data generated by the imaging unit 118b, using the triangulation method, the center position of the pupil ( Deriving coordinates).

  In addition, the pupil calculation unit 154 derives the position of the corneal reflection image (virtual image generated on the cornea when the eyeball is irradiated with light) in the vicinity of the pupil from the diagnostic image data generated by the imaging units 118a and 118b.

  The line-of-sight calculation unit 156 calculates line-of-sight information indicating the line of sight of the subject 10 based on the position of the pupil and the position of the cornea reflection image calculated by the pupil calculation unit 154.

  FIG. 5 is a diagram for explaining a corneal reflection image calculation process by the pupil calculation unit 154 and a line-of-sight calculation process by the line-of-sight calculation unit 156.

  As shown in FIG. 5, an infrared irradiation unit 116a is disposed on the right side as viewed from the eyeball EB, and an infrared irradiation unit 116b is disposed on the left side. An imaging unit 118a is disposed near the infrared irradiation unit 116a, and an imaging unit 118b is disposed near the infrared irradiation unit 116b.

  As described above, the cornea reflection image is a virtual image generated on the cornea CR when the eyeball EB is irradiated with light. Therefore, the cornea reflection image 170a generated by the infrared rays (light) irradiated by the infrared irradiation unit 116a exists on the connection LL1 between the curvature center 172 of the cornea CR and the infrared irradiation unit 116a (the light source of the infrared irradiation unit 116a). The position where the cornea reflection image 170a is generated is the midpoint between the cornea CR and the center of curvature 172 of the cornea CR on the connection LL1. Similarly, the cornea reflection image 170b generated by the infrared rays (light) irradiated by the infrared irradiation unit 116b exists on the connection LL2 between the curvature center 172 of the cornea CR and the infrared irradiation unit 116b (the light source of the infrared irradiation unit 116b). The position where the cornea reflection image 170b is generated is the midpoint between the cornea CR and the center of curvature 172 of the cornea CR on the connection line LL2.

  That is, the infrared irradiation unit 116a, the cornea reflection image 170a, and the curvature center 172 of the cornea CR are located on the same straight line (on the connection LL1), and the infrared irradiation unit 116b, the cornea reflection image 170b, and the curvature center 172 of the cornea CR are the same straight line. It is located on the line (on connection LL2).

  In addition, as described above, the infrared irradiation unit 116a emits infrared light of 850 nm, and the infrared irradiation unit 116b emits infrared light of 950 nm. The imaging unit 118a receives only 850 nm infrared light, and the imaging unit 118b receives only 950 nm infrared light. Therefore, the imaging unit 118a captures only the cornea reflection image 170a by the infrared irradiation unit 116a, and the imaging unit 118b captures only the cornea reflection image 170b by the infrared irradiation unit 116b.

  Here, since the infrared irradiation unit 116a is provided in the vicinity of the imaging unit 118a and the infrared irradiation unit 116b is provided in the vicinity of the imaging unit 118b, the position of the imaging unit 118a is regarded as the position of the infrared irradiation unit 116a. The position of the imaging unit 118b can be regarded as the position of the infrared irradiation unit 116b.

  Therefore, the line-of-sight calculation unit 156 uses a triangulation method to connect the predetermined imaging unit 118a and the corneal reflection image 170a (that is, the connection LL1), the predetermined imaging unit 118b, and the corneal reflection image 170b. The center of curvature 172 of the cornea CR is calculated by calculating the intersection with the connection (that is, the connection LL2).

  Further, it is known that a line connecting the eyeball EB center 174 and the center 176 of the pupil is a line indicating the line of sight. On the other hand, it is also known that the center of curvature 172 of the cornea CR is located on the line showing the line of sight. Therefore, a line LL3 passing through the coordinates of the pupil center 176 calculated by the pupil calculation unit 154 and the coordinates of the curvature center 172 of the cornea CR calculated by the line-of-sight calculation unit 156 is a line indicating the line of sight. Accordingly, the line-of-sight calculation unit 156 calculates, for example, the coordinates of the center of curvature 172 and the inclination of the line LL3 as line-of-sight information.

  Returning to FIG. 2, the gaze point deriving unit 158 derives whether or not there is a gaze point of the subject 10 for each of a plurality of blocks based on the gaze information calculated by the gaze calculation unit 156. Here, in order to reduce the processing load, the number of times the gazing point is included (hereinafter referred to as the number of gazing times) is counted up only for the block including the gazing point. More specifically, the gaze point derivation unit 158 determines the coordinates of the intersection (hereinafter simply referred to as a gaze point) between the line of sight indicated by the line-of-sight information calculated by the line-of-sight calculation unit 156 and the display surface 114a for a predetermined sampling time. The number of gazes is counted up for each block that includes the coordinates of the derived gaze point.

  The information storage unit 160 stores the result derived by the gaze point deriving unit 158 for each block. In the present embodiment, the information storage unit 160 stores the result in the storage unit 112 in association with the address set for each block.

  The calculation unit 162 performs a predetermined calculation based on the result accumulated by the information accumulation unit 160 and determines whether or not there is a possibility that the subject 10 has a brain function disease. The predetermined calculation used by the calculation unit 162 is any one selected from the group of a neural network, a naive Bayes, a Bayesian network, and a decision tree method.

  The calculation unit 162 performs calculation by weighting corresponding to the type of disease to be diagnosed based on the content of the image for each block. The arithmetic processing by the arithmetic unit 162 will be described in detail later.

  The sound output unit 122 converts sound data for prompting the subject 10 to view the image into sound and outputs the sound. The diagnostic display unit 124 is an inspection image in which all the gazing points within a predetermined time and an image showing a block are superimposed, and a trajectory of all the gazing points in a predetermined time and an image showing a block are superimposed. Display an image. The diagnostic display unit 124 is used when a doctor diagnoses a brain function disease or when the doctor explains a diagnosis result of the brain function disease to a person concerned (for example, a family member) of the subject 10.

(Brain function disease diagnosis support method)
FIG. 6 is a flowchart for explaining the flow of processing of the brain function disease diagnosis support method using the brain function disease diagnosis support apparatus 100.

  As shown in FIG. 6, first, in response to an operation input by the inspector, the image selection unit 150 causes the image display unit 114 to display an inspection image based on predetermined inspection image data (S200). The block dividing unit 152 divides the inspection image displayed on the display surface 114a of the image display unit 114 into a plurality of blocks, sets a unique address for each divided block, and displays the image displayed on the block. (For example, a foreground structure, a distant structure, a human eye, and a mouth) are associated with an address and stored in the storage unit 112 (S202).

  Subsequently, in response to an operation input by the inspector, the central control unit 120 causes the audio output unit 122 to output audio data in accordance with the inspection procedure recorded in advance in the storage unit 112 (S204). The voice output by the voice output unit 122 is, for example, “Please look at the nearest one” when diagnosing Alzheimer-type dementia. Then, the subject 10 starts gazing at the image display unit 114 based on the voice.

  Subsequently, the infrared irradiation units 116a and 116b emit infrared rays toward the subject 10 (S206), and the imaging units 118a and 118b generate diagnostic image data (S208). Note that the infrared irradiation by the infrared irradiation units 116a and 116b and the generation of diagnostic image data by the imaging units 118a and 118b are continued until it is determined in the inspection end determination S222 described later that the inspection has ended.

  Then, based on the diagnostic image data generated in the image generation process S208, the pupil calculation unit 154 calculates the center position of the pupil (S210), and calculates the position of the corneal reflection image (S212). The line-of-sight calculation unit 156 calculates line-of-sight information indicating the line of sight based on the center position of the pupil and the position of the corneal reflection image (S214).

  The gaze point deriving unit 158 derives the coordinates of the gaze point (S216), and determines whether or not the coordinates of the gaze point are within the display surface 114a (S218). If the coordinates of the gaze point are within the display surface 114a (YES in S218), the number of gazes n of the block corresponding to the coordinates of the gaze point is counted up, and the information storage unit 160 stores the gaze number n in the storage unit 112. (S220). Here, the information accumulating unit 160 also accumulates the movement order s in which the coordinates of the gazing point move the block. The movement order s is derived by counting up when the gazing point moves across blocks.

  Then, the central control unit 120 determines whether or not there is an instruction to end the inspection from the inspector or whether or not a predetermined inspection time has ended (S222). If the examination is not completed (NO in S222), the process waits until a predetermined sampling time SI elapses (NO in S224). When the sampling time SI elapses (YES in S224), the processes after the pupil position calculation process S210 are performed. repeat.

  On the other hand, when the inspection is completed (YES in S222), the calculation unit 162 performs a calculation based on the result accumulated in the result accumulation process S220 (S230).

  Here, a case where the calculation unit 162 performs a calculation based on a neural network will be described as an example.

  In the present embodiment, a neural network is adopted as a model imitating a neural circuit of a human brain. In a neural network, learning is performed on a known input signal so that a known output can be obtained, and a network having an input / output relationship that outputs a desired value for a given input is realized. ing. In the present embodiment, a plurality of blocks are configured in a network structure, and the load of the block is set by learning, and the output is derived from the number of gazes n and the gaze time t for each block.

FIG. 7 is a diagram for explaining the relationship between a block and a neural network. As shown in FIG. 7, each block of the inspection image 130a corresponds to the number of gazes n and the gaze time t stored in the storage unit 112 by the information storage unit 160, and the following formula (1) is set. doing. The gaze time t can be calculated by adding the sampling time SI to the number of gazes n.
... Formula (1)

As shown in FIG. 7, the value of each block of the inspection image 130a can be expressed by the following mathematical formula (2).
... Formula (2)

If the neural network is a hierarchical type having N layers, the first layer is an input layer and the Nth layer is an output layer. The layer between them is an intermediate layer. If the number of neurons in the nth layer is Ln, the output of each neuron can be expressed by the following equation (3).
... Formula (3)
However, since the neurons in the input layer only transmit the input signal to the intermediate layer, and the 0th neuron in each layer always outputs 1, Equation (3) can be expressed by Equations (4) and (5) below. Become.
... Formula (4)
... Formula (5)

The output signals of other neurons can be calculated using the following formulas (6) and (7).
... Formula (6)
The function of the output signal is a sigmoid function as shown in the following formula (8).
... Formula (8)

When using the error back-propagation method, which is a learning algorithm used for multi-layer perceptron learning, if the teacher signal in the output signal is y, then the scale representing how much the output of the output layer is close to the teacher signal is as follows: Learning is performed by defining the square error E in the equation (9).
... Formula (9)

That is, the weight W (see the following formula (10)) may be determined so that the square error E approaches 0.
... Formula (10)

  And the calculating part 162 calculates the possibility of a brain function disease with the neural network which performed learning as demonstrated above.

  Here, the load (weighting) of the test image in the neural network may be performed corresponding to the type of disease to be diagnosed. For example, in the examination of Alzheimer-type dementia, when the examiner gives an instruction “please see a distant object”, the load of the block associated with the image showing the distant view is set small. By carrying out like this, in the test | inspection of Alzheimer-type dementia, the ratio diagnosed as having Alzheimer-type dementia can be set high. Therefore, it is possible to reduce the possibility of being misdiagnosed as not suffering from Alzheimer-type dementia despite having Alzheimer-type dementia.

  Further, in the examination of autism, a large load is set on a block associated with human eyes and images near the eyes. By carrying out like this, in the test | inspection of autism, the ratio diagnosed as not having autism can be set high. Therefore, it is possible to reduce the possibility of being misdiagnosed as suffering from autism even though it is not suffering from autism.

  As described above, according to the brain function disease diagnosis support apparatus 100 and the brain function disease diagnosis support method according to the present embodiment, the position of the gaze point of the subject 10 is derived with a simple configuration, and the gaze point is determined. By analyzing, it is possible to quantitatively derive the possibility of suffering from a brain function disease.

(Modification)
In the above-described embodiment, the case where the neural network is employed as an example of the computation method used by the computation unit 162 has been described as an example. However, other computation methods may be employed. In the modification, the case where the calculation unit 162 performs calculation using naive Bayes will be described as an example.

  FIG. 8 is a flowchart for explaining the processing flow of the brain function disease diagnosis support method according to the modification. Note that the processes from the image display process S200 to the sampling period elapsed determination process S224 described above are substantially the same, and thus the same reference numerals are given and description thereof is omitted.

  As shown in FIG. 8, when the processing from the image display process S200 to the sampling period elapsed determination process S224 is completed, the arithmetic unit 162 determines the coordinates of the gazing point in each block based on the result accumulated in the result accumulation process S220. The time t continuously included in the block is calculated, the time t is a predetermined time (for example, 0.5 seconds) or more, and the number of times that is the predetermined time or more is the predetermined number of times or more. A block is detected (S232). Then, the calculation unit 162 creates a gaze block pattern shown in FIG. 9 (S234). Here, as shown in FIG. 9, the block dividing unit 152 divides the inspection image 130 a in a table form and shows a plurality of blocks in the form of a table. A block (FIG.) In which either or both satisfy a predetermined condition is indicated by Y. Here, the gaze block pattern is stored in the storage unit 112 as a table.

  Subsequently, the calculation unit 162 calculates the possibility of the disease using a naive Bayes classifier. More specifically, as shown in FIG. 10, the central control unit 120 has a conditional probability P (c | x) based on the disease block pattern Cx that the subject 10 having a certain brain function disease X will take. Is set in advance for each block Ci.

For example, as shown in FIG. 9, in the case where disease classification is performed in a state where a detection result indicating that the coordinates of the gazing point are included is input in each block of the test image 130 a, a certain brain function disease X The probability that the subject 10 is diagnosed as having Alzheimer-type dementia by a block (gaze block) that matches the disease block pattern Cx in the test image 130a is represented by P (x | c), If a certain prior probability is P (x), the probability P (x | c) to be diagnosed can be calculated by the following formula (11) according to Bayes' theorem.
... Formula (11)

Since the inspection image 130a is divided as a block Ci, the right side of the equation (11) is replaced by the following equations (12) and (13) and expressed by the equation (14). The probability of the possibility of a brain function disease is calculated by using (S236).
... Formula (12)
... Formula (13)
... Formula (14)
Then, the result P (x _j | C _j ) of the j-th test is used as the next prior probability P (x _{j + 1} ), and a predetermined probability (for example, the possibility of Alzheimer type dementia is 20). Less than% or 80% or more) (NO in S238), the processes after the image display process S200 are repeated.

  As described above, the information about where the subject 10 is gazing at the examination image 130a, how long it takes to reach the point to be gazed, and how long it is gazing is consistent with the disease block pattern Cx Or it is reflected in the prior probability by the conditional probability of the deviating block.

  For example, when the possibility of Alzheimer's dementia is less than 20% or 80% or more (YES in S238), the brain function disease diagnosis support method is terminated.

  As described above, the calculation unit 162 performs calculation using naive Bayes, so that the possibility of suffering from Alzheimer-type dementia can be diagnosed quickly and accurately. Note that the calculated numerical value indicating the possibility of morbidity is not absolute, and the examiner may diagnose whether or not the afflicted person is using the numerical value.

  In addition, the conditional probability of the test image in the Naive Bayes method is changed according to the operation input by the examiner when brain functional diseases such as Alzheimer-type dementia, autism, and depression are examined. Is done. For example, in the examination of Alzheimer-type dementia, when the examiner gives an instruction “please see a nearby object”, the conditional probability of a block associated with an image showing a near view is set small. By carrying out like this, in the test | inspection of Alzheimer-type dementia, the ratio diagnosed as having Alzheimer-type dementia can be set high. On the other hand, by setting a large conditional probability of a block associated with an image showing a foreground, it is possible to set a low ratio of being diagnosed as suffering from Alzheimer's dementia.

  Also, in testing for autism, by setting a large conditional probability for blocks that are associated with human eyes and images near the eyes, the proportion of people diagnosed with autism is reduced. Can be set to

  Further, the calculation of the possibility of the brain function disease may be calculated using only the conditional probability of the block in which the gaze block pattern and the disease block pattern Cx coincide.

  As described above, even if the calculation method according to the modified example is used, the position of the gazing point of the subject 10 is derived with a simple configuration, and the gazing point is analyzed by analyzing the gazing point. The possibility of being present can be derived quantitatively.

  Furthermore, a brain function disease diagnosis support program for causing a computer to function as the brain function disease diagnosis support device 100, a computer-readable flexible disk, magneto-optical disk, ROM, EPROM, EEPROM recording the brain function disease diagnosis support program Recording media such as CD, DVD, and BD are also provided. Here, the program refers to data processing means described in an arbitrary language or description method.

  As mentioned above, although preferred embodiment of this invention was described referring an accompanying drawing, it cannot be overemphasized that this invention is not limited to this embodiment. It will be apparent to those skilled in the art that various changes and modifications can be made within the scope of the claims, and these are naturally within the technical scope of the present invention. Understood.

  For example, the method is not limited to the gaze calculation method described above, and other methods may be adopted as long as the gaze point of the subject 10 can be derived.

  In the above-described embodiment, the case where a neural network or naive Bayes is employed as an example of the calculation method used by the calculation unit 162 has been described. However, calculation is performed using a Bayesian network or a decision tree method. Also good.

  For example, as described in `` Outline of Bayesian Network by Kazuo Shigeru '', when a Bayesian network observes some variables, it obtains the probability distribution for the other variables, and determines the state with the highest probability value. This is an arithmetic technique that can be obtained as a variable prediction result. That is, from the observed variable information (e), the probability of the random variable (X) to be obtained, that is, the posterior probability P (X | e) is obtained, and the X expected value or the maximum value of the posterior probability, a certain hypothesis is obtained from the posterior probability. Of certainty (simultaneous probability that several variables take a specific set of values) and the like.

  When the decision tree method is used, a tree diagram or a diagram called a tree structure is created, and a rule for classifying the output is created. Here, a tree structure may be configured in which nodes are related to whether or not blocks are watched as instructed in association with inspection images.

  Note that each step in the brain function disease diagnosis support method of the present specification does not necessarily have to be processed in time series in the order described in the flowchart, and may include processing in parallel or by a subroutine.

  The present invention can be used for a brain function disease diagnosis support apparatus and a brain function disease diagnosis support method for supporting diagnosis of a brain function disease.

DESCRIPTION OF SYMBOLS 100 ... Brain function disease diagnosis support apparatus 114 ... Image display part 114a ... Display surface 116a, 116b ... Infrared irradiation part 118a, 118b ... Imaging part 152 ... Block division part 154 ... Pupil calculation part 156 ... Gaze calculation part 158 ... Gaze point derivation Unit 160 ... Information storage unit 162 ... Calculation unit

Claims (3)

An image display unit that displays an image for the subject to visually recognize;
A block dividing unit that divides an image displayed on the display surface of the image display unit into a plurality of blocks;
An infrared irradiation unit that emits infrared rays toward the subject; and
Two imaging units for imaging at least the reflected infrared light reflected by the subject and generating two image data having horizontal parallax;
A pupil calculation unit that calculates the position of the pupil of the subject based on the two image data generated by the imaging unit;
A line-of-sight calculation unit that calculates line-of-sight information indicating the line of sight of the subject, based on the position of the pupil calculated by the pupil calculation unit;
Based on the line-of-sight information calculated by the line-of-sight calculation unit, for each of the plurality of blocks, a gaze point derivation unit that derives whether or not there is a gaze point of the subject;
An information accumulating unit for accumulating the result derived by the gaze point deriving unit for each block;
A calculation unit that performs a predetermined calculation based on the result accumulated by the information storage unit and determines whether or not there is a possibility of a brain function disorder of the subject; and
Equipped with a,
The image displayed by the image display unit in the block is associated with each of the plurality of blocks,
The brain function disease diagnosis support apparatus, wherein the calculation unit performs the predetermined calculation with weighting corresponding to the type of disease to be diagnosed based on the content of the image for each block . The brain function disease diagnosis support apparatus according to claim 1, wherein the predetermined calculation is any one selected from the group of a neural network, a naive Bayes, a Bayesian network, and a decision tree method. A brain function disease diagnosis support method using an image display unit that displays an image for a subject to visually recognize,
The image display unit displays an image for the subject to visually recognize,
Dividing the image displayed on the display surface in the image display unit into a plurality of blocks;
The image displayed by the image display unit in the block is associated for each of the plurality of blocks,
Irradiate the subject with infrared rays,
Imaging at least the reflected infrared light reflected by the subject to generate two image data having a horizontal parallax;
Based on the generated two image data, the pupil position of the subject is calculated,
Based on the calculated position of the pupil, calculates gaze information indicating the gaze of the subject,
Based on the calculated line-of-sight information, for each of the plurality of blocks, derive whether there is a gaze point of the subject,
Accumulate the derived results for each block,
Referring to accumulated the result, the diagnosis target and brain function disorder diagnosis support method by weighting corresponding to the type of disease characterized by the TURMERIC line a predetermined operation to be based on the contents of the image of each block.