JP5456132B2 - Diagnosis support device, diagnosis support device control method, and program thereof - Google Patents

Description

  The present invention relates to a medical diagnosis support system that computer-processes data obtained from a subject and presents the obtained diagnostic information.

In the medical field, a doctor displays a medical image obtained by photographing a patient on a monitor, interprets the displayed medical image, and observes a state of a lesioned part and a change with time. As an apparatus for generating this kind of medical image,
-CR (Computed Radiography) device,
-CT (Computed Tomography) device,
-MRI (Magnetic Resonance Imaging) device,
-Ultrasonic device (US; Ultrasound System) etc. are mentioned.

  For the purpose of reducing the burden on doctors' interpretation, a diagnosis support apparatus has been developed that automatically detects a lesioned part by digitizing a medical image and analyzing the image to perform computer-aided diagnosis. Hereinafter, the computer-aided diagnosis is referred to as CAD (Computer-Aided Diagnosis). In such CAD, abnormal shadow candidates are automatically detected as lesions. In this abnormal shadow detection process, image data representing a radiographic image is processed by a computer to detect an abnormal tumor shadow representing cancer or the like, a high-density microcalcification shadow, or the like. Then, by presenting the detection result, it is possible to reduce the burden on the doctor's interpretation and improve the accuracy of the interpretation result.

  In order for doctors to avoid misdiagnosis at the time of interpretation, guidelines such as those described in Non-Patent Document 2 are provided.

  A diagnosis support apparatus that performs computer-aided diagnosis always considers the contradictory “sensitivity” and “misdiagnosis detection” balance when calculating abnormal shadow candidates [Patent Document 1].

  For example, if “sensitivity”, which is a parameter for adjusting the number of tumor shadow candidates extracted, is increased, the number of “misdiagnosis detections” that are actually extraction of shadows that are not tumors also increases.

  As described above, when “sensitivity” is increased, oversight can be reduced, but “misdiagnosis detection” (false positive lesion candidates are sometimes referred to simply as FP hereinafter).

Japanese Patent No. 3417595

Kawada / Niki / Omatsu, “Analysis of internal structure using 3D curvature of small lung masses by 3D chest CT images”, IEICE Transactions, D-II, Vol. J83-D-II, no. 1, pp. 209-218, January 2000 Judgment criteria and follow-up guidelines for CT screening for lung cancer using Single slice helical CT, "NPO Japan CT Screening Society"

  However, a system such as a diagnosis support apparatus that performs computer-aided diagnosis described in Patent Document 1 performs computer-aided diagnosis using only data obtained from a subject. That is, the examination history of the subject is not taken into consideration.

  The present invention has been made in view of the above problems, and an object of the present invention is to provide a mechanism for obtaining diagnostic information by computer processing that also considers the examination history of a subject.

In order to achieve the above object, a diagnosis support apparatus according to an aspect of the present invention comprises the following arrangement. That is, a database that associates the diagnosis support processing means for obtaining information on a lesion from the data of the subject and determining the medical importance of the lesion, and the information on the lesion and the progress indicating the severity of the lesion The diagnosis support processing means obtains the degree of progression from the database based on information about the lesion obtained from the data of the subject, and determines the medical importance of the lesion based on the degree of progression. .

  With the configuration of the present invention, it is possible to provide a mechanism for providing diagnostic information by computer processing in consideration of the examination history of the subject.

It is a figure which shows the apparatus structure of a diagnostic assistance apparatus system. It is a figure which shows the function structure of a diagnosis assistance apparatus. It is a flowchart which shows the process sequence of a diagnostic assistance apparatus. It is a figure which shows the example of an output of a diagnostic assistance apparatus. It is a flowchart which shows the process sequence which calculates the magnitude | size of the lesion candidate in a diagnosis assistance apparatus. It is a figure which shows the example of the data stored in the medical knowledge database of a diagnostic assistance apparatus. It is a flowchart which shows the process sequence which searches the similar image of a case in a diagnosis assistance apparatus.

  Hereinafter, preferred embodiments of a medical diagnosis support apparatus and method according to the present invention will be described in detail with reference to the accompanying drawings. However, the scope of the invention is not limited to the illustrated examples.

(First embodiment)
The CPU 100 mainly controls the operation of each component of the diagnosis support apparatus 1. The main memory 101 stores a control program executed by the CPU 100 and provides a work area when the CPU 100 executes the program. The magnetic disk 102 stores an operating system (OS), device drives for peripheral devices, various application software including a program for performing diagnosis support processing, which will be described later, and the like. The display memory 103 temporarily stores display data for the monitor 104. The monitor 104 is, for example, a CRT monitor or a liquid crystal monitor, and displays an image based on data from the display memory 103. The mouse 105 and the keyboard 106 are used by the user for pointing input and character input, respectively. The above components are connected to each other via a common bus 107 so that they can communicate with each other.

  In the present embodiment, the diagnosis support apparatus 1 can read image data and the like from the database 3 via the LAN 4. Alternatively, a storage device such as an FDD, a CD-RW drive, an MO drive, a ZIP drive or the like may be connected to the diagnosis support apparatus 1 and image data or the like may be read from these drives. Alternatively, a medical image or the like may be acquired directly from the medical image photographing apparatus 2 via the LAN 4.

  FIG. 2 shows a configuration example of the diagnosis support apparatus 1.

  In FIG. 2, the diagnosis support apparatus 1 includes a medical examination data input unit 201, a case database 202, a medical knowledge database 203, a diagnosis support processing unit 204 as a diagnosis support processing unit, a processing method change unit 207, and a history record as a storage unit. A unit 208 and a data output unit 209 are provided.

  The medical examination data input unit 201 is data related to the subject, such as image data acquired from, for example, an X-ray imaging apparatus, CT apparatus, MR apparatus, ultrasound or ultrasonic diagnostic apparatus, measurement data such as electrocardiogram, electroencephalogram data, and white blood cell count. To get.

  Moreover, the structure which can input the medical examination data containing the information relevant to lesion candidate acquisition, such as medical chart information, may be sufficient. In that case, the configuration may be such that these data can be directly input by the user, or can be read from various storage media such as FDD, CD-RW drive, MO drive, and ZIP drive in which information is recorded. There may be. Further, it may be configured such that it can be connected to a database storing these data via a LAN and received.

  The case database 202 stores image data taken by the medical examination data acquisition device 2, numerical data such as an electrocardiogram and a white blood cell count, and text data such as a medical record relating to the subject. Each case data can include a value of a data processing result by the diagnosis support apparatus and a definite diagnosis result, and the information can be used for similar case retrieval.

  Also, templates classified into categories used for recognition and identification processing are stored. This template is used for pattern recognition processing to determine what the lesion is (malignant, benign, etc.). Further, when a lesion is extracted, it is used to extract a region that is a lesion from similarity (for example, a correlation value) with data extracted from a subject.

  The medical knowledge database 203 stores diagnostic criteria for lesion areas and information on diseases to be examined including metastatic lesions and complications. Furthermore, data such as a diagnostic procedure such as a procedure for examining the presence or absence of a related disease when a primary lesion is detected may be stored.

  The diagnosis support processing unit 204 acquires information for supporting diagnosis from data obtained from the subject.

  When processing the image data, the processing unit 205 acquires information on the lesion from the data acquired from the subject. For example, lesions such as tumor shadow candidate regions and calculus candidate regions are extracted.

  In addition, when acquiring a lesion part candidate that is a candidate for a lesion part, the case database 202 may be referred to and the degree of similarity with past case data may be compared. In this case, the processing unit 205 calculates a feature amount from the image, and acquires a lesion candidate from the matching degree of the feature amount for each case stored in the case database 202.

  Data relating to diagnostic criteria and diagnostic procedures stored in the medical knowledge database 203 may be referred to. For example, if it is a CT image as a diagnostic criterion, there is a range of CT values corresponding to a lesioned portion, and there is a high possibility that a lesioned portion exists in that range. Therefore, the lesioned part is extracted from the range. Moreover, as a diagnostic procedure, a doctor's diagnostic trial is used as a flow and it is grammarized. For example, in the case of two-dimensional image data, a processing procedure required for computer processing such as extraction of a region of interest of a subject and quantification of texture features in the region of interest is grammarized according to the doctor's thoughts. Yes.

  The processing target of the processing unit 205 is not limited to image data. For example, measurement data or medical record data other than an image related to a subject is also a processing target, and a lesion candidate may be acquired based on the data. .

  When the image processing target is image data, the output processing unit 206 acquires information for supporting diagnosis such as the lesion likelihood of the lesion candidate extracted by the processing unit 205. In the case of measurement data, data obtained from the subject is directly analyzed, and information for supporting diagnosis is acquired and output.

  The history recording unit 208 records diagnosis history information related to the subject. As the diagnosis history information, an inspection method performed in the past, a diagnosis result at that time, and the like are recorded. In addition, the onset probability estimated in the past, the next diagnosis time according to the onset probability, and the like are also stored.

  In the inspection method, information on the purpose of inspection such as screening, detailed inspection, or progress monitoring of the lesion is also recorded.

  The processing method changing unit 207 determines the processing method of the processing unit 204 and the output processing unit 205 according to the history information of the subject obtained from the history recording unit 208.

  The data output unit 209 outputs the lesion candidate data obtained from the processing unit 205 and the lesion candidate determination information obtained by the output processing unit 206 as a specified format.

  Next, how the CPU 100 controls the diagnosis support apparatus 1 will be described with reference to the flowchart of FIG.

  In step S <b> 31, the CPU 100 inputs medical examination data from the medical examination data input unit 201 to the diagnosis support apparatus 1.

  In addition, since the lesion detection criteria and the lesion detection procedure at the time of acquiring the lesion candidate differ depending on the purpose of the examination, the examination history of the 208 subject from the history recording unit (for example, the history of the examination from the previous imaging date to the examination date) ) Get information for inspection purposes.

  In step S <b> 32, the processing method changing unit 207 estimates the current onset probability according to the history information of the subject recorded in the history recording unit 208 under the control of the CPU 100. Here, the onset probability refers to the probability that a lesion will develop. It also refers to the probability that a lesion will shift from benign to malignant.

  Symptom changes that can occur according to the time of the examination of the subject are stored as probabilistically useful information from past medical knowledge. Therefore, the symptom change can be predicted within a probabilistically significant range according to the information of the passage of time. That is, according to the examination history of the subject, the change in the symptom of the subject can be predicted within a probabilistically significant range.

  In estimating the onset probability, history information such as the past and current test data of the subject recorded in the history recording unit 208, the time interval for performing diagnosis such as the number of screenings, the onset probability estimated by the past screening, etc. Then, using the lesion candidate data of the image processing result, the progress degree of the lesion candidate is calculated.

  For example, the current size of the lesioned part is predicted from the information on the lesioned part (tumor, stone, etc.) of the subject recorded in the history recording unit 208. This prediction is performed using the past lesion size and growth rate. Here, as the growth rate, a statistical amount recorded in the medical knowledge database 203 using the size and type of a past lesion is used. Further, if a change in the size of the lesion is recorded in the history recording unit 208, it can be obtained from the size of the lesion at the past time point. For example, if the predicted size is 5 mm or more, it is determined that the onset probability is high. The onset probability here is the probability that the extracted lesion candidate is malignant. The processing method changing unit 207 also changes the size of the lesion part extracted by the processing unit 205 according to the onset probability. When this onset probability is high, there is a possibility that a new lesion part may occur, so that a lesion part of the entire size range is extracted. In some cases, the processing function is changed according to the size of the lesion.

  Further, when it is determined from the history of the subject recorded in the history recording unit 208 that the first visit is made, a lesion having the entire range of sizes is extracted with a high probability of onset.

  On the other hand, if there is no history of extraction of the lesion from the history of the subject recorded in the history recording unit 208, the onset probability changes according to the elapsed time from the previous diagnosis. If the time has not elapsed since the previous examination, the probability that a large lesion will occur is significantly reduced. Therefore, the processing unit 205 does not extract the large lesion because the onset probability is low. The parameter of the method change unit 207 is changed. Thereby, extraction according to a specific size can be performed, so that the false positive rate of the lesion part extracted by the processing unit 205 has an effect of decreasing.

  As described above, the processing method changing unit 207 changes the processing method of the diagnosis support processing unit 204 based on the history of the subject recorded in the history recording unit 208.

  As a method of changing the processing method of the diagnosis support processing unit 204, the probability distribution that gives the reliability is adjusted, the threshold value of the probability when it is recognized as abnormal, or the recognition function such as a Bayes discriminator is adjusted. There are methods such as changing parameters and moving the boundary surface in the feature space of the classifier. It is also possible to change the determination algorithm itself. In addition to the linear discriminant function, selecting a discriminating method such as a support vector machine, a Bayes discriminator, a neural network, or AdaBoost also corresponds to changing the processing method.

  The following setting is made as an example of the onset probability.

Onset probability = 1.0: Onset probability value cannot be estimated, or onset probability value that is recognized as default; in this state, the discriminator parameter used for default is used. Onset probability = 0.9: Onset probability is low Lowers the sensitivity of the discriminator and sets the discernment so that 90% false positive of the default state is obtained. Onset probability = 1.1: If the onset probability is high, increase the sensitivity of the discriminator and set the default. Setting the discriminative power so that 110% of false positives of the state are generated The above setting is merely an example and does not limit the present embodiment.

  In step S33, the processing unit 205 extracts lesion candidates by the processing method defined in step S32 according to the control of the CPU 100. At this time, by using the information of the diagnostic criteria stored in the medical knowledge database 203, lesion candidate data such as the reliability indicating the likelihood of lesion from the feature amount calculated from the lesion candidate and the progression degree are obtained. Here, the reliability indicates, for example, the probability of being a lesion from a correlation value with a feature value obtained from a past lesion.

  In the case of a chest CT image, the processing unit 205 divides regions such as lung field, diaphragm, bronchus, pulmonary artery, and pulmonary vein from the image, and classifies the lung field into upper lobe, middle lobe, lower lobe, and area. Here, as a method for detecting an organ region from a medical image, a level set method which is a kind of a dynamic contour method will be described as an example. In the case of the level set method, a level set function that is one dimension higher than the dimension of the detection target region is defined, and the region to be extracted is regarded as its zero contour line. Then, by updating this function based on the following evolution equation called a level set equation, the contour is controlled and the region is detected.

φ _t + F | ▽ φ | = 0
Here, φ _t represents a value obtained by first-order differentiation of the level set function in the time axis direction, F represents the growth rate of the contour, and | ▽ φ | represents the absolute value of the gradient of the level set function.

  In this way, an organ region can be detected from a medical image. In the above description, the organ region detection has been described by taking the level set method as an example. However, the region detection method includes a threshold processing method, a region expansion method, a dynamic contour method, clustering, a graph minimum cutting method, and the like. . An organ region is detected using any of these methods or other techniques.

  And these methods may be used by switching according to the site. Furthermore, not only using the image feature amount, but also region detection may be performed using a probabilistic atlas or a human body shape model as prior knowledge.

  In addition, as a method of detecting abnormalities such as lung masses from organ areas, filter processing for detecting abnormalities, pattern matching, abnormal detection by a discriminator, registration of past images and average shape images and diagnostic images, etc. There is a process of performing difference detection. In addition, as image feature amounts for specifying a lung field mass, there are a shape index and a curvedness obtained from a CT value and a three-dimensional curvature (Gaussian curvature, average curvature, principal curvature) of each pixel inside the tumor.

  A lesion candidate is detected in combination with any of the above or using other techniques.

  In step S <b> 35, the CPU 100 evaluates the lesion candidate detected by the processing unit 205 according to the processing method determined by the processing method changing unit 207.

  In order to perform abnormal disease classification and benign / malignant discrimination, the possibility of malignancy (probability value) is determined using a discriminator of a discrimination method (linear discriminant function, support vector machine, AdaBoost, Bayes discriminator, neural network, etc.) Disease classification and benign / malignant discrimination. These abnormality detection, disease classification, and benign / malignant discrimination are not limited to the above methods.

  Furthermore, it is possible not only to change the weight of the feature quantity used for the discriminator but also to consider or replace the feature quantity itself.

  Here, an example is shown in which a feature quantity extracted from an X-ray CT image is applied to the following linear discriminant function to perform benign / malignant discrimination of lung field tumor candidates. That is, candidate lesions are classified as true lesions and false positives:

Where x is a feature vector of one pattern, m ₁ and m ₂ are average vectors of classes 1 and 2, and Σ _W is an intra-class covariance matrix. The value of the linear discriminant function f is used as a discriminant score. When this value is negative, it is discriminated as class 1 (benign), and when it is positive, it is discriminated as class 2 (malignant). Then, change the parameters sigma _W of the linear discriminant function depending on the onset probability. That is, when it is determined that the onset probability is high, Σ _W is set so that f can be more positive, and when it is determined that the onset probability is low, Σ _W so that f can be more negative. Set.

  That is, if it is estimated that the onset probability is high, more lesion candidates are noted. That is, the sensitivity of the output processing unit 206 is increased. On the other hand, if it is estimated that the onset probability is low, the sensitivity is lowered so that fewer lesion candidates that are relatively unimportant are presented in order to reduce false-positive lesion candidates.

  The processing of the output processing unit 206 may be performed in conjunction with the sensitivity of the processing unit 205 or may be performed independently. That is, the sensitivity of the processing unit 205 can be maximized to extract many tumor shadows, and the output processing unit 206 can change the threshold value to reduce false positives. It is also possible to link with the sensitivity of the processing unit 205.

In step S36,
The candidate lesion data detected in step S32 is narrowed down to the candidate lesion data determined to satisfy the criteria recognized as a lesion in step S35 and output.

  Here, the data output unit 209 converts the output data according to the output destination under the control of the CPU 100. As output destinations, there are storage devices such as paper, memory, hard disk, and monitors.

  FIG. 4 shows an output example of lesion candidate data. A mark indicating a lesion candidate is displayed on the medical examination data, and an image feature amount, patient attribute / time-change data are displayed next to the mark. When the lesion candidate data is output to a display device such as a monitor, it may be displayed in a pop-up format or in a separate window.

  Next, referring to FIG. 5, the lesion candidate size prediction method in step S33 executed by the processing method changing unit 207 described above will be described. The predicted size is recorded in the history recording unit 208 as the history of the subject, and is used for the next diagnosis process or used for determining the next diagnosis time.

  Here, taking a tumor as an example, a method for prediction based on its size and growth rate will be described. Here, the longest diameter is used as the size of the tumor.

In step S51,
The tumor detected in step S32 is acquired from the current chest image data, and the longest diameter of the tumor is calculated.

In step S52,
It is examined whether or not there is a description of a lesion corresponding to the tumor acquired in step S51 in the subject's past chest image data or past diagnosis report. If the lesion has already occurred in the past, the process proceeds to step S53.

  If the lesion candidate detected in step S51 is detected for the first time in the current examination, the process proceeds to step S55.

In step S53,
The past information of the tumor is acquired from the history recording unit 208. Information to be acquired includes size (in this case, the longest diameter), case diagnosis, reliability, and the like. The information may be calculated by the diagnosis processing unit 204 if there is no description in the past diagnosis report.

In step S54
The growth rate is calculated by the following formula according to the corresponding longest diameter value of the current tumor, the past longest diameter value, and the period from the past examination. Here, the following formula is given as an example. However, the formula is not limited to this formula. For example, a weight or an adjustment term is added according to the site, and other growth rates that are recognized in the medical field are calculated. It may be a method.

V _C = (MR _present −MR _past ) / t where V _C is the growth rate, MR _present is the current longest diameter value of the mass, MR _past is the longest diameter value of the corresponding mass when examined in the past, t is the time elapsed since the previous examination.

In step S55,
If there is no past information of the detected mass, the value of the growth rate of the mass known empirically in the medical field is used. Of course, other test data of the subject or a formula or method for deriving the growth rate of the tumor in accordance with information such as past cases may be used.

In step S56,
The next examination scheduled date (next examination period) of the subject recorded in the history recording unit 208 is acquired, and at the next examination based on the current tumor size, growth rate, and the number of days until the next examination. Estimate the size of the mass at. As a size estimation method, for example, there are the following equations. However, the size may be estimated according to the experience value compiled in the table as medical knowledge without being limited to the formula.

MR _future = MR _present + VC ^* t _f where MR _future is the prediction of the size of the tumor at the next scheduled screening, and t _f is the time until the next scheduled screening.

In step S57,
The prediction of the size of the tumor at the next screening calculated in step S56 is compared with the medical guidelines to estimate the onset probability. For example, if the tumor size is predicted to be less than 5 mm, it can be determined that the onset probability is low. And if there is a prediction that the size of the tumor will be 10 mm or more, it can be determined that the onset probability is high.

  As described above, the size of the lesion at the next screening can be predicted by the processing as shown in FIG. 5, and the onset probability can be estimated by the processing method changing unit 207 using the information.

  This time, in order to express the size, it has been described using the longest diameter of the tumor, but the present invention is not limited to the above as long as the size and growth rate of the tumor can be expressed, such as using the volume of the tumor.

  Next, an example using medical importance as a method of estimating the onset probability in step S32 will be described.

  The processing method changing unit 207 estimates the onset probability according to the medical importance of the lesion candidate acquired by the processing unit 205. Alternatively, the onset probability can also be estimated according to the change in the medical importance of the lesion candidate.

  Here, the medical importance I calculated by the processing method changing unit 207 will be described.

The medical importance I can be defined as follows.
I = A ^* B ^* C
However,
A: Critical illness (relative severity among different diseases)
B: Progression level (stage) (Severity within the same disease)
C: Relevant disease severity.

  For example, when a plurality of types of diseases are detected in the processing unit 205, the related disease degree C for each lesion is set as follows according to the examination purpose. That is,

となる。

It becomes.

Here, examples of the important illness A and the progression B are shown in FIG. For example, in the case of follow-up of lung cancer, if one malignant tumor (stage 0) in the lung field is detected as a lesion candidate by the processing unit 205, the medical importance I _C regarding the malignant tumor is I _C = 10 ^* 2 ^*. 1 = 20
Next, to estimate the probability of onset, the patient compares to the medical significance of the corresponding lesion that was in the past. From the history recording unit 208, if the same tumors in the past examination has been determined that the benign tumor, its past medical importance I _P is IP = 5 * 2 * 1 = 10
It becomes. That is, since the medical importance has increased within a certain period, it is estimated that the onset probability is high. Or even if the medical importance does not change, if the value of a certain value or more is maintained for a certain period, it is estimated that the onset probability is high.

  Furthermore, the table (medical importance, change in medical importance, duration) made from cases collected by medical research is used to estimate the probability of occurrence due to changes in medical importance as described above. It is also possible to do.

  The content of the medical importance described here is merely an example, and is not limited to the definition in the present embodiment. For example, when estimating the onset probability based on medical importance, examination purpose information for screening, detailed examination, and follow-up observation may be considered.

  Next, a method for estimating the onset probability using the probability of a lesion candidate as a lesion (sometimes referred to as certainty factor) will be described.

  Here, as a method of calculating the probability of a lesion candidate as a lesion, detection of a lesion candidate by a similar image search can be mentioned. With reference to FIG. 7, the procedure of the similar image search process will be described.

In step S71,
The processing unit 205 processes medical examination data and calculates an image feature amount. Examples of the image feature amount include a luminance distribution obtained from general image processing, a 2D or 3D feature amount of a region of interest in the image, and the like. Further, a “shape index” value and a “curvedness” value may be used.

In step S72,
The case database 202 is searched using the feature amount of the image calculated by the processing unit 205, and an image case having a close feature amount is acquired. In addition, a case is associated with a feature amount and stored in advance, and case information is acquired from the feature amount with the case.

In step S73,
The searched images are arranged in order of the feature amount, and the similarity of the images is calculated in the order of the most similar images. Here, as an example of the similarity, an inner product value of normalized feature vectors of the reference image and the similar image in the feature space used for the search is used. That is, in the feature space, the closer the feature amount between the reference image and the similar image is, the closer the images are. The similarity calculation method described here is merely an example, and is not limited to the method in the present embodiment.

In step S74,
A finding associated with a case of the retrieved image or a feature amount obtained from the image is acquired. In the findings, the elements A, B, and C for obtaining the above-mentioned medical importance are attached. For example, cancers that progress rapidly, cancers with high metastasis, etc. are malignant and have a high medical importance. In addition, for the selected case, the certainty factor of the case when selected by pattern matching is acquired. Here, the certainty factor indicates, for example, the probability that a case is malignant when a plurality of doctors diagnose this case. In addition, the correct answer rate (establishment that the doctor's findings coincide with the output findings of the discrimination function) when the case is judged by the discrimination function may be used as the certainty factor as a statistical value. Next, the change T of the lesion similarity is defined as follows based on the obtained similarity and the certainty of the case.

T = S _present ^* C _present -S _past ^* C _past
However,
S _present : Similarity obtained in the current examination C _present : Confidence obtained in the current examination S _past : Similarity of the corresponding case in the past recorded in the history recording unit C _past : History If the certainty of the corresponding case in the past of the patient recorded in the recording unit and the above T has risen within a certain period, it is estimated that the onset probability is high.

  In addition to the above, in order to calculate the probability of a lesion candidate, for example, a support vector machine or other classifiers used in pattern recognition can be used, and the method is not limited to the above method using the similarity. .

  As described above, according to the first embodiment, sensitivity of diagnosis support can be adjusted by considering the history of a patient, and the burden on an interpreting doctor can be reduced. As a result, oversight and unnecessary biopsy are reduced, so that the burden on the patient can be reduced.

(Second embodiment)
In the first embodiment, it has been described that the sensitivity of diagnosis support is adjusted based on the onset probability of a patient. However, it has been assumed that test data obtained from a medical test data acquisition apparatus is acquired with standard parameters.

  In the second embodiment, the change of the acquisition parameter of the medical examination data acquisition apparatus according to the onset probability of the patient will be described.

  Although depending on the type of medical examination data acquisition device, if the clarity of the acquired examination data is increased, the load on the patient (invasion of examination) may be increased. For example, in the case of an X-ray CT apparatus, if an attempt is made to reduce the amount of exposure of a patient by reducing the voltage or the amount of electricity of the X-ray tube, the noise of the obtained cross-sectional image increases. Alternatively, it is possible to increase the moving speed of the bed as a method of reducing the total exposure dose to the patient without reducing the dose of the X-ray tube (X-ray source). There is a possibility that it will not be taken out of the image.

  In this embodiment, the acquisition parameter of the medical examination data acquisition apparatus is changed according to the onset probability of the patient calculated in the first embodiment.

  In the X-ray CT apparatus, when it is determined that the onset probability is low, the dose from the X-ray source is reduced or the cross-sectional interval is increased. On the other hand, when it is determined that the onset probability is high, the X-ray dose is increased or the cross-sectional interval is decreased in order to obtain a clearer medical image.

  Alternatively, if the onset probability at a specific site is high, it is possible to clearly perform imaging of only that site, or to inspect other sites by a method with less load.

  Alternatively, the inspection data calculation algorithm can be changed. In the case of an X-ray CT system, the reconstruction algorithm can be changed, so if the probability of onset is high, select an algorithm that can obtain a clearer image even if it is late, or select an early algorithm if the probability of onset is low. May be.

  Further, acquisition parameters such as the amount of contrast agent for MRI may be adjusted.

  Furthermore, the onset probability can be used to select a medical examination data acquisition device. For example, when it is estimated that the onset probability is high, it is possible to perform a test (for example, PET) more appropriate for the test of the case.

  As described above, according to the second embodiment, the characteristics of the case can be acquired in more detail by adjusting the acquisition parameter of the inspection apparatus according to the onset probability estimated from the history of the patient. it can.

(Third embodiment)
In the second embodiment, it has been described that a lesion having a high onset probability can be detected more accurately by changing an acquisition parameter of the medical examination data acquisition apparatus according to the onset probability of the patient.

  However, since the final diagnosis is performed by a doctor who is qualified to make a diagnosis, such as an interpreting physician or an attending physician, in order to confirm the diagnosis, the image diagnosis support device also provides information on the criteria for making the lesion a candidate in addition to the information on the lesion candidate. Must be presented.

  Here, the diagnostic imaging support apparatus 1 changes the patient's onset probability and sensitivity parameter (changed parameter, its value, or changed algorithm) together with the related information of the detected lesion candidate and the interpretation report and Present the reason and evidence to the doctor.

  Alternatively, the result output method may be displayed on a screen of a terminal used for interpretation in addition to the interpretation report on a paper medium, or other presentation method.

  As described above, according to the third embodiment, more accurate diagnosis can be performed by presenting the onset probability and change information of the sensitivity parameter together with the diagnosis report. In addition, the doctor can better grasp the patient's health and manage the information.

(Other examples)
An object of the present invention is to supply a computer-readable storage medium storing a program for realizing the functions of the above-described embodiments to a system or apparatus. Needless to say, this can also be achieved by the computer (or CPU or MPU) of the system or apparatus reading and executing the program code stored in the storage medium.

  In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiments, and the storage medium storing the program code constitutes the present invention.

  As a storage medium for supplying the program code, for example, a floppy disk, a hard disk, an optical disk, a magneto-optical disk, a CD-ROM, a CD-R, a magnetic tape, a nonvolatile memory card, a ROM, or the like can be used.

  Further, by executing the program code read by the computer, not only the functions of the above-described embodiments are realized, but also an OS (operating system) operating on the computer based on the instruction of the program code. Perform some or all of the actual processing. Needless to say, the process includes the case where the functions of the above-described embodiments are realized.

  Further, after the program code read from the storage medium is written into a memory provided in a function expansion board inserted into the computer or a function expansion unit connected to the computer, the function expansion is performed based on the instruction of the program code. It goes without saying that the CPU 100 or the like provided in the board or the function expansion unit performs part or all of the actual processing and the functions of the above-described embodiments are realized by the processing.

  In addition, the description in this Embodiment mentioned above is an example of the suitable diagnosis assistance apparatus which concerns on this invention, and this invention is not limited to this.

202 Case Database 203 Medical Knowledge Database 204 Diagnosis Support Processing Unit 205 Processing Unit 206 Output Processing Unit 207 Processing Method Change Unit 208 History Recording Unit

Claims (11)

Diagnostic information processing means for obtaining information on a lesion from data of a subject and determining the medical importance of the lesion;
A database associating information about the lesion with the degree of progression indicating the severity of the lesion,
The diagnosis support processing means obtains the degree of progression from the database based on information about the lesion obtained from the data of the subject, and determines the medical importance of the lesion based on the degree of progression. Diagnosis support device. The database stores important illnesses that are relative severity among different diseases as A, the progression as B, and the related illness as C,
The diagnosis support apparatus according to claim 1, wherein the diagnosis support processing unit calculates a medical importance I from an expression of I = A * B * C.   The diagnosis support apparatus according to claim 1, further comprising a changing unit that changes a size of a lesion portion to be processed in the diagnosis support processing unit according to an examination history of the subject. .   The changing means determines the size of the lesion portion of the subject stored in the storage means and the size of the lesion portion to be processed according to the elapsed time from the date when the lesion portion was imaged to the examination date. The diagnosis support apparatus according to claim 3, wherein the size of a lesion part to be processed is changed as a processing method of the diagnosis support processing unit. The changing means is
5. The diagnosis support according to claim 3, wherein the onset probability of a lesion is calculated according to the test history of the subject, and the processing method of the diagnosis support processing means is changed according to the onset probability. apparatus. The changing means is
6. The diagnosis support apparatus according to claim 3, wherein a parameter of a processing function, an algorithm of the processing function, or a threshold value of an output value of the processing function used in the diagnosis support processing unit is changed. Furthermore, it has a medical examination data acquisition device that acquires data from the subject,
The diagnosis support apparatus according to any one of claims 3 to 6, wherein the changing unit changes an acquisition parameter of the medical examination data acquisition apparatus. Furthermore, it has data output means,
The diagnosis support apparatus according to any one of claims 1 to 7, wherein change information of a processing method of the diagnosis support processing unit or a processing result of the diagnosis support processing unit is output. Diagnostic information processing means for obtaining information on a lesion from data of a subject and determining the medical importance of the lesion;
A control method in which a CPU controls a diagnosis support apparatus including information relating to a lesion and a database that associates a degree of progression indicating the severity of the lesion,
The diagnosis support processing means acquiring the progress degree from the database based on information on a lesion acquired from the data of the subject;
The diagnostic support processing means determining a medical importance of a lesion based on the degree of progression;
A diagnostic support apparatus control method comprising:   A program that causes a computer to execute the control method of the diagnosis support apparatus according to claim 9.   A computer-readable storage medium storing the program according to claim 10.

Images