JP3871144B1 - Diagnosis support device - Google Patents

Abstract

The present invention provides information about a future disease predicted for a patient.
In a diagnosis support apparatus 10, first, a candidate detection unit 16 performs a process of detecting a lacunar infarction candidate from a captured image of a patient (step S1). When the patient's biometric attribute information is input by the operator via the operation unit 12 (step S2), the control unit 11 controls the detection result of the lacunar infarction candidate and the biometric attribute information based on the diagnosis result of the doctor. Is stored in either the case patient DB 151 or the target patient DB 153 (step S3). Further, the diagnosed disease name, the family history, and case information for the period until the onset of disease are stored in the case information DB 152. Next, based on the case patient DB 151 and the case information DB 152, the control unit 11 calculates a predicted value for each target patient registered in the target patient DB 153, and stores the predicted value in association with each target patient in the target patient DB 153. (Step S4).
[Selection] Figure 6

Description

  The present invention relates to a diagnosis support apparatus that processes information about a patient, outputs diagnosis support information, and performs diagnosis support of a doctor.

In the medical field, a diagnosis support apparatus has been developed to support a doctor's diagnosis work from various aspects. For example, there is a method in which examination data and attribute data for a plurality of patients are made into a database, and a disease name that is highly likely to be affected is determined and output based on this database (for example, see Patent Document 1). . Along with the advancement of image processing technology, a diagnosis support apparatus has been developed that detects an image region that is expected to be a lesion from a medical image obtained by imaging a patient as an abnormal shadow candidate (see, for example, Patent Document 2).
JP 2003-310557 A JP 2002-112985 A

  However, the above-described diagnosis support apparatus merely analyzes a disease that has already occurred, and has only a function of assisting a doctor's diagnosis of the disease.

  The subject of this invention is providing the diagnosis assistance apparatus which can provide the information about the disease estimated in the future about a patient.

The invention according to claim 1 is a diagnosis support apparatus,
An abnormal shadow candidate detecting means for detecting an image region of an abnormal shadow candidate from a medical image obtained by photographing a patient;
Patient information storage means for storing, for each patient, biometric attribute information obtained for a plurality of affected patients and the detection result of the abnormal shadow candidate;
As a target sample with biometric attribute information and abnormal shadow candidate detection results for a plurality of patients as a mother sample, and biometric attribute information and abnormal shadow candidate detection results for a patient to be diagnosed as a variable by the nearest neighbor method A predictive value calculating means for calculating a predictive value indicating a possibility of future disease of the patient to be diagnosed based on a mother sample adjacent to the target sample ;
Display means for displaying, along with the calculated predicted value, the detection result of the patient's biological attribute information and abnormal shadow candidate that became the mother sample adjacent to the target sample;
It is characterized by providing.

The invention according to claim 2 is the diagnosis support apparatus according to claim 1 ,
The biometric attribute information includes at least information on a patient's age, sex, smoking, family history of illness, height, weight, body fat percentage, and blood pressure.

The invention according to claim 3 is the diagnosis support apparatus according to claim 1 ,
The biological attribute information includes information on a period from when an abnormal shadow candidate is detected by the abnormal shadow candidate detection means to when the disease occurs.

The invention according to claim 4 is the diagnosis support apparatus according to claim 3 ,
The predicted value calculating means calculates a period during which disease is expected to occur using the detection result of the abnormal shadow candidate and the biological attribute information, and converts this into a predicted value.

The invention according to claim 5 is the diagnosis support apparatus according to any one of claims 1 to 4,
The predicted value calculation means groups the plurality of patients based on the biological attribute information, and calculates a predicted value for each group.

The invention according to claim 6 is the diagnosis support apparatus according to any one of claims 1 to 5 ,
Setting means for setting temporary information to replace the biometric attribute information stored in the storage means;
The predicted value calculation means calculates a predicted value using the detected abnormal shadow candidate and the set temporary information.

The invention according to claim 7 is the diagnosis support apparatus according to any one of claims 1 to 6 ,
The lesion to be detected by the abnormal shadow candidate detection means is lacunar infarction,
The predicted value calculated by the predicted value calculating means is an index value indicating the possibility of developing any of the following diseases, from dementia, cerebral infarction, or subarachnoid hemorrhage, from the lacunar infarction.

The invention according to claim 8 is the diagnosis support apparatus according to any one of claims 1 to 7 ,
Output means for outputting the predicted value calculated by the predicted value calculation means for the patient to be diagnosed.

The invention according to claim 9 is the diagnosis support apparatus according to claim 8 ,
The output means outputs information related to detection performance of the abnormal shadow candidate detection means together with the predicted value.

The invention according to claim 10 is the diagnosis support apparatus according to claim 8 or 9 , wherein
The temporary information is biometric attribute information assuming a case where the living situation of a patient to be diagnosed is improved,
Improvement instruction information storage means for storing improvement instruction information indicating the content of improvement of the life of the patient corresponding to the improvement condition of the biometric attribute information in the case of improvement,
The output means acquires biometric attribute information of a patient to be diagnosed and improvement instruction information corresponding to the input temporary information from the improvement instruction information storage means, and outputs the improvement instruction information together with the predicted value. And

The invention according to claim 11 is the diagnosis support apparatus according to any one of claims 1 to 10 ,
The display means displays the biometric attribute information for each patient, the predicted value calculated by the predicted value calculating means, and the detection result by the abnormal shadow candidate detecting means so as to be comparable.
The invention according to claim 12 is the diagnosis support apparatus according to any one of claims 1 to 11,
The display means displays a medical image of the target patient for which an abnormal shadow candidate has been detected, together with a predicted value calculated for the target patient.

According to the inventions described in claims 1 , 2 , 7 , 8 , and 12 , the predicted value can be calculated using the detection result of the abnormal shadow candidate that is an important finding of the disease, and the reliability of the predicted value is improved. In addition to improving, it is possible to provide doctors and patients with useful information about the possibility of getting sick in the future. Further, by further using the biometric attribute information for calculation of the predicted value, it is possible to calculate the predicted value in consideration of detailed information. Moreover, since the biometric attribute information of the patient who became a mother sample is displayed together with the predicted value, it can be used as reference information for diagnosis.

According to the third and fourth aspects of the invention, it is possible to predict a period until a doctor becomes sick from a predicted value.

According to the invention described in claim 5 , by grouping for each patient whose characteristics are close to each other, it is possible to suppress variations in information for calculating the predicted value, thereby improving the reliability of the predicted value. it can.

According to the sixth aspect of the present invention, the condition for calculating the predicted value can be changed based on the provisional information. For example, the predicted value assuming the case where the patient's situation has changed, such as when life is improved, is provided to the doctor. be able to.

According to the invention described in claim 9 , the doctor can confirm the reliability of the detection result of the abnormal shadow candidate used for calculating the predicted value.

According to the tenth aspect of the present invention, the cause of illness and the direction of improvement can be shown to the patient by the improvement instruction information.

According to the invention of claim 1 1, the physician may biometric attribute information for each patient, the predicted value, it is possible to compare the detection result of the abnormal shadow candidate, improve the efficiency of a physician.

First, the configuration will be described.
In FIG. 1, the functional structure of the diagnosis assistance apparatus 10 in this embodiment is shown.
The diagnosis support apparatus 10 analyzes an image of a patient's captured image to detect abnormal shadow candidates, creates a database of the results, and calculates a predicted value indicating the possibility of future disease for an unaffected patient. .

  In the present embodiment, an example will be described in which lacunar infarctions that cause various diseases such as stroke, dementia, and subarachnoid hemorrhage are detected as abnormal shadow candidates. Lacunar infarction is a blood vessel that narrows the blood flow in the blood vessels in the brain, necrotic cells become necrotic and become vascular aneurysms. When the number of lacunar infarcts increases or the aneurysm ruptures and the blood flow in the blood vessels flows, the disease such as the above-mentioned stroke occurs, and the lacunar infarction itself is a finding of these diseases.

  As shown in FIG. 1, the diagnosis support apparatus 10 includes a control unit 11, an operation unit 12, a display unit 13, a communication unit 14, a storage unit 15, a candidate detection unit 16, an image memory 17, and an I / F 18. ing. A printer 20 is connected to the diagnosis support apparatus 10 via the I / F 18, and an image generation apparatus that generates a captured image of the patient via the communication unit 14, the captured image and other information related to the patient are integrated. It can be connected to a server to be managed.

Hereinafter, each component will be described.
The control unit 11 includes a CPU (Central Processing Unit), a RAM (Random Access Memory), and the like. The control unit 11 reads out a processing program stored in the storage unit 15 by the CPU, expands it in the RAM, and performs various operations according to the processing program. In addition, the operation of each unit 12 to 18 is comprehensively controlled.

  The operation unit 13 includes a keyboard, a mouse, and the like. When these operations are performed, an operation signal corresponding to the operation is generated and output to the control unit 11.

  The display unit 13 includes a display such as an LCD (Liquid Crystal Display), and displays various operation screens and display information such as processing results of the control unit 11 on the display in accordance with instructions from the control unit 11.

  The communication unit 14 includes a communication interface such as a network interface card, and communicates with an external device. For example, information related to a lesion candidate detected by the candidate detection unit 16, information on a predicted value calculated by the control unit 11, and the like in a server that receives a captured image of a patient from the image forming apparatus or manages and manages information related to the patient. Send.

The storage unit 15 stores a processing program executed by the control unit 11, various parameters necessary for executing the processing program, data, and the like.
Further, the storage unit 15 stores a case patient DB (Data Base) 151, a case information DB 152, a target patient DB 153, and an improvement information DB 154.

  The case patient DB 151 is a DB for managing attribute information and biological information about a diseased patient (hereinafter referred to as a case patient). The attribute information refers to information related to the patient such as the patient's name, the assigned patient ID, age, and sex. Biometric information refers to information related to the patient's living body, for example, information related to the patient's body weight, height, etc., information obtained from examinations such as blood pressure, body fat percentage, and triglyceride, smoking, drinking, etc. Includes information related to daily life. Hereinafter, the attribute information and the biometric information may be collectively referred to as biometric attribute information. The biometric attribute information is obtained by the operator through the operation unit 12 and input to the diagnosis support apparatus 10 from information obtained from examination results and patient answers to the questionnaire.

  As shown in FIG. 2, the case patient DB 151 stores attribute information such as the name of the case patient and biological information such as blood pressure in association with the patient ID. In addition, the number of lacunar infarct candidates, which is a detection result by the candidate detection unit 16, is stored. Note that the attribute information and the biometric information shown in FIG. 2 show an example, and the present invention is not limited to this. For example, in the information item “smoking”, not only the presence / absence of smoking but also more detailed information such as a daily smoking amount (the number of cigarettes) and a smoking history may be stored.

  In the case information DB 152, as shown in FIG. 3, for the case patient stored in the case patient DB 151, the disease name diagnosed by the doctor, the family history of the disease, and information on the period until the onset correspond to the patient ID. It is remembered.

The target patient DB 153 is a DB for managing attribute information and biological information about a patient who has not yet developed a disease and calculates a predicted value indicating the possibility of future disease (hereinafter referred to as a target patient). is there.
As shown in FIG. 4, the target patient DB 153 stores attribute information such as a patient name and biological information such as blood pressure in association with the patient ID of the target patient. In the family history item, when there is a family history, the disease name is stored. Moreover, the information of the prediction value calculated by the control part 11 in the database process mentioned later and the number of lacunar infarction candidates detected by the candidate detection part 16 (or certified by a doctor's diagnosis after that) are memorize | stored.

  The improvement instruction information DB 154 is a DB that manages improvement instruction information such as instruction messages for instructing patients to improve their lives based on predicted values. For example, as shown in FIG. 5, the improvement instruction information DB 154 stores an improvement condition and two improvement instruction information 1 and 2 corresponding to the improvement condition for each item that needs improvement. For example, in the “weight” item, the improvement condition in which the weight greater than the standard weight is reduced to the standard weight includes the improvement instruction information 1 of “Let's reduce the weight to the standard weight” and “representative for reducing the weight” Corresponds to the improvement instruction information 2 "

  The candidate detection unit 16 is configured by a CPU or the like, and detects a candidate region of a lacunar infarction shadow by performing image analysis of a captured image of a patient in cooperation with a candidate detection processing program stored in the storage unit 15.

Hereinafter, a method for detecting a lacunar infarction candidate will be described. Here, detection is performed using a T1-weighted image and a T2-weighted image taken by MRI (Magnetic Resonance Imaging).
First, the T1-weighted image and the T2-weighted image of the detection target patient are read from the image memory 17. In the T1-weighted image, the lacunar infarction exhibits a lower signal than the brain tissue, whereas in the T2-weighted image, the higher signal than the brain tissue. Since the lacunar infarct exists on the brain parenchyma around the ventricle, it appears as a circular shadow on the image with a contrast different from that of the surrounding area.

  Therefore, the candidate detection unit 16 repeats binarization while changing the threshold using the T2-weighted image, and extracts a circular region having a diameter of 3 to 10 mm that exhibits a high signal in each binarized image. Then, a feature amount analysis such as a circularity and an area is calculated for the extracted region and a feature amount analysis is performed to perform primary detection of a lacunar infarction candidate.

  Next, a histogram is created in the T1-weighted image to obtain the most frequent pixel value, and a brain parenchymal region is extracted by the region expansion method using this most frequent pixel value as an extension point. Next, among the primary candidates for lacunar infarction, the primary candidates that exist outside the extracted brain parenchyma region are determined to be false positive candidates. This is because the lacunar infarction exists on the brain parenchyma region, and therefore it is possible to determine that the primary candidates detected in regions other than the extracted brain parenchyma region, such as the cerebral sulcus and brain margin, are false positives. And the remaining candidate which deleted the candidate judged to be a false positive candidate from a primary candidate is output as a final detection result.

  Note that the above detection method is an example, and other detection methods may be applied, and the detection method to be applied is not particularly limited.

  The image memory 17 stores a captured image of the patient input via the communication unit 14.

The I / F 18 is an interface for connecting the printer 20.
The printer 20 performs print processing based on output data input from the diagnosis support apparatus 10. For example, the improvement instruction sheet output data is received from the diagnosis support apparatus 10 and the improvement instruction sheet is printed out.

Next, the operation of the diagnosis support apparatus 10 will be described.
The diagnosis support apparatus 10 creates a database of biometric attribute information input by an operator for a patient who has completed an examination, in cooperation with the control unit 11 and each processing program stored in the storage unit 15, and based on the database. Software processing of database processing for calculating the predicted value of the target patient, predicted value display processing for displaying the calculated predicted value, and sheet output processing for outputting an improvement instruction sheet describing improvement items corresponding to the predicted value is executed. .

First, database processing will be described with reference to FIG.
In the database process shown in FIG. 6, first, the candidate detection unit 16 performs a process for detecting a lacunar infarction candidate from a captured image of the patient (step S1). The detection result is displayed on the display unit 13 and, for example, a marker such as an arrow indicating the position of the detected lacunar infarction candidate is displayed on the captured image. The doctor interprets the photographed image while referring to the detection result, makes a diagnosis of lacunar infarction, whether or not the patient is ill, and inputs the diagnosis result to the diagnosis support apparatus 10 via the operation unit 12. . In addition, when it is diagnosed that the disease has occurred, information on the name of the diagnosed disease, the family history of the disease, and the period until the onset of disease is input.

On the other hand, the operator inputs biometric attribute information of the patient who has completed the examination to the diagnosis support apparatus 10.
In the diagnostic support apparatus 10, when the patient's biometric attribute information is input by the operator via the operation unit 12 (step S2), the control unit 11 controls the detection result of the lacunar infarction candidate based on the diagnosis result of the doctor. The biometric attribute information is stored in either the case patient DB 151 or the target patient DB 153 (step S3).

  That is, when a diagnosis result indicating that the patient is ill has been input, the detection result of the lacunar infarction candidate and the biometric attribute information are stored in the case patient DB 151, and the diagnosis result that the patient is not ill has been input. If it exists, it is stored in the target patient DB 153. When it is diagnosed that the patient is ill, the name of the diagnosed disease, the family history, and case information for the period until the illness are stored in the case information DB 152. In addition, when a reexamination is performed on a target patient who has not yet developed disease, it may be diagnosed that the disease has occurred. Therefore, when the case patient DB 151 and the target patient DB 153 are collated and the same patient is searched, The biometric attribute information stored in the target patient DB 153 is deleted.

  When each of the DBs 151 to 153 is updated as described above, a predicted value is calculated for each target patient registered in the target patient DB 153 by the control unit 11 based on the case patient DB 151 and the case information DB 152. The calculated predicted value is stored in the target patient DB 153 in association with each target patient (step S4).

Here, a method using a multiple regression analysis method and a method using a nearest neighbor method will be described as methods for calculating a predicted value.
<When multiple regression analysis method is used>
When calculating the predicted value by multiple regression analysis, the prediction period Y until the onset of disease is calculated by the following multiple regression analysis formula 1.
Y = a ₁ x ₁ + a ₂ x ₂ +... + _An x _n + e (1)
Here, x _n is the n independent variables, a _n is a weighting factor.
E is a constant.

Coefficient a _n is predetermined by the biological attribute information and lacunar infarction number of candidates in each case patients stored in the case patients DB 151. That is, since the period until the onset of disease is known for the case patient, data for that period is input to Y, and n item data of the biometric attribute information and the number of lacunar infarction candidates for the case patient are input to x _n respectively. the formula 1 prepared for a plurality of cases patients, optimal a _n satisfying these multiple expressions 1 is determined.

In Formula 1 is a _n obtained this way, by inputting each item data of the biometric attribute information stored and lacunar infarction number of candidates to the subject patient DB153 for target patients for calculating the predicted value x _n The prediction period Y is calculated. The calculated prediction period Y is converted into a predicted value Z normalized to 1 to 100. In the present embodiment, the higher the predicted value, the higher the possibility of disease occurrence, and the conversion to the normalized value is simply linear from 0 to 5 years of the prediction period Y to the normalized value of 100 to 0. Conversion is performed (the prediction value 100 is assigned to the prediction period 0, and the prediction value 0 is assigned to the prediction period 5).

<When using the nearest neighbor method>
In the nearest neighbor method, the biometric attribute information and the lacunar infarction candidate of the target patient are stored in the feature quantity space of the mother sample with n items of the biometric attribute information and the number of lacunar infarct candidates stored in the case patient DB 151 as variables. When a target sample having a variable number of n item data is set, a mother sample closest to the target sample is obtained. The period until the onset of the case patient of this mother sample is approximated as the predicted period Y until the onset of the target patient. The prediction period Y obtained in this way is normalized as in the case of the multiple regression analysis, and is calculated as a predicted value Z of 0 to 100.

  As for the distance between the target sample and the mother sample, the Euclidean distance is obtained in consideration of the variance of the mother sample. Note that even if the Euclidean distance is the closest to the target sample, depending on the size of the Euclidean distance, the error may increase if the prediction period Y is approximated to the period until the occurrence of this mother sample. The value of the predicted value Z is corrected according to the magnitude of. The correction is performed using a correction table prepared in advance. In the correction table, for example, correction values for the Euclidean distance are determined in advance so that the predicted value Z is subtracted by 1, 3, and 10 as the Euclidean distance becomes 1, 2, and 3, for example. This correction value may be set optimally experimentally and empirically. To explain a correction example, the predicted value Z = 80 is calculated. However, when the Euclidean distance from the nearest sample is 2, the predicted value Z = 80-3 = 77 is calculated based on the correction table. The

  As described above, the period until the onset of the nearest mother sample may be regarded as the prediction period Y. However, a plurality of mother samples are selected in order from the smallest Euclidean distance to the target sample, and It is good also as averaging the period until onset and making the average period into the prediction period Y.

  In addition, in any method, for example, when “blood pressure” was not measured in this examination, when data was not obtained for all information items to be input for the target patient, the information for which the data was not obtained Standard alternative data may be used for the item, or no data may be input for the information item.

  When the predicted value Z for each target patient is calculated as described above, this predicted value Z is stored in the target patient DB 153 in association with the patient ID of the target patient (step S4). Since the predicted value Z varies depending on the case patient DB 151, the predicted value Z may be calculated according to the update of the case patient DB 151, or may be recalculated at regular intervals.

Next, the predicted value display process will be described with reference to FIG.
In the predicted value display process shown in FIG. 7, the patient list screen d1 shown in FIG. 8 is displayed on the display unit 13 under the control of the control unit 11 (step S11).

  On the patient list screen d1, a list of target patients stored in the target patient DB 153 is displayed, and a list d11 that can compare information such as patient ID, biometric attribute information, and the number of lacunar infarctions of each target patient is displayed. . In the initial setting, target patients are listed in the order of patient IDs. For the number of lacunar infarct candidates, the detection performance d12 of the candidate detection unit 16 is displayed at the bottom of the list. As a result of detection by the candidate detection unit 16 using an image whose true positive or false positive is known in advance, the probability of detecting true positive is the true positive rate, and the number of detected false positives is divided by the total number of cases. This value is used as the number of false positives per case, and is displayed as an index value for the doctor to confirm the reliability of the detection result of the candidate detection unit 16.

  In addition, on the patient list screen d1, predicted values calculated for each target patient are displayed in a list. The predicted values can be sorted (rearrangement processing), and the instruction key k1 is displayed on the upper right of the screen. It is also possible to select a target patient and display more detailed information such as a captured image, and the instruction key k2 is displayed.

  Here, when the predicted value sort instruction key k1 is operated via the operation unit 12 (step S12; Y), the list d11 in which the target patients are rearranged in descending order of the predicted value by the control unit 11 is the patient list. It is displayed on the screen d1 (step S13).

  Further, when one target patient is selected from the patient list d11 via the operation unit 12 and then the image display instruction key k2 is operated, the detail screen d2 shown in FIG. 9 is displayed, and the selected patient is imaged. An image, a predicted value, and other detailed information are displayed (step S14).

  On the detailed screen d2, the captured image d21 of the selected target patient and the calculated predicted value d22 are displayed, and the biological attribute information of the target patient and the number of detected lacunar infarct candidates are displayed as the detailed information d23. In addition, when the predicted value is calculated by the nearest neighbor method, the biometric attribute information of the case patient who is the mother sample related to the calculation of the predicted value, the number of detected lacunar infarct candidates, and case information from the case patient DB 151 and the case information DB 152 It is good also as reading and displaying the list | wrist d24. Thereby, the doctor can use biometric attribute information and the like about the case patient whose tendency is similar to the target patient as reference information for diagnosis.

  Further, on the detailed screen d2, an instruction key k3 for outputting an improvement instruction sheet is displayed. By operating this instruction key k3, an improvement instruction sheet for instructing improvement on the life of the target patient is displayed on the printer 20. Can be output.

Hereinafter, the sheet output process will be described with reference to FIG. The sheet output process is a process that is started when the instruction key k3 is operated on the detailed screen d2.
In the sheet output process shown in FIG. 10, provisional information is set for an information item (hereinafter referred to as an improvement item) that becomes a prediction value improvement parameter by the control unit 11, and a prediction value is calculated again using the provisional information ( Step S101). Hereinafter, the former is referred to as a current predicted value and the latter is referred to as an improved predicted value in order to distinguish between the predicted value calculated previously and the predicted value calculated using temporary information.

  The improvement item is determined in advance from among information items used for calculating the improvement prediction value as an item that can be improved by a change in the life of the patient. For example, blood pressure, weight, etc. can be changed by the patient's improvement efforts, so the information items of the biometric attribute information are determined as improvement items, but the patient's improvement such as height, age, gender, number of lacunar infarction shadow candidates, etc. Information items for which efforts are not made are not determined as improvement items.

  As the temporary information, a standard index value of a healthy body is set. For example, the provisional information on the improvement item “weight” uses the weight defined as the standard based on the age, sex, height, and the like of the patient. For the standard weight, the formula defined by the Japanese Society of Obesity can be used: standard weight = height (m) × height (m) × 22.

  The improved predicted value is calculated by the same method as the current predicted value calculation method. That is, when the multiple regression analysis method is used, only the improvement items are input instead of the temporary information using the multiple regression analysis formula to which the same explanatory variable and independent variable are applied. In addition, an improvement prediction value is calculated for each improvement item for which provisional information is input so that it can be specified how much the improvement item affects the prediction value. That is, the patient is expected to live for each improvement item, such as an improvement prediction value when provisional information is used for the improvement item of “blood pressure” and an improvement prediction value when provisional information is used for the improvement item of “weight”. An improvement prediction value in the case of improvement is calculated.

  Next, a rate of change with respect to the current predicted value is calculated for the improved predicted value calculated for each improved item by the control unit 11, and the presence / absence of an improved item having a large changed rate is determined (step S102). The rate of change is the ratio of improved predicted value / current predicted value. An improvement item having a small change rate, that is, an improvement item having a small influence on the predicted value is a standard (healthy body) for the item, or an item that has little influence on the disease, and therefore needs to be improved. Therefore, an improvement item having a small change rate is determined based on the threshold value, and an improvement instruction is given only for an improvement item having a large change rate. What is necessary is just to set suitably the threshold value for judging the magnitude of a change rate suitably from experience.

As a result of the determination, if there is no improvement item having a large change rate (step S102; N), the improvement instruction sheet is not output and the present process is terminated.
On the other hand, when at least one improvement item with a large change rate is included (step S102; Y), the improvement corresponding to the improvement item and the improvement condition determined by the control unit 11 that the change rate is large. The instruction information is read from the improvement information DB 154 (step S103).

  Referring to the example shown in FIG. 5, when the improvement prediction value is calculated for the improvement item “weight” having a large change rate instead of the provisional information of the standard weight from the weight higher than the standard weight, the improvement instruction information 1 “weight” is calculated. Are reduced to the standard weight ”, and the improvement instruction information 2“ Representative method for losing weight ... ”is read out.

  Next, when the patient biometric attribute information to be displayed on the improvement instruction sheet is read from the DB 153 by the control unit 11, the data of the improvement instruction sheet format is read from the storage unit 15, and according to this format, the patient biometric attribute information, The output data of the improvement instruction sheet in which the current prediction value calculated for the patient, the improvement prediction value, the read improvement instruction information, and the like are described is generated. Then, this output data is transmitted to the printer 20, and an improvement instruction sheet as shown in FIG. 11 is printed out by the printer 20 (step S104).

  As shown in FIG. 11, in the improvement instruction sheet R, the name of the patient is described, the biometric attribute information and the number of lacunar infarct candidates for the patient are described as a list r11 of the consultation results, and the calculated current predicted value r12 is combined. It is described. Moreover, the improvement instruction information r13 and the improvement prediction value r14 about the improvement item judged that the change rate from the present prediction value is large among the improvement items are described. Further, more specific improvement instruction information r15 is also described. The improvement instruction information r13 corresponds to the improvement instruction information 1 read from the improvement information DB 154, and the improvement instruction information r15 corresponds to the improvement instruction information 2.

  As described above, according to the present embodiment, information such as the detection result of a lacunar infarction candidate, biometric attribute information, and the period until the onset of a diseased case patient is stored in a DB. Since the prediction value indicating the possibility of disease occurrence is calculated, it is possible to provide the doctor with prediction information of a disease that may become diseased in the future. In addition, because the prediction value is calculated using various information such as the detection result of the lacunar infarction candidate and the biometric attribute information, it is possible to make a comprehensive prediction about the onset of the disease, and to provide high-quality diagnosis support It becomes possible.

  Moreover, since the improvement prediction value is calculated using the temporary information and output as an improvement instruction sheet together with the improvement instruction information, the cause of the disease can be identified from the improvement items, and the direction to be improved is given to the patient. Can show.

  Further, when the predicted value is calculated by the nearest neighbor method, the biometric attribute information and the like of the case patient close to the target patient is displayed, so that it is possible to provide the doctor with reference information related to the disease. For example, the doctor can predict the type of disease that the target patient is predicted to develop from the disease name of the case patient.

In addition, embodiment mentioned above is a suitable example to which this invention is applied, and is not limited to this.
For example, case patients may be grouped when calculating a predicted value. For example, case patients are grouped by sex, age group, diseased disease, etc., and a multiple regression analysis formula is obtained for each group, or a population of the nearest neighbor method is formed. Then, when calculating the predicted value of the target patient, the multiple regression analysis formula of the group to which the target patient corresponds or the population in the nearest neighbor method is used. By grouping case patients that are more prone to trends, the accuracy of disease prediction can be improved.

  Further, in the present embodiment, the predicted value is obtained to provide as reference information at the time of diagnosis by the doctor, but it can also be used for diagnosis training of the doctor.

  In addition, when the nearest neighbor method is applied, the average value, variance value, minimum period, maximum period, etc. of the period until onset are obtained for each case patient used to calculate the predicted value, and this is used as a reference screen for detailed information. It may be displayed at d2. The doctor can obtain statistical information related to the disease by this display.

  In addition, information such as the result of the doctor's determination regarding the position and size of the lacunar infarction and the thickness of the existing blood vessel is stored in the case information DB 152 for the case patient, and this is an information item for calculating a predicted value. It is good also as setting to. The possibility of a lacunar infarction varies depending on the position, size, blood vessel, and the like, such as the larger the blood vessel, such as the main blood vessel, the higher the severity of the disease. Therefore, for the target patient from which a lacunar infarction candidate is detected, information such as the position, size, and vessel thickness of the detected lacunar infarction candidate can be obtained in the same manner and used for calculation of a predicted value. It is possible to obtain a highly predictive value.

It is a figure which shows the functional structure of the diagnosis assistance apparatus in this embodiment. It is a figure which shows the case patient DB example memorize | stored in the memory | storage part of FIG. It is a figure which shows case information DB example memorize | stored in the memory | storage part of FIG. It is a figure which shows the object patient DB example memorize | stored in the memory | storage part of FIG. It is a figure which shows the example of improvement information DB memorize | stored in the memory | storage part of FIG. It is a flowchart explaining the flow of the database process performed by a diagnosis assistance apparatus. It is a flowchart explaining the flow of the predicted value display process performed by a diagnosis assistance apparatus. It is a figure which shows the example of a patient list screen displayed on the display part of FIG. It is a figure which shows the example of a detailed screen displayed on the display part of FIG. It is a flowchart explaining the flow of the database process performed by a diagnosis assistance apparatus. It is a figure which shows the example of an improvement instruction | indication sheet.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Diagnosis support apparatus 11 Control part 12 Operation part 13 Display part 14 Communication part 15 Storage part 16 Candidate detection part 17 Image memory 18 I / F
20 Printer

Claims (12)

An abnormal shadow candidate detecting means for detecting an image region of an abnormal shadow candidate from a medical image obtained by photographing a patient;
Patient information storage means for storing, for each patient, biometric attribute information obtained for a plurality of affected patients and the detection result of the abnormal shadow candidate;
As a target sample with biometric attribute information and abnormal shadow candidate detection results for a plurality of patients as a mother sample, and biometric attribute information and abnormal shadow candidate detection results for a patient to be diagnosed as a variable by the nearest neighbor method A predictive value calculating means for calculating a predictive value indicating a possibility of future disease of the patient to be diagnosed based on a mother sample adjacent to the target sample ;
Display means for displaying, along with the calculated predicted value, the detection result of the patient's biological attribute information and abnormal shadow candidate that became the mother sample adjacent to the target sample;
A diagnostic support apparatus comprising: The diagnosis support apparatus according to claim 1 , wherein the biological attribute information includes at least information on a patient's age, sex, smoking, family history of disease, height, weight, body fat percentage, and blood pressure. The diagnosis support apparatus according to claim 1 , wherein the biological attribute information includes information on a period from when an abnormal shadow candidate is detected by the abnormal shadow candidate detection unit to when a disease occurs. The predicted value calculating means uses the detection result and the biometric attribute information of the abnormal shadow candidate, calculates a period that is expected to onset, according to claim 3, wherein the conversion of this to the predicted value Diagnosis support device. The diagnosis according to any one of claims 1 to 4, wherein the predicted value calculation unit groups the plurality of patients based on the biological attribute information and calculates a predicted value for each group. Support device. Setting means for setting temporary information to replace the biometric attribute information stored in the storage means;
The predicted value calculation unit, diagnostic support according to any one of claims 1 to 5, characterized in that to calculate the predicted value using the detected abnormal shadow candidate and the temporary information the set apparatus. The lesion to be detected by the abnormal shadow candidate detection means is lacunar infarction,
The predicted value predicted value calculating means calculates the claim 1, characterized in that said lacunar infarction dementia, which is an index value indicating the likelihood of developing future any disease of cerebral infarction or subarachnoid hemorrhage The diagnosis support apparatus according to any one of to 6 . The diagnosis support apparatus according to any one of claims 1 to 7 , further comprising an output unit that outputs a predicted value calculated by the predicted value calculation unit for the patient to be diagnosed. The diagnosis support apparatus according to claim 8 , wherein the output unit outputs information on detection performance of the abnormal shadow candidate detection unit together with the predicted value. The temporary information is biometric attribute information assuming a case where the living situation of a patient to be diagnosed is improved,
Improvement instruction information storage means for storing improvement instruction information indicating the content of improvement of the life of the patient corresponding to the improvement condition of the biometric attribute information in the case of the improvement,
The output means acquires biometric attribute information of a patient to be diagnosed and improvement instruction information corresponding to the inputted temporary information from the improvement instruction information storage means, and outputs the improvement instruction information together with the predicted value. The diagnosis support apparatus according to claim 8 or 9. The display means, according to claim 1 to 1, characterized biometric attribute information, the predicted value calculated by the predictive value calculating unit, that comparably displays the detection result of the abnormal shadow candidate detecting means for each patient The diagnosis support apparatus according to any one of 0 . The said display means displays the medical image of the target patient from which the abnormal shadow candidate was detected with the predicted value calculated about the target patient. Diagnosis support device.

Images