JP2021144603A - Wound examination support method, trained model generation method, wound examination support device, and computer program - Google Patents

Abstract

To provide a wound examination support method, a trained model generation method, a wound examination support device and a computer program, capable of supporting examination of wounds by a doctor with little knowledge or experience on foot wounds.SOLUTION: A wound examination support method disclosed herein comprises acquiring inspection information on a foot wound and, upon entering the inspection information on the foot wound, inputting the acquired inspection information to a trained model designed to output information on an examination result including a cause(s) of the wound, and outputting the examination result.SELECTED DRAWING: Figure 1

Description

The present invention relates to a wound care support method, a trained model generation method, a wound care support device and a computer program.

With the increase in the aging rate and the number of patients with lifestyle-related diseases, the number of patients with intractable foot lesions (chronic wounds in the lower limbs) is also increasing. Chronic wounds in the lower limbs (hereinafter referred to as "foot wounds") include various pathological conditions, and conventionally, wound treatment based on the doctor's empirical judgment based on macroscopic observation by the doctor has been performed (patented). Reference 1).

Japanese Unexamined Patent Publication No. 10-206424

However, the wound on the foot may be affected by various pathological conditions in a complex manner, and the clinical department differs depending on the pathological condition. If medical treatment is performed by a doctor who has little knowledge or experience regarding foot wounds, correct diagnosis and treatment cannot be performed, and as a result, wounds at high risk of aggravation may be overlooked, resulting in aggravation of the wound.

The present invention has been made in view of such circumstances, and is a wound treatment support method, a learned model generation method, and a wound treatment support device that can support wound treatment by a doctor who has little knowledge or experience regarding wounds on the legs. And to provide computer programs.

The wound medical care support method according to the embodiment of the present invention is a learned model that acquires test information on a foot wound and outputs information on a medical care result including the cause of the wound when the test information on the foot wound is input. The acquired test information is input to, and the medical treatment result is output.

The wound medical treatment support method according to the embodiment of the present invention is a learned model that acquires test information including an image of a foot wound and outputs information on the medical care result when the test information on the foot wound is input. The acquired test information is input to, and the medical treatment result is output. As the medical treatment result, the cause of the wound including arterial ulcer, venous ulcer, diabetic ulcer and compression ulcer, the estimated accuracy of each cause, and infection. Shows the estimated accuracy of the wound, as well as the recommended treatment method and recommended department for the wound.

The trained model generation method according to the embodiment of the present invention acquires a plurality of diagnostic images in which the cause of a wound on the foot is diagnosed, and identifies the cause of the wound and the presence or absence of infection corresponding to each diagnostic image. A trained model is generated that acquires a label and uses the acquired diagnostic image and a specific label to output information about the medical outcome, including the cause of the wound.

The wound medical treatment support device according to the embodiment of the present invention outputs information on the medical treatment result including the cause of the wound when the acquisition unit for acquiring the inspection information on the wound on the foot and the inspection information on the wound on the foot are input. The trained model is provided with an output unit that inputs acquired test information and outputs medical results.

The wound medical treatment support device according to the embodiment of the present invention outputs the information on the medical treatment result when the acquisition unit for acquiring the inspection information including the image of the wound on the foot and the inspection information on the wound on the foot are input. By inputting the test information acquired by the acquisition unit into the trained model, the causes of wounds including arterial ulcer, venous ulcer, diabetic ulcer and compression ulcer and the causes of each cause can be used as information on the clinical results. It is provided with an estimated probability, an estimated probability of infection, and a display unit for displaying a recommended treatment method for a wound and a recommended clinical department.

The computer program according to the embodiment of the present invention has been learned to acquire test information on foot wounds and output information on medical results including the cause of the wound when the test information on foot wounds is input to the computer. Input the acquired test information to the model, output the medical treatment result, and execute the process.

According to the present disclosure, it is possible to support the wound treatment of a doctor who has little knowledge and experience regarding the wound on the foot.

It is a block diagram which shows an example of the structure of the wound medical treatment support device of this embodiment. It is a schematic diagram which shows an example of a wound of a foot. It is a schematic diagram which shows the 1st example of the learning method of a learner. It is a schematic diagram which shows the 2nd example of the learning method of a learner. It is explanatory drawing which shows an example of the relation between the wound evaluation information and the cause of a wound. It is a schematic diagram which shows the 3rd example of the learning method of a learner. It is explanatory drawing which shows an example of the relation between the blood flow evaluation information and the cause of a wound. It is explanatory drawing which shows an example of the relationship between the body fluid component value and the wound evaluation information, and the healing period / healing rate. It is a schematic diagram which shows the 4th example of the learning method of a learner. It is a schematic diagram which shows the 1st example of the method of estimating the medical treatment result by a learning device. It is a schematic diagram which shows the 2nd example of the method of estimating the medical treatment result by a learning device. It is a schematic diagram which shows the 3rd example of the method of estimating the medical treatment result by a learning device. It is explanatory drawing which shows an example of the relationship between the cause of a wound and the treatment method. It is a schematic diagram which shows the 1st display example of the medical treatment result by the wound medical treatment support device. It is a schematic diagram which shows the 2nd display example of the medical treatment result by the wound medical treatment support device. It is a flowchart which shows an example of the processing procedure of the method of generating a learning device by a wound medical treatment support device. It is a flowchart which shows an example of the processing procedure of the method of estimating the medical treatment result by the wound medical treatment support device.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing an example of the configuration of the wound medical care support device 50 of the present embodiment. The wound medical care support device 50 includes a control unit 51 that controls the entire device, an interface unit 52, an image processing unit 53, a storage unit 54, a learning device 55 as a learning model, a display panel 56, and an operation unit 57. The control unit 51 can be composed of a CPU, a ROM, a RAM, and the like. A camera 10 and a plurality of measuring instruments 20, 21, ... Can be connected to the wound medical care support device 50. The wound medical care support device 50 and the camera 10 and the measuring instruments 20, 21, ... May be connected via a communication network, or may be directly connected by a wired cable. The wound medical treatment support device 50 may be installed in a medical examination room of a hospital or a clinic, or may be installed in another management room or the like. When the wound medical treatment support device 50 is installed in a room other than the examination room, a terminal device (for example, a personal computer, a tablet terminal, etc.) that is connected to the wound medical treatment support device 50 and has a display panel should be installed in the examination room. Just do it.

The camera 10 is installed in, for example, a medical examination room or an examination room of a hospital or a clinic, and can take a picture of a patient's foot. The measuring instruments 20, 21, ... Are devices that are installed in a laboratory or the like and can measure a patient's blood flow evaluation information, body fluid component values, and the like. Details of blood flow evaluation information and body fluid component values will be described later.

The interface unit 52 has an interface function between the camera 10 and the measuring instruments 20, 21, .... The interface unit 52 can acquire an image (image data) taken by the camera 10 and data (measurement data) measured by the measuring instruments 20, 21, ....

The image processing unit 53 can perform the required image processing. The details of the image processing unit 53 will be described later.

The storage unit 54 can be configured by a semiconductor memory, a hard disk, or the like, and can store image data and measurement data acquired via the interface unit 52. In addition, the storage unit 54 can store the data obtained as a result of the processing performed by the wound medical treatment support device 50 and the data in the process of processing.

The learner 55 can be configured by, for example, a convolutional neural network (CNN), a deep neural network (DNN), or a combination thereof, but other machine learning can also be used. good. The learner 55 is, for example, hardware such as a CPU (for example, a multi-processor in which a plurality of processor cores are mounted), a GPU (Graphics Processing Units), a DSP (Digital Signal Processors), and an FPGA (Field-Programmable Gate Arrays). Can be configured by combining. The details of the learner 55 will be described later. In this specification, the learning device 55 includes those before learning, those in the middle of learning, and those that have already been learned.

The display panel 56 can be composed of a liquid crystal panel, an organic EL (Electro Luminescence) display, or the like. The operation unit 57 can be configured by, for example, a touch panel incorporated in the display panel 56, and can perform a predetermined operation on an icon or the like displayed on the display panel 56. Further, the operation unit 57 may be a hardware keyboard, a mouse, or the like.

FIG. 2 is a schematic view showing an example of a wound on the foot. Wounds refer to physical damage to body surface tissue caused by external or internal factors, especially intractable foot lesions (chronic wounds in the lower extremities), which include a variety of pathologies and are experienced in foot wounds. Medical treatment by a good doctor is indispensable. On the other hand, it may be difficult for a doctor who has little knowledge and experience about foot wounds to make a correct diagnosis and treatment. The wound medical treatment support device 50 can output the examination result of the wound on the foot of the patient, and can support the wound medical treatment of a doctor who has little knowledge and experience about the wound on the foot.

In FIG. 2, the shaded area is a schematic representation of the wound. The actual wound may differ from that shown in FIG. FIG. 2A is an image P of the back of the patient's foot, which has a wound on the instep and toes. FIG. 2B is an image Q of the sole of the patient, which has a wound on the heel and the base of the finger. FIG. 2C is an image R of the patient's lower leg with a wound on the upper side of the ankle. The image R includes an image of the inside of the lower leg and an image of the outside of the lower leg, depending on the location of the wound. Images P, Q, and R are taken including the wound and the skin around the wound.

By photographing the wound with the camera 10 according to the location of the wound on the patient's foot, the camera 10 can output at least one image of the images P, Q, and R to the wound medical care support device 50. For example, when the patient's wound is present on the sole, sole, and lower leg, the images P, Q, and R are output to the wound medical care support device 50. If there is leaked liquid from the wound, the leaked liquid may be wiped off and photographed.

Next, a learning method of the learning device 55, that is, a method of generating the learning device 55 will be described.

FIG. 3 is a schematic diagram showing a first example of a learning method of the learning device 55. FIG. 3 shows an example in which a learning device 55 composed of CNN551 is generated by using the diagnostic image Pt of the sole of the foot, the diagnostic image Qt of the sole, and the diagnostic image Rt of the lower leg. The diagnostic image is an image in which the diagnosis result is shown by a wound specialist, and the diagnosis result by the specialist includes, for example, the cause of the wound (for example, arterial ulcer, venous ulcer, diabetic ulcer, compressive ulcer and others. (Can be classified into 5 categories), presence or absence of infection, wound severity (for example, can be classified as severity I (mild), severity II (moderate), severity III (severe)), healing Findings such as difficulty (for example, it can be expressed by the wound healing period divided into 1 month, 2 months, 3 months, etc., or the healing rate divided into 50%, 70%, 90%, etc.) included. Learning of CNN551 is performed by collecting a large amount of diagnostic images Pt, Qt, and Rt.

FIG. 3A shows the diagnostic images Pt, Qt, and Rt in which the cause of the wound is diagnosed as arterial ulcer (ischemic), the cause of the wound corresponding to each diagnostic image, the presence or absence of infection, and the severity of the wound. CNN551 can be generated by using the specific label to be specified as training data. In this case, the specific label is a teacher label indicating that the cause of the wound is an arterial ulcer (for example, label = 1), a teacher label indicating the presence or absence of infection (for example, label = 6 if infected, no infection). In the case of, the label = 7), and the teacher label indicating the severity (in the case of severity I, II, III, the label = 8, 9, 0, respectively) can be used. When the diagnostic images Pt, Qt, and Rt are input to the CNN551, an identification flag indicating whether the input diagnostic image is the sole, the sole, or the lower leg may also be input to the CNN551.

FIG. 3B shows the diagnostic images Pt, Qt, and Rt in which the cause of the wound is diagnosed as venous ulcer (congestive), the cause of the wound corresponding to each diagnostic image, the presence or absence of infection, and the severity of the wound. CNN551 can be generated by using the specific label to be specified as training data. In this case, the specific label can be a teacher label (eg, label = 2) indicating that the cause of the wound is a venous ulcer. Other specific labels are the same as in FIG. 3A.

FIG. 3C shows diagnostic images Pt, Qt, and Rt in which the cause of the wound is diagnosed as diabetic ulcer (neuropathy), the cause of the wound corresponding to each diagnostic image, the presence or absence of infection, and the severity of the wound. CNN551 can be generated by using a specific label for specifying the above as training data. In this case, the specific label can be a teacher label (eg, label = 3) indicating that the cause of the wound is a diabetic ulcer. Other specific labels are the same as in FIG. 3A.

FIG. 3D shows diagnostic images Pt, Qt, and Rt in which the cause of the wound is diagnosed as compressive ulcer, and specific labels that identify the cause of the wound, the presence or absence of infection, and the severity of the wound corresponding to each diagnostic image. Can be used as training data to generate CNN551. In this case, the specific label can be a teacher label (eg, label = 4) indicating that the cause of the wound is a compressive ulcer. Other specific labels are the same as in FIG. 3A.

FIG. 3E generates CNN551 using the diagnostic images Pt, Qt, and Rt diagnosed as having the cause of the wound as “other” and the specific label identifying the cause of the wound corresponding to each diagnostic image as training data. can do. In this case, as the specific label, a teacher label (for example, label = 5) indicating that the cause of the wound is “other” can be used.

Although not shown, if the findings of "difficulty in healing" are obtained by a specialist doctor corresponding to each of the diagnostic images Pt, Qt, and Rt, the difficulty in healing is determined. A specific label to be specified can also be used as a teacher label.

Further, for the diagnostic images Pt, Qt, and Rt, both an image with a wound and an image without a wound are included.

The CNN551 can output information on the medical treatment result when at least one of the images P, Q, and R of the wound on the patient's foot is input. Information about the outcome of the procedure includes the cause of the wound, the presence or absence of infection, and the severity.

FIG. 4 is a schematic diagram showing a second example of the learning method of the learning device 55. FIG. 4 shows an example in which a learner 55 composed of DNN552 is generated by using the wound evaluation information. Wound evaluation information is information representing the evaluation of a wound on the foot, and the wound can be evaluated by parameters such as the position of the wound, the shape of the wound, the skin around the wound, the leaked fluid, and the hair growth of the toes.

FIG. 5 is an explanatory diagram showing an example of the relationship between the wound evaluation information and the cause of the wound. If the location of the wound is toe, ankle, heel, etc., there is a possibility of arterial ulcer as the cause of the wound, and if the location of the wound is ankle, lower leg, etc., the cause of the wound is venous ulcer. If there is a possibility and the position of the wound is the sole of the foot etc., there is a possibility of diabetic ulcer as the cause of the wound, and if the position of the wound is the heel, ankle, ball of the finger, etc., the cause of the wound As there is a possibility of compressive ulcer. The shape of the wound, the skin around the wound, the leaked fluid, and the hair growth of the toes may also be related to the cause of the wound, as shown in FIG.

When the wound evaluation information is input to DNN552, the state of each parameter of the wound evaluation information (wound position, wound shape, skin around the wound, leaked fluid, hair growth of toes) is quantified and the wound is evaluated. The information may be converted into a vector and input to DNN552.

As shown in FIG. 4, DNN552 identifies wound evaluation information indicating the evaluation of a wound on the foot, and the cause of the wound, the presence or absence of infection, the severity of the wound, the healing period, and the healing rate corresponding to the wound evaluation information. It can be generated by using a specific label as training data. Wound evaluation information is information in which the diagnosis result is shown by a wound specialist, and the diagnosis result by the specialist includes, for example, the cause of the wound (arterial ulcer, venous ulcer, diabetic ulcer, compression ulcer and others). , Presence or absence of infection, severity of wound (severity I (mild), severity II (moderate), severity III (severe)), difficulty of healing (eg, 1 month, 2 months, 3 months, etc.) It can be expressed by the wound healing period classified into, or the healing rate classified into 50%, 70%, 90%, etc.). Learning of DNN552 is performed by collecting a large amount of wound evaluation information.

As the specific label, a teacher label (for example, label = 11, 12, 13, etc.) indicating that the healing period is, for example, 1 month, 2 months, 3 months, etc. can be used, and the healing rate is, for example, Teacher labels (eg, labels = 21, 22, 23, etc.) indicating that they are 50%, 70%, 90%, etc. can be used. Other specific labels are the same as in the case of FIG.

The wound evaluation information can be extracted using an image processing device (for example, the image processing unit 53 or a separate image processing device). For example, an image of a wound on a foot can be extracted by performing necessary image processing. Some of the leaked fluid can be obtained by macroscopic observation. The position of the wound can be identified from the captured image, and the shape, size, color, and depth (depth of focus) of the wound can be extracted. In addition, the color and luster of the skin around the wound, redness and swelling of the skin, and the dry state of the skin can be extracted from the captured image. In addition, the presence or absence of hair growth on the toes can be extracted from the captured image. In this way, various conditions such as the position of the wound, the shape of the wound, the skin around the wound, the leaked fluid, and the hair growth of the toes can be determined in a complex manner, and the cause of the wound, the presence or absence of infection, and the difficulty of healing are determined. Etc. can be estimated. Further, the wound evaluation information may be acquired by macroscopic observation by a doctor, and the acquired wound evaluation information may be input from a terminal device or the like and output to the wound medical treatment support device 50.

The DNN552 can output information on the medical treatment result when the wound evaluation information is input. Information on the outcome of the treatment includes the cause of the wound, the presence or absence of infection, the severity of the wound, the healing period and healing rate of the wound.

FIG. 6 is a schematic diagram showing a third example of the learning method of the learning device 55. FIG. 6 shows an example in which the learning device 55 composed of DNN552 is generated by using the blood flow evaluation information and the body fluid component value in addition to the wound evaluation information in the second example shown in FIG. The blood flow evaluation information can be parameters such as pulse pressure of the foot, skin perfusion pressure, blood flow and blood flow velocity of the foot, skin surface temperature of the foot, and tissue oxygen saturation in the vicinity of the wound. The body fluid component value includes, for example, a CRP value (C-reactive protein value), a serum albumin value, a mineral, a vitamin, a cell growth factor, and the like.

FIG. 7 is an explanatory diagram showing an example of the relationship between the blood flow evaluation information and the cause of the wound. As shown in FIG. 7, regarding the pulse pressure of the foot, if the ABI value is less than 0.9, the cause of the wound may be arterial ulcer, and if the SPP value is less than 50 mmHg, the cause of the wound may be. It may be an arterial ulcer. Regarding the pulse pressure, the strength of the pulse pressure can be determined by palpation. Each parameter of the blood flow evaluation information can be measured using the respective measuring device. Foot blood flow and velocity, foot skin surface temperature, and tissue oxygen saturation near the wound may also be associated with arterial ulcers. Further, the blood flow evaluation information may be output from the measuring device to the wound medical care support device 50, or the data obtained by the measuring device may be input from a terminal device or the like and output to the wound medical care support device 50. ..

FIG. 8 is an explanatory diagram showing an example of the relationship between the body fluid component value and the wound evaluation information and the healing period / healing rate. As shown in FIG. 8, if the CRP value is greater than 5.0 mg / dL, the wound may be difficult to heal. Also, if the skin around the wound is red or swollen, the wound is large and deep, the wound may be difficult to heal. The body fluid component value can be obtained by a blood test, a urine test, a spectrometer, or the like.

As shown in FIG. 6, the DNN552 identifies wound evaluation information indicating the evaluation of a wound on the foot, and the cause of the wound, the presence or absence of infection, the severity of the wound, the healing period, and the healing rate corresponding to the wound evaluation information. It can be generated by using a specific label as training data. Further, the DNN552 can be generated by using the blood flow evaluation information indicating the evaluation of the blood flow of the foot and the specific label for identifying the cause of the wound corresponding to the blood flow evaluation information as training data. In this case, the particular label can be a teacher label indicating whether or not the cause of the wound on the foot is an arterial ulcer.

Further, the DNN552 can be generated by using the body fluid component value and the specific label corresponding to the body fluid component value as training data. The specific label can be, for example, a teacher label indicating the presence or absence of infection, a teacher label indicating the healing period of the wound, and a teacher label indicating the healing rate of the wound.

When the blood flow evaluation information and the body fluid component value are input to the DNN552, each parameter of the blood flow evaluation information and each parameter of the body fluid component value may be quantified, converted into a vector, and input to the DNN552.

The DNN552 can output information on the medical treatment result when the wound evaluation information, the blood flow evaluation information, and the body fluid component value are input. Information on the outcome of the treatment includes the cause of the wound, the presence or absence of infection, the severity of the wound, the healing period and healing rate of the wound. In particular, by further considering the blood flow evaluation information, it is possible to make a definitive diagnosis as to whether or not the cause of the wound is an arterial ulcer. Wounds caused by arterial ulcers (ischemic) are at particularly high risk of aggravation, and without proper treatment at the right time, debridement and small amputations are required, and ischemia When it becomes widespread, it is necessary to make a large amputation to save lives. For this reason, it is especially important for doctors to diagnose whether or not the cause of the wound is an arterial ulcer, and for doctors who have little knowledge or experience about foot wounds, important medical support is provided. Can be provided.

FIG. 9 is a schematic diagram showing a fourth example of the learning method of the learning device 55. FIG. 9 shows an example in which a learner 55 composed of DNN553 is generated by using blood flow evaluation information and body fluid component values. The difference from the third example of FIG. 6 is that the wound evaluation information is used as training data in the third example, whereas the wound evaluation information is not used in the fourth example. The specific label can be a teacher label indicating whether or not the cause of the wound is an arterial ulcer, a teacher label indicating the presence or absence of infection, and a teacher label indicating the healing period and the healing rate. As will be described later, the DNN553 shown in FIG. 9 can output information on the medical treatment result in combination with the CNN551 shown in FIG.

Although not shown, the presence or absence of possessed diseases (for example, lower limb obstructive arterial disease, diabetic foot ulcer, heart failure, renal failure, etc.), morbidity / wound period, and history of foot amputation were conducted by interview with a specialist. Alternatively, information such as tissue fluid / water content around the wound, muscle mass, location of obstructed blood vessels in the lower limbs, and degree of stenosis from the measured data may be used as training data.

The test information acquired when learning the learner 55 includes data such as a diagnostic image of a wound on the foot (an image in which the cause of the wound is diagnosed), a wound evaluation information, a blood flow evaluation information, and a body fluid component value. Etc. are included. In the present specification, the acquisition of the inspection information includes not only the case of acquiring the inspection information at the time of learning but also the case of using the inspection information acquired in advance.

Next, a method of estimating the medical treatment result of the wound on the foot will be described.

FIG. 10 is a schematic diagram showing a first example of a method of estimating a medical treatment result by the learning device 55. In the example of FIG. 10, the CNN 551 generated in the example of FIG. 3 is used as the learner 55. The control unit 51 acquires an image of the patient to be treated (an image of a wound on the foot), and inputs the acquired image to the CNN 551. The image of the patient to be treated can be, for example, the image Px of the back of the foot, the image Qx of the sole of the foot, and the image Rx of the lower leg. It is not necessary to take an image, and the image of the non-wounded portion does not have to be input to the CNN551, but in the sense of narrowing down the cause of the wound, the image of the non-wounded portion may be input to the CNN551.

CNN551 outputs each item of arterial ulcer, venous ulcer, diabetic ulcer, compressive ulcer, infection, severity, healing period, and healing rate as information on medical treatment results. For example, for an arterial ulcer, a numerical value indicating the possibility of an arterial ulcer can be output as an estimated accuracy. The numerical value indicating the possibility is, for example, a numerical value from 0 to 1, and for example, 0.8 indicates that the possibility of arterial ulcer is 80% certainty. The same applies to other wound causes and infections. As for the severity, for example, numerical values indicating the possibility of each of severity I (mild), severity II (moderate), and severity III (severe) can be output. As for the healing period, for example, it is possible to output a numerical value indicating each possibility of a period such as 1 month, 2 months, or 3 months. As for the healing rate, for example, it is possible to output a numerical value indicating the possibility of each ratio such as 50%, 70%, 90%.

As a result, even a doctor who has little knowledge and experience about foot wounds can accurately know the cause of the wound, the presence or absence of infection, and the severity of the wound based on the image of the patient's foot wound. In addition, the healing period and healing rate of wounds can be estimated.

FIG. 11 is a schematic diagram showing a second example of a method of estimating a medical treatment result by the learning device 55. In the example of FIG. 11, the DNN552 generated in the example of FIG. 6 is used as the learner 55. The control unit 51 acquires the wound evaluation information, the blood flow evaluation information, and the body fluid component value representing the evaluation of the wound on the foot of the patient to be treated, and inputs the acquired wound evaluation information, the blood flow evaluation information, and the body fluid component value into the DNN552. do. When the blood flow evaluation information and the body fluid component value of the patient to be treated are not used, the DNN552 generated in the example of FIG. 4 may be used.

The DNN552 can output each item of arterial ulcer, venous ulcer, diabetic ulcer, compressive ulcer, infection, severity, healing period, and healing rate as information on the medical treatment result. Since each of these items is the same as in FIG. 10, the description thereof will be omitted. As a result, even a doctor who has little knowledge and experience about foot wounds can accurately know the cause of the wound, the presence or absence of infection, and the severity of the wound based on the image of the patient's foot wound. In addition, the healing period and healing rate of wounds can be estimated. In particular, by using the blood flow evaluation information, it is possible to make a definitive diagnosis as to whether or not the cause of the wound is an arterial ulcer.

FIG. 12 is a schematic diagram showing a third example of a method of estimating a medical treatment result by the learning device 55. In the example of FIG. 12, as the learner 55, the CNN 551 generated in the example of FIG. 3 and the DNN 553 generated in the example of FIG. 9 are used in combination. The output of CNN551 and the output of DNN553 are combined by the output layer 55a and output. The image of the patient to be treated is the same as in FIG. 10, and the blood flow evaluation information and the body fluid analysis value of the patient to be treated are the same as in FIG. 11, so the description thereof will be omitted.

In the output layer 55a, the output value can be weighted. For example, as an output item of CNN551, the output value of "arterial ulcer" is 0.7, and as an output item of DNN553, the output value of "arterial ulcer" is 0.8, which is "not an arterial ulcer". The output value of "" is 0.2. Since the DNN553 outputs information on the medical treatment result by inputting the blood flow evaluation information, it is possible to make a definitive diagnosis. Therefore, weighting is performed so as to apply a bias so that the output value of the "arterial ulcer" of CNN551 increases, and the output value of the "arterial ulcer" of the output layer 55a is set to CNN551, for example, 0.8. Can be increased from the output value of. In this case, since it is highly possible that the cause of the wound is an arterial ulcer, the output value of the cause of another wound as an output item of CNN551 can be reduced and output from the output layer 55a.

On the contrary, it is assumed that the output value of "arterial ulcer" is 0.7 as the output item of CNN551, and the output value of "arterial ulcer" is 0.2 as the output item of DNN553, and "arterial ulcer". It is assumed that the output value of "not" is 0.8. In this case, since the possibility that the cause is arterial ulcer is low, weighting is performed so that the output value of "arterial ulcer" of CNN551 is reduced, and the output value of "arterial ulcer" of the output layer 55a is reduced. Can be reduced from the output value of CNN551, for example, 0.3.

As described above, the medical treatment result can be output by acquiring the examination information regarding the wound on the patient's foot and inputting the acquired examination information into the learning device 55. The examination information can be information that can be obtained by an examination, an interview, or the like performed when a doctor treats a wound. When the test information is input, the learner 55 can output information on the medical treatment result similar to that of an experienced doctor who specializes in wounds on the foot. In addition to the cause of the wound, the information on the medical treatment result can include the presence or absence of infection, the severity of the wound, and the like. As a result, even a doctor who has little knowledge and experience about wounds on the legs can accurately know the cause of the wounds and can perform correct diagnosis and treatment. It can support the wound care of poor doctors.

The test information to be acquired when estimating the medical treatment result of the wound of the foot includes an image of the wound of the foot, wound evaluation information, blood flow evaluation information, data such as a body fluid component value, and the like. In the present specification, the acquisition of test information includes not only the case of acquiring when estimating a medical treatment result but also the case of using the test information acquired in advance.

FIG. 13 is an explanatory diagram showing an example of the relationship between the cause of the wound and the treatment method. The control unit 51 can recommend a wound treatment method and a clinical department based on the cause of the wound estimated by the learner 55. As shown in FIG. 13, when the cause of the wound is an arterial ulcer, the treatment methods include revascularization (intravascular treatment, peripheral vascular bypass), antiplatelet drug, and wound treatment (external agent, wound dressing). ), Debridement, large cuts, etc. are included. For example, if the wound is caused by an arterial ulcer and the ischemic area is narrow, revascularization can be recommended if there is no infection. In addition, if the ischemic area is narrow and there is infection, debridement (removal of necrotic tissue) can be performed, and then revascularization can be recommended. After revascularization, wound treatment (prevention of infection with topical agents, control of exudate with wound dressing and promotion of granulation) can be recommended. In addition, it is possible to recommend a clinical department that specializes in each treatment.

For venous ulcers, diabetic ulcers, compressive ulcers, and other causes, treatment methods as shown in FIG. 13 can be recommended.

FIG. 14 is a schematic view showing a first display example of the medical treatment result by the wound medical treatment support device 50. The wound medical treatment support device 50 can display the medical treatment result on the display panel 56 or an external terminal device (not shown). As shown in FIG. 14, the area 101 displays the wound image of the patient, the area 102 displays the cause of the wound, the area 103 displays the severity of the wound, and the area 104 shows the severity of the wound. The recommended treatment method / department can be displayed, and the estimated value of the healing period / healing rate of the wound can be displayed in the area 105.

The wound image displayed in the area 101 can be enlarged and displayed as a desired wound portion by a doctor's operation (for example, pinch out). In region 102, arterial ulcers, venous ulcers, diabetic ulcers, compressive ulcers and other estimated probabilities (predicted probabilities, presumed certainty), and estimated probabilities of infection are displayed as causes of wounds. In the example of FIG. 14, it can be seen that the patient's wounds are 80% and 60% likely to be caused by arterial ulcers and diabetic ulcers, respectively, and the infection is also 70% likely.

By manipulating the name of the treatment method in region 104 (for example, tapping), it is possible to switch to a screen for explaining the detailed procedure of the treatment method, and the treatment method can be explained to the patient in an easy-to-understand manner.

In this way, even a doctor who has little knowledge and experience about wounds on the legs can accurately know the cause of the wounds and can perform correct diagnosis and treatment. Therefore, knowledge and experience about wounds on the legs. It can support the wound care of poor doctors. In addition, displaying the recommended treatment method and recommended clinical department of the wound facilitates explanation to the patient.

In addition, by displaying the captured wound image and the severity of the wound, the condition of the wound on the foot of the patient and the risk of aggravation can be easily confirmed.

In addition, by displaying the estimated wound healing period and the estimated healing rate, it is possible for the patient to understand how long it takes for the wound to heal and at what rate the wound is completely healed. , Can give the patient a sense of security.

FIG. 15 is a schematic view showing a second display example of the medical treatment result by the wound medical treatment support device 50. The area 101 displays the wound image of the patient taken at the time of the first medical examination or the previous medical examination, and the area 106 displays the wound image of the patient taken at the time of the re-examination or the current medical examination. The area 103 displays the severity of the wound at the time of the first medical examination or the previous medical examination, and the area 107 displays the severity of the wound at the time of the re-examination or the current medical examination. The period between the previous time and this time is a period determined by the doctor according to the condition of the wound, and may be, for example, 2 weeks or 1 month. The wound image of each region can be automatically adjusted to have the same size, angle, and layout so that comparison and control can be easily performed.

The wound medical care support device 50 can acquire images of the wound on the patient's foot taken at different time points, and input each of the acquired images to the learning device 55 to display the transition of the severity of the wound. can. For example, by showing the patient that the severity at this visit is less severe than at the previous visit, the patient should ensure that the wound is being treated successfully. Can be done.

FIG. 16 is a flowchart showing an example of a processing procedure of a method of generating the learning device 55 by the wound medical care support device 50. Hereinafter, for convenience, the subject of processing will be described as the control unit 51. The control unit 51 acquires a plurality of diagnostic images in which the cause of the wound on the foot is diagnosed (S11), and acquires blood flow evaluation information corresponding to each diagnostic image (S12).

The control unit 51 acquires the body fluid component value corresponding to each diagnostic image (S13), and acquires the specific label corresponding to each diagnostic image (S14). The specific label can be a teacher label indicating the cause of the wound, the presence or absence of infection, the severity of the wound, the healing period and the healing rate corresponding to each diagnostic image.

The control unit 51 generates a trained model by using each acquired diagnostic image, blood flow evaluation information, body fluid component value, and specific label as training data (S15). The control unit 51 determines the presence / absence of other training data (S16), and if there is other training data (YES in S16), continues the processing after step S11.

When there is no other training data (NO in S16), the control unit 51 stores the generated trained model in the storage unit 54 (S17), and ends the process.

In the above example, the wound medical care support device 50 is configured to generate a trained model, but is not limited to this. For example, a trained model may be generated by another device different from the wound medical care support device 50, and the generated learned model may be stored in the wound medical care support device 50.

FIG. 17 is a flowchart showing an example of a processing procedure of a method of estimating a medical treatment result by the wound medical treatment support device 50. The control unit 51 acquires an image of a wound on the patient's foot (S21) and determines the presence or absence of blood flow evaluation information of the patient (S22). When there is blood flow evaluation information of the patient (YES in S22), the control unit 51 acquires the blood flow evaluation information (S23) and performs the process of step S24 described later.

When there is no blood flow evaluation information of the patient (NO in S22), the control unit 51 determines the presence or absence of the body fluid component value of the patient (S24). When there is a body fluid component value of the patient (YES in S24), the control unit 51 acquires the body fluid component value (S25) and performs the process of step S26 described later.

When there is no body fluid component value of the patient (NO in S24), the control unit 51 inputs the test information (blood flow evaluation information, body fluid component value) including the acquired image into the trained model (learner 55) (S26). ), Output information on the medical treatment result including the cause of the wound on the leg (S27), display the estimation result based on the output information (S28), and end the process. The display of the estimation result can be, for example, as illustrated in FIG.

The wound medical care support device 50 can also be realized by using a general-purpose computer including a CPU (processor), a GPU, a RAM (memory), and the like. That is, as shown in FIGS. 16 and 17, a computer program defining the procedure of each process is loaded into a RAM (memory) provided in the computer, and the computer program is executed by the CPU (processor) on the computer. The wound medical care support device 50 can be realized in the above. The computer program may be recorded and distributed on a recording medium.

According to this embodiment, even a doctor who has little knowledge or experience about a leg wound (for example, a doctor who diagnoses diabetes, a doctor who performs dialysis, etc.) can accurately grasp the cause of the wound and perform correct treatment according to the cause. It can be managed and can support wound care. Also, if the doctor does not have treatment skills, the patient can be referred to the appropriate department.

Further, according to the present embodiment, a system surpassing an experienced doctor can be constructed by including detailed test parameters (blood flow evaluation information, body fluid component value) in addition to the initial examination and palpation results.

The wound medical treatment support method of the present embodiment acquires the test information on the wound on the foot, and when the test information on the wound on the foot is input, it is acquired in the trained model that outputs the information on the medical treatment result including the cause of the wound. Input the test information and output the medical treatment result.

The wound medical treatment support device of the present embodiment is a trained model that outputs information on the medical treatment result including the cause of the wound when the acquisition unit for acquiring the examination information on the wound on the foot and the examination information on the wound on the foot are input. Is provided with an output unit for inputting the acquired examination information and outputting the medical treatment result.

The computer program of the present embodiment acquires test information on foot wounds in a computer, and when the test information on foot wounds is input, it outputs information on medical results including the cause of the wound to a trained model. Input the acquired test information, output the medical treatment result, and execute the process.

The trained model is generated to output information about medical results, including the cause of the wound, when test information about the wound on the foot is input. The examination information can be information that can be obtained by an examination, an interview, or the like performed when a doctor treats a wound. In addition to the cause of the wound, the information on the medical treatment result can include the presence or absence of infection, the severity of the wound, and the like. As the teacher data (training data), the trained model can use a large amount of data (information on examination information and medical treatment results) diagnosed by an experienced specialist doctor who specializes in wounds on the legs.

The examination information regarding the wound on the patient's foot is acquired, the acquired examination information is input to the trained model, and the medical treatment result is output. That is, by inputting the test information into the trained model, the trained model can output information on the medical treatment result similar to that of an experienced doctor who specializes in foot wounds. As a result, even a doctor who has little knowledge and experience about wounds on the legs can accurately know the cause of the wounds and can perform correct diagnosis and treatment. It can support the wound care of poor doctors.

In the wound medical treatment support method of the present embodiment, the inspection information includes an image of a wound on the foot, and when an image of the wound on the foot is input, the trained model outputs information on the medical treatment result. , Input the acquired image and output the medical treatment result.

In the trained model generation method of the present embodiment, a plurality of diagnostic images in which the cause of the wound on the foot is diagnosed are acquired, and a specific label for identifying the cause of the wound and the presence or absence of infection corresponding to each diagnostic image is acquired. Then, using the acquired diagnostic image and the specific label, a trained model that outputs information on the medical result including the cause of the wound is generated.

The trained model can generate a plurality of diagnostic images in which the cause of the wound on the foot is diagnosed, and a specific label for identifying the cause of the wound and the presence or absence of infection corresponding to each diagnostic image as training data. can. The specific label can be a teacher label for each cause of the wound or a teacher label expressing the presence or absence of infection. The trained model can output information on the medical treatment result when an image of a wound on the foot is input. Information on the outcome of the procedure includes the cause of the wound as well as the presence or absence of infection.

It is possible to acquire an image of a wound on the foot, input the acquired image to the trained model, and output the medical treatment result. Since the information on the medical treatment results includes the presence or absence of infection, the presence or absence of infection can also be estimated. As a result, even a doctor who has little knowledge and experience about foot wounds can accurately know the cause of the wound and the presence or absence of infection based on the image of the patient's foot wound.

The wound medical treatment support method of the present embodiment acquires the wound evaluation information representing the evaluation of the wound on the foot, and outputs the information on the medical treatment result when the wound evaluation information representing the evaluation of the wound on the foot is input. The acquired wound evaluation information is input to the model, and the medical treatment result is output.

In the trained model generation method of the present embodiment, wound evaluation information representing the evaluation of a wound on the foot is acquired, and a specific label for identifying the cause of the wound and the presence or absence of infection corresponding to the wound evaluation information is acquired. The acquired wound evaluation information and the specific label are further used to generate the trained model that outputs information on the medical treatment result.

The trained model can be generated by using the wound evaluation information indicating the evaluation of the wound on the foot and the specific label corresponding to the wound evaluation information to identify the cause of the wound and the presence or absence of infection as training data. The wound evaluation information can be information including a plurality of items (parameters) that can evaluate the state such as the position of the wound, the shape of the wound, the skin around the wound, the leaked liquid, and the hair growth of the toes. When the wound evaluation information is used as training data, the evaluation categories for evaluating the state may be defined in advance for each parameter, and the evaluation values for each evaluation category may be collectively vectorized. The specific label can be a teacher label for each cause of the wound or a teacher label expressing the presence or absence of infection. The trained model can output information on the medical treatment result when the wound evaluation information is input. Information on the outcome of the procedure includes the cause of the wound as well as the presence or absence of infection.

Wound evaluation information representing the evaluation of the wound on the foot can be acquired, the acquired wound evaluation information can be input to the trained model, and the medical treatment result can be output. The wound evaluation information may be acquired by performing image processing on an image of a wound on the foot, or may be acquired by observing the patient's foot with the naked eye. As a result, even a doctor who has little knowledge or experience regarding foot wounds can accurately know the cause of the wound and the presence or absence of infection based on the patient's foot wound evaluation information.

In the wound medical treatment support method of the present embodiment, the wound evaluation information is extracted from an image of a foot wound and acquired.

The wound evaluation information can be extracted by performing necessary image processing on the photographed image of the wound on the foot. Some of the leaked fluid can be obtained by macroscopic observation. The position of the wound can be identified from the captured image, and the shape, size, color, and depth (depth of focus) of the wound can be extracted. In addition, the color and luster of the skin around the wound, redness and swelling of the skin, and the dry state of the skin can be extracted from the captured image. In addition, the presence or absence of hair growth on the toes can be extracted from the captured image. This makes it possible to comprehensively determine various conditions such as the position of the wound, the shape of the wound, the skin around the wound, the leaked liquid, and the hair growth of the toes, and estimate the cause of the wound and the presence or absence of infection. ..

In the wound medical treatment support method of the present embodiment, when the test information includes blood flow evaluation information representing the evaluation of the blood flow of the foot and further input the blood flow evaluation information representing the evaluation of the blood flow of the foot, the medical treatment is performed. Output information on the result The acquired blood flow evaluation information is input to the trained model, and the medical treatment result is output.

The learned model generation method of the present embodiment acquires blood flow evaluation information representing the evaluation of the blood flow of the foot for which the cause of the wound on the foot has been diagnosed, and determines the cause of the wound corresponding to the blood flow evaluation information. A specific label to be specified is acquired, and the acquired blood flow evaluation information and the specific label are further used to generate the trained model that outputs information on the medical treatment result.

The trained model uses blood flow evaluation information representing the evaluation of blood flow in the foot for which the cause of the wound on the foot has been diagnosed, and a specific label for identifying the cause of the wound corresponding to the blood flow evaluation information as training data. Can be generated. The blood flow evaluation information can be, for example, pulse pressure of the foot, skin perfusion pressure, blood flow and blood flow velocity of the foot, skin surface temperature of the foot, tissue oxygen saturation in the vicinity of the wound, and the like. The specific label can be a teacher label expressing the presence or absence of arterial ulcer, which is one of the causes of wounds. When the blood flow evaluation information is input, the trained model can output information on the medical treatment result (specifically, the presence or absence of arterial ulcer).

It is possible to acquire blood flow evaluation information representing the evaluation of blood flow in the foot, further input the acquired blood flow evaluation information into the trained model, and output the medical treatment result. In particular, by further considering the blood flow evaluation information, it is possible to make a definitive diagnosis as to whether or not the cause of the wound is an arterial ulcer.

In the wound medical treatment support method of the present embodiment, the test information includes the body fluid component value, and when the body fluid component value is further input, the acquired body fluid component value is output to the learned model that outputs information on the medical treatment result. Is input and the medical treatment result is output.

The trained model is generated to output information about the medical treatment result when the body fluid component value is further input. Body fluid component values include, for example, CRP (C-reactive protein) levels, serum albumin levels, minerals, vitamins and the like. The trained model can be generated by using the body fluid component value and the specific label corresponding to the body fluid component value as training data. The specific label can be, for example, a teacher label expressing the presence or absence of infection, a teacher label expressing the healing period of the wound, and a teacher label expressing the healing rate of the wound.

It is possible to acquire the body fluid component value, further input the acquired body fluid component value into the trained model, and output the medical treatment result. This makes it possible for even a doctor with little knowledge or experience regarding foot wounds to estimate the presence or absence of infection based on the patient's body fluid component value, the wound healing period, the wound healing rate, and the like.

In the wound medical treatment support method of the present embodiment, the cause of the wound includes an arterial ulcer, a venous ulcer, a diabetic ulcer, and a compressive ulcer, and the information on the medical treatment result is the estimated accuracy and infection of each cause. Includes estimated accuracy of.

Wound causes include arterial ulcers, venous ulcers, diabetic ulcers, and compressive ulcers, and information about clinical outcomes includes an estimated probability of each cause and an estimated probability of infection. This allows even a doctor with little knowledge or experience of foot wounds to know the most probable cause of the wound and how likely it is to be infected. ..

In the wound medical care support method of the present embodiment, the information regarding the medical care result includes the severity of the wound.

Information about the outcome includes the severity of the wound. Severity can be classified as, for example, I (mild), II (moderate), III (severe). This allows even a doctor with little knowledge or experience of foot wounds to know the severity of the wound.

The wound treatment support method of the present embodiment recommends a wound treatment method and a clinical department based on the presumed cause of the wound.

Wound treatment methods and departments can be recommended based on the presumed cause of the wound. For example, if the wound is caused by an arterial ulcer and the ischemic area is narrow, revascularization can be recommended if there is no infection. In addition, if the ischemic area is narrow and there is infection, debridement (removal of necrotic tissue) can be performed, and then revascularization can be recommended. After revascularization, wound treatment (prevention of infection with topical agents, control of exudate with wound dressing and promotion of granulation) can be recommended. In addition, it is possible to recommend a clinical department that specializes in each treatment.

The wound medical treatment support method of the present embodiment acquires the test information including the image of the wound on the foot, and when the test information on the wound on the foot is input, it is acquired in the trained model that outputs the information on the medical treatment result. Input the test information and output the medical treatment result, and as the medical treatment result, the cause of the wound including arterial ulcer, venous ulcer, diabetic ulcer and compression ulcer, the estimated probability of each cause, and the estimated probability of infection. , As well as the recommended treatment methods and recommended departments for wounds.

The wound medical treatment support device of the present embodiment has an acquisition unit that acquires examination information including an image of a wound on the foot, and a trained model that outputs information on the medical treatment result when the examination information on the wound on the foot is input. By inputting the test information acquired by the acquisition unit, the cause of the wound including arterial ulcer, venous ulcer, diabetic ulcer and compression ulcer and the estimated accuracy of each cause can be used as information on the clinical result. It is provided with an estimated probability of infection and a display unit for displaying recommended treatment methods and recommended clinical departments for wounds.

The test information including the image of the wound on the foot is acquired, the acquired test information is input to the trained model, and the medical treatment result is output. As the medical treatment result, the cause of the wound including arterial ulcer, venous ulcer, diabetic ulcer and compression ulcer, the estimated probability of each cause, the estimated probability of infection, and the recommended treatment method and recommended clinical department of the wound are displayed.

As a result, even a doctor who has little knowledge and experience about wounds on the legs can accurately know the cause of the wounds and can perform correct diagnosis and treatment. It can support the wound care of poor doctors. In addition, displaying the recommended treatment method and recommended clinical department of the wound facilitates explanation to the patient.

The wound medical care support method of the present embodiment further displays a captured image and the severity of the wound.

Furthermore, by displaying the captured image and the severity of the wound, the condition of the wound on the patient's foot and the risk of aggravation can be easily confirmed.

The wound care support method of this embodiment further displays an estimated wound healing period and an estimated healing rate.

Furthermore, by displaying the estimated wound healing period and the estimated healing rate, it is possible for the patient to understand how long it takes for the wound to heal and at what rate the wound is completely healed. , Can give the patient a sense of security.

In the wound medical treatment support method of the present embodiment, images of wounds on the foot taken at different time points are acquired, and each of the acquired images is input to the trained model to change the severity of the wound. indicate.

Images of the wound on the foot taken at different time points are acquired, and each of the acquired images is input to the trained model to display the transition of the severity of the wound. For example, by showing the patient that the severity at this visit is less severe than at the previous visit, the patient should ensure that the wound is being treated successfully. Can be done.

The trained model generation method of the present embodiment acquires a specific label that specifies the wound healing period corresponding to at least one of the diagnostic image and the wound evaluation information representing the evaluation of the wound on the foot, and further uses the specific label. To generate the trained model that outputs the healing period of the wound.

The trained model can be generated using specific labels as training data that identify the wound healing period corresponding to at least one of the diagnostic image and the wound assessment information representing the assessment of the wound on the foot. The wound healing period can be 1 month, 2 months, 3 months, or the like, and the label corresponding to each wound healing period can be used as the teacher label. The trained model can output the healing period of the wound if at least one of the diagnostic image and the wound evaluation information representing the evaluation of the wound on the foot is input.

10 Cameras 20, 21 Measuring instruments 50 Wound medical care support device 51 Control unit 52 Interface unit 53 Image processing unit 54 Storage unit 55 Learning device 551 CNN
552, 553 DNN
55a Output layer 56 Display panel 57 Operation unit

Claims (20)

Get inspection information about wounds on the foot,
When the test information about the wound on the leg is input, the acquired test information is input to the trained model that outputs the information about the medical result including the cause of the wound, and the medical result is output.
Wound medical care support method. The inspection information includes an image of a wound on the foot.
When an image of a wound on the foot is input, the acquired image is input to the trained model that outputs information on the medical treatment result, and the medical treatment result is output.
The wound medical treatment support method according to claim 1. Obtain wound evaluation information that represents the evaluation of a wound on the foot,
When the wound evaluation information representing the evaluation of the wound on the foot is input, the acquired wound evaluation information is input to the trained model that outputs the information on the medical treatment result, and the medical treatment result is output.
The wound medical treatment support method according to claim 2. The wound evaluation information is extracted from an image of a foot wound and acquired.
The wound medical treatment support method according to claim 3. The test information includes blood flow evaluation information representing an evaluation of blood flow in the foot, and includes blood flow evaluation information.
When the blood flow evaluation information representing the evaluation of the blood flow of the foot is further input, the acquired blood flow evaluation information is input to the trained model that outputs the information on the medical treatment result, and the medical treatment result is output.
The wound medical treatment support method according to any one of claims 1 to 4. The test information includes body fluid component values and includes
When the body fluid component value is further input, the acquired body fluid component value is input to the learned model that outputs information on the medical treatment result, and the medical treatment result is output.
The wound medical treatment support method according to any one of claims 1 to 5. Causes of the wound include arterial ulcers, venous ulcers, diabetic ulcers, and compressive ulcers.
The information on the medical results includes the estimated accuracy of each cause and the estimated accuracy of infection.
The wound medical treatment support method according to any one of claims 1 to 6. The information regarding the medical results includes the severity of the wound.
The wound medical treatment support method according to any one of claims 1 to 7. Wound treatment methods and departments are recommended based on the presumed cause of the wound,
The wound medical treatment support method according to any one of claims 1 to 8. Obtain inspection information, including images of foot wounds,
When the test information about the wound on the foot is input, the acquired test information is input to the trained model that outputs the information about the medical treatment result, and the medical treatment result is output.
As the medical treatment result, the cause of the wound including arterial ulcer, venous ulcer, diabetic ulcer and compression ulcer, the estimated probability of each cause, the estimated probability of infection, and the recommended treatment method and recommended clinical department of the wound are displayed. ,
Wound medical care support method. In addition, the captured image and the severity of the wound are displayed.
The wound medical treatment support method according to claim 10. In addition, the estimated wound healing period and estimated healing rate are displayed.
The wound medical treatment support method according to claim 10 or 11. Get images of foot wounds taken at different times,
By inputting each acquired image into the trained model, the transition of the severity of the wound is displayed.
The wound medical treatment support method according to any one of claims 10 to 12. Obtain multiple diagnostic images where the cause of the wound on the foot was diagnosed,
Obtain a specific label that identifies the cause of the wound and the presence or absence of infection corresponding to each diagnostic image.
Using the acquired diagnostic images and specific labels, generate a trained model that outputs information about medical results, including the cause of the wound.
Trained model generation method. Obtain wound evaluation information that represents the evaluation of a wound on the foot,
Obtain a specific label that identifies the cause of the wound and the presence or absence of infection corresponding to the wound evaluation information.
Further using the acquired wound evaluation information and a specific label, the trained model that outputs information on the medical treatment result is generated.
The trained model generation method according to claim 14. Obtain blood flow evaluation information representing the evaluation of blood flow in the foot for which the cause of the wound on the foot has been diagnosed.
Obtain a specific label that identifies the cause of the wound corresponding to the blood flow evaluation information,
Further using the acquired blood flow evaluation information and the specific label, the trained model that outputs information on the medical treatment result is generated.
The trained model generation method according to claim 14 or 15. Obtain a specific label specifying the wound healing period corresponding to at least one of the diagnostic image and the wound evaluation information representing the evaluation of the wound on the foot.
The specific label is further used to generate the trained model that outputs the healing period of the wound.
The trained model generation method according to any one of claims 14 to 16. The acquisition department that acquires inspection information on foot wounds,
Wound medical care support provided with an output unit that inputs the acquired test information to the trained model that outputs information on the medical results including the cause of the wound when the test information on the wound on the foot is input and outputs the medical results. Device. An acquisition unit that acquires inspection information including images of foot wounds,
When the test information related to the wound on the leg is input, the test information acquired by the acquisition unit is input to the trained model that outputs the information related to the medical treatment result, so that the information related to the medical treatment result is arterial ulcer or venous. A wound medical care support device including a cause of a wound including an ulcer, a diabetic ulcer and a compressive ulcer, an estimated probability of each cause, an estimated probability of infection, and a display unit for displaying a recommended treatment method and a recommended medical department of the wound. On the computer
Get inspection information about wounds on the foot,
When the test information about the wound on the leg is input, the acquired test information is input to the trained model that outputs the information about the medical result including the cause of the wound, and the medical result is output.
A computer program that executes processing.

Images