JP6422142B1 - Image diagnostic system for fertilized egg, image diagnostic program for fertilized egg, and image diagnostic method for fertilized egg. - Google Patents

Abstract

Provided is an image diagnostic system for a fertilized egg that can obtain a diagnosis result that can predict whether or not a normal child can be acquired by using only image data of a fertilized egg without subjective judgment.
The image data of a fertilized egg includes image data 10 whose pregnancy prognosis is known, and the image data 10 includes normal image data that has resulted in the acquisition of a newborn and abnormal image data of a chromosomal abnormality. The analyzing means 6 selects a plurality of teacher data from normal image data and abnormal image data, creates a discriminator from the teacher data, extracts a plurality of test data from the normal image data and abnormal image data, and identifies them Each test data is diagnosed by the tester, the correct diagnosis rate of the test data is calculated, and the creation of the discriminator, the extraction of the test data and the calculation of the correct check rate are repeated by changing the teacher data to be selected. Based on the plurality of correct diagnosis rates, an optimum classifier is determined from the plurality of classifiers obtained by repetition.
[Selection] Figure 1

Description

  The present invention relates to an image diagnostic system for fertilized eggs, an image diagnostic program for fertilized eggs, and an image diagnostic method for fertilized eggs.

  While Japan is in danger of declining birthrate, many patients suffer from infertility. In vitro fertilization or microinsemination can be used as a treatment for establishing pregnancy in infertile patients. In vitro fertilization is a treatment that removes egg cells from the ovaries (egg collection) and gives sperm outside the body to make a fertilized egg. It is a treatment to make. In any treatment, fertilized eggs are cultured and grown, frozen and stored once, thawed when the patient's condition is good, and returned to the uterus to establish pregnancy.

  The selection of the fertilized egg to be returned to the uterus is made by a doctor, laboratory technician, or cultivator subjectively by microscopic observation before thawing and after thawing. Only 30% of people get pregnant. That is, the accuracy of subjective judgment is low, and even if the judgment method is devised, there is a certain limit to raising the precision.

  On the other hand, the cost of treatment per treatment is high and is a self-paying burden. For this reason, if the accuracy of the determination of fertilized egg selection is low, the number of treatments increases and time is wasted, and in addition to psychological stress, economic stress is also great. Under such circumstances, infertility treatment requires both patients and society to reach pregnancy as soon as possible and with less economic burden.

  In this regard, in the medical field, various techniques for analyzing and evaluating diagnostic images using a computer have been proposed. For example, Patent Literature 1 proposes a diagnosis support technology that improves the accuracy of diagnosis by causing a computer to search for similar cases using image feature amounts when interpreting a diagnostic image. Patent Document 2 proposes a fertilized egg quality evaluation technique that uses a computer to realize information useful for quality evaluation of a fertilized egg without identifying cells in the fertilized egg using only the captured image. Yes.

JP 2002-230518 A JP 2010-181402 A

  However, the technique described in Patent Document 1 does not target a fertilized egg image, but the target image is a CT image targeting an organ, and it is possible to search for similar cases to improve diagnostic accuracy. It was a thing. That is, from the description of Patent Document 1, it has been difficult to derive a method for determining whether or not pregnancy can be established for an image of a fertilized egg. On the other hand, the technique described in Patent Document 2 is intended for an image of a fertilized egg, but it is limited to a technique for evaluating the quality of a fertilized egg, and is not a technique for directly determining whether pregnancy is established.

  The present invention solves the conventional problems as described above, and it is possible to obtain a diagnostic result capable of predicting whether or not a normal child can be acquired by using only image data of a fertilized egg without subjective judgment. An object of the present invention is to provide an image diagnostic system, an image diagnostic program for a fertilized egg, and an image diagnostic method for a fertilized egg.

In order to achieve the above object, an image diagnostic system of the present invention is an image diagnostic system for a fertilized egg using a machine learning technique, a storage means for storing image data of the fertilized egg, and image data of the fertilized egg. Receiving means for storing the data in the storage means, analysis means for analyzing the image data, and output means for outputting the analysis result by the analysis means, and the image data of the fertilized egg has a known pregnancy prognosis The image data that includes the image data that is known, and the pregnancy prognosis is known, is composed of normal image data that has resulted in the acquisition of a newborn, and abnormal image data of chromosomal abnormalities, A plurality of teacher data is selected from the normal image data and the abnormal image data, a discriminator is created from the teacher data, and a plurality of test data is generated from the normal image data and the abnormal image data. The test data is diagnosed by the classifier, the correct diagnosis rate of the test data is calculated, and the creation of the classifier, the extraction of the test data and the calculation of the correct diagnosis rate are selected. the repeated by changing the number of training data, based on the diagnostic accuracy of different said computed for each teacher data of the number of, and determine the optimal number of the teacher data, the number of the training data as the optimum number The creation of the discriminator, the extraction of the test data, and the calculation of the correct diagnosis rate are repeated again by changing the teacher data to be selected , based on a plurality of correct check rates obtained by the repeat again , An optimum classifier is determined from a plurality of the classifiers obtained by the repetition again .

The image diagnostic program of the present invention is a diagnostic image program for a fertilized egg for causing a computer to perform image diagnosis of a fertilized egg using a machine learning technique, and fetching image data of the fertilized egg into a storage means And an analysis step for analyzing the image data and an output step for outputting an analysis result by the analysis means. The image data of the fertilized egg includes image data whose pregnancy prognosis is known. The image data for which the pregnancy prognosis is known is composed of normal image data that has resulted in the acquisition of a live child and abnormal image data of chromosomal abnormality, and in the analyzing step, the normal image data and the abnormal image Selecting a plurality of teacher data from the data, creating a discriminator from the teacher data, the normal image data and the abnormal image A plurality of test data is extracted from the data, each test data is diagnosed by the classifier, a correct diagnosis rate of the test data is calculated, the classifier is created, the test data is extracted, and the correct diagnosis is performed The calculation of the rate is repeated by changing the number of the teacher data to be selected, and the optimum number of the teacher data is determined based on the correct diagnosis rate calculated for each teacher data having a different number, and the teacher data Using the number as the optimum number, the creation of the discriminator, the extraction of the test data, and the calculation of the correct diagnosis rate are repeated again while changing the selected teacher data, and a plurality of the correct values obtained by the second iteration are obtained. based on the diagnostic rate, you and determining the optimal classifier from a plurality of the classifiers obtained by repeating the re.

The diagnostic imaging method of the present invention is a diagnostic imaging method of a fertilized egg for causing a computer to perform diagnostic imaging of a fertilized egg using a machine learning technique, and includes a step of taking in image data of the fertilized egg into a storage means; An analysis step of analyzing the image data; and an output step of outputting an analysis result by the analysis means, wherein the image data of the fertilized egg includes image data whose pregnancy prognosis is known, and The image data with a known prognosis is composed of normal image data that has resulted in the acquisition of a live child and abnormal image data with chromosomal abnormalities. In the analysis step, a plurality of teachers are obtained from the normal image data and the abnormal image data. Select data, create a classifier from the teacher data, and extract a plurality of test data from the normal image data and the abnormal image data To diagnose the respective test data in the discriminator, calculates the diagnostic accuracy of the test data, creation of the classifier, an operation of extraction and the diagnostic accuracy of the test data, the teacher data to be selected repeatedly changing the number, the number of different said computed for each teacher data based on the diagnostic accuracy, to determine the optimum number of the teacher data, the number of the training data as the optimum number, the discriminator create, operations of extraction and the diagnostic accuracy of the test data, repeated again by changing the teacher data to be selected, a plurality of the said obtained by again repeated on the basis of the diagnostic accuracy by repeating the re determining an optimal classifier from a plurality of the classifiers obtained you characterized.

According to the present invention, it is possible to diagnose a fertilized egg image without subjective judgment by simply executing a computer program without using equipment such as photographing and measurement. The classifier that diagnoses the data is statistically created by machine learning based on the pregnancy prognosis image data whose pregnancy prognosis is known. It will be possible. Moreover, since the optimal number of teacher data is determined based on the correct diagnosis rate, it is possible to prevent a decrease in diagnostic accuracy due to overlearning of machine learning.

  The effects of the present invention are as described above, and according to the present invention, subjective judgment is made by simply executing a computer program on image data of a fertilized egg without using equipment such as photographing and measurement. Diagnosis is possible without any problems. The classifier that diagnoses the data is statistically created by machine learning based on the pregnancy prognosis image data whose pregnancy prognosis is known, so the diagnosis result predicts the possibility of acquiring a normal child with high accuracy. It will be possible.

The block diagram of the diagnostic imaging system of a fertilized egg which concerns on one Embodiment of this invention. The flowchart of the diagnostic imaging system which concerns on one Embodiment of this invention. The schematic diagram of an example of the normal image data of a fertilized egg. The schematic diagram of an example of the abnormal image data of a fertilized egg. The schematic diagram which showed the image data which amplified this in addition to the normal image data shown in FIG. 6 is a flowchart illustrating a process of determining the optimum number of teacher data in an embodiment of the present invention. The figure which showed the relationship between the number of teacher data and the correct diagnosis rate in one Embodiment of this invention. 6 is a flowchart illustrating a process for determining an optimal classifier in an embodiment of the present invention. The figure which showed the relationship between image data and the diagnostic result by an optimal discrimination device in one Embodiment of this invention. The figure which showed the parameter | index of the diagnosis which concerns on one Embodiment of this invention.

  Hereinafter, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 shows a configuration diagram of a fertilized egg image diagnostic system (hereinafter simply referred to as “image diagnostic system”) according to an embodiment of the present invention. The image diagnosis system 1 is a system that outputs determination information regarding the feasibility of pregnancy based on image data of a fertilized egg. The diagnostic imaging system 1 delivers various data such as image data of a fertilized egg to a plurality of medical institutions 1 and 2 via the Internet 8. Various calculation processes are performed by the analysis means 6 (CPU: central processing unit). Various data are input and set by the input means 4 such as a keyboard. An output result is output by output means 5 such as a display or a printer. The image data from the medical institution received by the receiving unit 2 is stored in the storage unit 7 for each medical institution. The analysis result by the analysis unit 6 is transmitted to a plurality of medical institutions 1, 2 and the like by the transmission unit 3.

  Image data handled by the image diagnostic system 1 includes image data whose pregnancy prognosis is known by performing artificial insemination or microinsemination (hereinafter referred to as “pregnancy prognosis known image data”) and image data to be diagnosed (hereinafter referred to as “pregnancy prognosis image data”). , “Diagnosis target image data.” The diagnosis target image data is data for which it is unknown whether pregnancy is established or not .. In Fig. 1, the receiving means 2 receives the diagnosis target image data 11 from a plurality of medical institutions 1, 2, etc. Are separately received via the Internet 8. Separately, the receiving means 2 receives the pregnancy prognosis finding image data 10. Both image data are in, for example, jpeg format, and both are stored in the storage means 7. .

  All the image data are subjected to image processing before being analyzed. The image processing is, for example, erasing of background artifacts and shadows of the storage container, and standardization processing of a digitized image. Therefore, the image data analyzed by the analysis means 6 is subjected to image processing in advance. The image processing only needs to be performed before analysis, may be performed by the same program, or may be performed by another program for image processing. Normally, in any case, the analysis unit 6 analyzes the image data after image processing. However, when performing image processing with the same program, the analysis unit 6 performs image processing on the image data before analysis. You may have the function to perform.

  The analysis means 6 derives judgment information regarding whether or not pregnancy is possible for the diagnostic object image data. As a premise for deriving such determination information, an optimum discriminator described later is created in the diagnostic imaging system 1 based on the pregnancy prognosis finding image data 10. That is, the information processing by the image diagnostic system 1 is roughly divided into information processing for creating an optimal classifier based on the pregnancy prognosis image data 10 and information processing for diagnosing diagnosis target image data using the optimal classifier. . In the present embodiment, an example in which these two information processes are executed by one program will be described, but each information process may be executed by another program.

  FIG. 2 shows a flowchart of the diagnostic imaging system 1 according to an embodiment of the present invention. FIG. 6 described later shows a flowchart showing a process for determining the optimum number of teacher data, and FIG. 8 shows a flowchart showing a process for determining the optimum discriminator. The image diagnosis system 1 is used by causing a computer to execute an image diagnosis program. In the present embodiment, the present invention will be described as the diagnostic imaging system 1. The flow of information processing of the diagnostic imaging system 1 shown in FIGS. 2, 6, and 8 shows steps that the diagnostic imaging program causes the computer to execute. It is also a thing. Therefore, the contents of information processing by the image diagnosis system 1 described in the present embodiment are also contents that the image diagnosis program causes the computer to execute.

  In FIG. 2, after initial setting (step 100), image data is read (step 101). As described above, the image data handled by the image diagnosis system 1 includes two types of image data 10 (see FIG. 1) for which the pregnancy prognosis is known and diagnosis target image data 11 (see FIG. 1) to be diagnosed. It is. Both image data may be read in succession, but if at least the pregnancy prognosis finding image data 10 is read, an optimum discriminator can be created (step 104). Thereafter, the diagnostic object image data 11 may be read. .

  Steps 102 to 104 are pre-stage steps before diagnosing the diagnosis target image data 11, and in this embodiment, in order to improve the accuracy of diagnosis, the optimum number of teacher data is determined and then optimized again. The number of teacher data is selected and the optimum classifier is determined. Although details will be described later, when the analysis process starts, the optimum number of teacher data is determined (step 103), the optimum classifier is subsequently determined (step 104), and the image data to be diagnosed is diagnosed by the optimum classifier. (Step 105).

  Both the analysis process for determining the optimum classifier and the analysis process for diagnosing image data by the optimum classifier utilize pattern recognition of supervised machine learning by artificial intelligence. An example of a program language for analysis processing is the Wolfram Language. As for machine learning pattern recognition, there are recognition methods such as logistic regression, naive Bayes, random forest, nearest neighbors, neural network, and deep learning. The Wolfram Language already has functions for handling these recognition methods, and you can select the best one from each recognition method.

  In FIG. 1, pregnancy prognosis finding image data 10 is classified into normal image data and abnormal image data. Normal image data is image data that resulted in the acquisition of newborns later. The abnormal image data is abnormal image data of chromosomal abnormality. It suffices that the abnormal image data has been found to be a chromosomal abnormality, and it does not matter whether the fertilized egg related to the abnormal image data is actually returned to the uterus to confirm the presence or absence of pregnancy. This is because it is clear that chromosomal abnormalities do not lead to the acquisition of normal live children. Therefore, abnormal image data may be those that became pregnant later but became miscarriage and proved to be chromosomal abnormality by chromosomal examination at that time. Pre-implantation screening (PGS) revealed chromosomal abnormality (aneuploid and mosaic). Or both of them may be mixed.

  When both the normal image data and the abnormal image data are analyzed by the analysis means 6, the formats are unified. For example, the image resolution is converted to a square of 100 × 100 pixels at, for example, 72 pixels / inch. Reference numeral 10 in FIG. 3 shows a schematic diagram of an example of normal image data, and reference numeral 11 in FIG. 4 shows a schematic diagram of an example of abnormal image data. Since each image data is converted into a square, the same square image data can be amplified by rotation and inversion.

  FIG. 5 shows image data obtained by amplifying the normal image data shown in FIG. As shown in FIG. 5, three pieces of image data are added by 90 degree rotation, 180 degree rotation, and 270 degree rotation, and one image data is inverted, inverted 90 degree rotated, inverted 180 degree rotated, inverted 270 degree. Four images are added by rotation, and when the original image data is included, eight image data are obtained. The same applies to the abnormal image data 11 shown in FIG. In this way, by amplifying the image data, one image data can be amplified to 8 times the number of image data, so that the characteristics of the image data can be expressed in many ways, and the classifier described later Accuracy can be increased.

  The unification of the format of the image data is not limited to the above, but may be changed as appropriate. For example, each image data may be converted into a circle, and the image data may be amplified by rotation and inversion. In the case of circular image data, by rotating the image data in small increments at a small angle, the number of amplified images can be increased, and the characteristics of the image data can be expressed in many ways, which is advantageous for improving the accuracy of the discriminator.

  The diagnostic imaging system 1 selects teacher data from pregnancy prognosis image data including normal image data and abnormal image data, and creates a discriminator from the data. In the present embodiment, the optimal number of teacher data is determined in advance in order to prevent a decrease in diagnostic accuracy due to overlearning of machine learning. FIG. 6 is a flowchart showing a process of determining the optimum number of teacher data.

  Hereinafter, the determination of the optimum number of teacher data will be described with reference to FIG. For the sake of convenience, a numerical example will be described. However, this is an example and the present invention is not limited to this. For example, in the following embodiment, the number of image data after amplification is 2600, but may be changed as appropriate at a level of several thousand or may be changed as appropriate at a level of several tens of thousands. In the following embodiment, the optimal number of teacher data is 1150, but if the original image data is tens of thousands, the optimal number of teacher data can be, for example, about 20,000.

  In this embodiment, the diagnostic imaging system 1 has received 125 normal image data and 200 abnormal image data as pregnancy prognosis finding image data 10 in FIG. 200 pieces of abnormal image data include image data that became pregnant later but became miscarriage and proved to be chromosomal abnormality by the chromosome examination at that time, and image data that was found to be chromosomal abnormality by preimplantation screening It is. As described above, the amplification of the image data amplifies the image data by eight times, so that the normal image data is 1000, the abnormal image data is 1600, and the image data is 2600 in total.

  In FIG. 6, 90 pieces of teacher data (n pieces) are set as initial values (step 110). Subsequently, 90 pieces of image data are selected as teacher data from 2600 pieces of image data (step 111). The breakdown of the 90 pieces of teacher data is the same number of 45 pieces of normal image data and abnormal image data (n / 2 pieces). A discriminator is created from the 90 pieces of teacher data (step 112).

  The discriminator is a discriminant function program that classifies input data into several classes. Here, classes are classified into two categories: normal and abnormal. An image feature is extracted from the teacher data, and a discriminator is obtained by setting parameters that can distinguish between normal and abnormal as much as possible. When image data to be diagnosed is input to the discriminator, either normal or abnormal results are output.

  As will be described later, the optimum discriminator in step 105 not only shows the two-class analysis results for normal and abnormal image data of each fertilized egg, but also outputs the probability of abnormality (and normal). The abnormality probability is a classifier that calculates the probability that the image data is abnormal based on the image characteristics of the image data to be diagnosed. When the discriminator diagnoses the image data as abnormal, the abnormality probability of the image data is 0.5 or more (the determination criterion may be changed as appropriate). When the abnormality probability is calculated, the normal probability is uniformly determined, and normal probability = 1−abnormal probability.

  Next, test data to be determined by the classifier is extracted (step 113). As test data, 500 (n1) image data are randomly extracted from 2510 image data obtained by excluding 90 image data selected as teacher data from 2600 image data. The breakdown of the 500 test data is the same number of normal image data and abnormal image data of 250 (n1 / 2). The total 500 test data is diagnosed by the discriminator created in step 112 (step 114).

There are two types of diagnosis, normal or abnormal, and 500 diagnosis results are obtained for 500 pieces of test data. Since the test data is image data that is known to be normal or abnormal, it is possible to determine whether the diagnosis result is correct. That is, when the diagnosis result is normal and the test data is normal, and when the diagnosis result is abnormal and the test data is also abnormal, a positive determination is made. On the other hand, when the diagnosis result is normal and the test data is abnormal, and when the diagnosis result is abnormal and the test data is normal, the determination is negative. In this embodiment, the correct diagnosis rate is defined by the following equation (1), and the correct diagnosis rate is calculated for the diagnosis results of 500 test data (step 115).
Formula (1) Correct diagnosis rate = number of test data for correct judgment / total number of test data

  After calculating the correct diagnosis rate for the 500 test data, 500 test data are extracted at random again, and Steps 113 to 115 are executed to calculate the correct diagnosis rate again. This is repeated 10 times (n2 times) including the calculation of the initial correct diagnosis rate. As a result, 10 correct diagnosis rates can be obtained.

  When the calculation of the correct diagnosis rate 10 times is completed, the number of teacher data, which is 90, is increased (step 116). Thereafter, steps 111 to 115 are repeated with the newly set number of teacher data, and 10 correct diagnosis rates are obtained again. Thereafter, by repeating the same procedure by increasing the number of teacher data, 10 correct diagnosis rates can be obtained for each number of teacher data.

  FIG. 7 is a diagram showing the relationship between the number of teacher data and the correct diagnosis rate. For example, when the teacher data is 200 pieces, the lower end 16 indicates a value obtained by subtracting the standard deviation value from the average value, the upper end 17 indicates a value obtained by adding the standard deviation value to the average value, and a point 15 indicates the average value. Is shown. The further the distance between the lower end 16 and the upper end 17, the greater the variation and the greater the standard deviation. In the result of FIG. 7, when the number of teacher data is 1000 to 1200, the correct diagnosis rate maintains a high value and the variation is small (see part A), and the high correct check rate is stable. Can be secured. On the other hand, the variation increases when the number of teacher data is 1200 or more. The analysis means 6 determines 1150 as the optimum number N of teacher data using the average value and standard deviation of the correct diagnosis rate as a scale (step 117).

  After determining the optimum number N of teacher data, the analysis means 6 performs a calculation process for determining the optimum classifier. FIG. 8 is a flowchart illustrating a process for determining an optimal classifier. First, 1150 (N) image data are selected as teacher data from 2600 image data (step 120). The breakdown of the 1150 teacher data was 575 (N / 2) of the same number of normal image data and abnormal image data. A discriminator is created from the 1150 teacher data (step 121).

  Next, test data to be diagnosed by the discriminator is extracted (step 122). As test data, 100 (N1) image data are extracted at random from 850 image data obtained by excluding 1150 image data selected as teacher data from 2600 image data. The breakdown of the test data of 100 sheets was 50 (N1 / 2 sheets) of the same number of normal image data and abnormal image data. The 50 test data are diagnosed by the discriminator created in step 121 (step 123).

  The diagnosis procedure is the same as in step 114 of FIG. 6, and there are two types, normal or abnormal. 100 diagnosis results are obtained for 100 test data, and the correct diagnosis rate defined by the above formula (1) The correct diagnosis rate is calculated for the diagnosis result of 100 pieces of test data (step 124). This is repeated 50 times (N2 times) including the calculation of the initial correct diagnosis rate.

  When the calculation of the correct diagnosis rate 50 times is completed, 1150 teacher data are newly selected at random, and steps 121 to 124 are repeated to obtain 100 correct diagnosis rates again. Thereafter, by repeating the same procedure, 100 correct diagnosis rates are obtained for each newly selected teacher data. The analysis means 6 employs the discriminator that yielded the highest correct diagnosis rate as the optimum discriminator (step 125). This step corresponds to step 104 in FIG.

  The analysis process described above is intended for pregnancy prognosis clarified image data whose pregnancy prognosis is known, but after that, the analysis target image data to be diagnosed is analyzed. That is, in FIG. 2, after determining the optimum classifier (step 104), diagnosis target image data is diagnosed by this optimum classifier (step 105). As shown in FIG. 1, diagnosis target image data 11 is individually received from a plurality of medical institutions 1, 2, etc. via the Internet 8.

  In the above analysis process, the breakdown of the teacher data and the test data has been described with an example in which the same number of normal image data and abnormal image data are used, that is, the ratio of the two image data is 1: 1. It is not a thing. The ratio of both image data should just determine the optimal ratio with which the correct diagnosis rate of a discriminator increases, for example, may be 2: 1 and may be 1: 2. Specifically, before determining the final ratio, the diagnostic imaging system 1 is operated on a trial basis, the ratio is appropriately changed, and the optimal ratio is determined while checking the correct diagnosis rate of the discriminator. do it.

  Hereinafter, the diagnosis index will be described. FIG. 9 shows the relationship between the image data and the diagnosis result by the optimum classifier. This figure is an arrangement of the diagnosis results for the pregnancy prognosis image data obtained by the optimum discriminator in the process for obtaining the optimum discriminator shown in FIG. “a” to “d” are breakdowns of 100 test data. a is the number of sheets diagnosed as positive (abnormal) with respect to abnormal image data and correctly diagnosed, b is the number of sheets diagnosed as negative (normal) with respect to abnormal image data, and is the number of sheets diagnosed incorrectly, c Is the number of sheets diagnosed as positive (abnormal) with respect to normal image data, and is the number of sheets diagnosed incorrectly. D is the number of sheets diagnosed as negative (normal) with respect to normal image data, and is the number of sheets correctly diagnosed.

  FIG. 10 shows diagnosis indexes, the correct diagnosis rate is the rate of correctly diagnosing as shown in the above formula (1), the sensitivity is the rate of correctly diagnosing abnormally fertilized eggs as “positive” (abnormal), and the specificity is The rate at which normal fertilized eggs are correctly diagnosed as “negative” (normal), the positive predictive value is the predictive value when positive is diagnosed, and the negative predictive value is the predictive value when negative is diagnosed. The calculation formula of each index is shown in the right column of FIG. 10, and can be obtained from a to d in FIG. 9a to 9d are verified from the pregnancy prognosis clarified image data whose pregnancy prognosis is known, and the value of each index obtained from a to d in FIG. 9 is highly reliable as the characteristic of the optimum discriminator. It will be a thing.

  For example, when the test data is normal image data and abnormal image data are halved, the positive predictive value is 0.775. When the diagnosis target image data is diagnosed by the image diagnosis system 1 according to the present embodiment, In addition, when “abnormal” is diagnosed, it is “abnormal” with a probability of 77.5%. Therefore, in this case, it can be determined that the fertilized egg should not be returned to the uterus. On the other hand, when the negative predictive value is 0.760, the probability that a live child can be acquired when diagnosis target image data is diagnosed by the image diagnosis system 1 according to the present embodiment is diagnosed as “normal”. Is 76%.

  As described above, the optimum discriminator obtained in step 105 not only shows two classification analysis results such as negative (normal) and positive (abnormal), but also calculates an abnormal probability of a fertilized egg related to individual image data as a premise thereof. doing. For example, if the abnormality probability of a fertilized egg related to certain image data is calculated as 0.678, this means that the probability that this fertilized egg is abnormal is 67.8%. For example, 0.5 is used as a negative or positive determination criterion, and 0.5 or higher is determined as positive (abnormal). Taking the case where the abnormality probability is calculated as 0.878 as an example, since the abnormality probability is 0.5 or more, it is determined that the fertilized egg related to the image data is qualitatively positive (abnormal). The Therefore, instead of using only the two-classification analysis result by the optimum classifier, if the abnormal probability (normal probability) is used, the fertilized eggs can be ranked so that the selection when returning the fertilized eggs to the uterus is performed. It becomes possible.

  As described above, the present invention uses a machine learning method, and it is possible to perform subjective judgment by simply executing a computer program on image data of a fertilized egg without using equipment such as photographing and measurement. Diagnosis is possible without doing so. The classifier that diagnoses the data is statistically created by machine learning based on the pregnancy prognosis image data whose pregnancy prognosis is known, so the diagnosis result predicts the possibility of acquiring a normal child with high accuracy. It will be possible.

DESCRIPTION OF SYMBOLS 1 Image diagnosis system of fertilized egg 2 Receiving means 3 Transmitting means 4 Input means 5 Output means 6 Analyzing means 7 Storage means

Claims (3)

An image diagnostic system for a fertilized egg using a machine learning technique,
Storage means for storing fertilized egg image data;
Receiving means for capturing the image data of the fertilized egg into the storage means;
Analyzing means for analyzing the image data;
Output means for outputting an analysis result by the analysis means,
The fertilized egg image data includes image data whose pregnancy prognosis is known,
The image data for which the pregnancy prognosis is known is composed of normal image data that has resulted in the acquisition of a live child, and abnormal image data of chromosomal abnormalities,
The analysis means includes
Selecting a plurality of teacher data from the normal image data and the abnormal image data;
Creating a classifier from the teacher data;
Extracting a plurality of test data from the normal image data and the abnormal image data;
Diagnose each test data with the classifier, calculate the correct diagnosis rate of the test data,
Creating the discriminator, extracting the test data, and calculating the correct diagnosis rate are repeated by changing the number of the teacher data to be selected,
Based on the correct diagnosis rate calculated for each teacher data having a different number, the optimum number of the teacher data is determined,
With the number of the teacher data as the optimum number, the creation of the discriminator, the extraction of the test data, and the calculation of the correct diagnosis rate are repeated again while changing the selected teacher data,
On the basis of the plurality of the diagnostic accuracy obtained by repeated again, embryo imaging system, characterized in that to determine the optimal classifier from a plurality of the classifiers obtained by repeating the re. An image diagnostic program for a fertilized egg for causing a computer to perform image diagnosis of a fertilized egg using a machine learning technique,
A step of taking in image data of a fertilized egg into a storage means;
An analysis step of analyzing the image data;
Causing the computer to execute an output step of outputting an analysis result by the analysis means;
The fertilized egg image data includes image data whose pregnancy prognosis is known,
The image data for which the pregnancy prognosis is known is composed of normal image data that has resulted in the acquisition of a live child, and abnormal image data of chromosomal abnormalities,
In the analysis step,
Selecting a plurality of teacher data from the normal image data and the abnormal image data;
Creating a classifier from the teacher data;
Extracting a plurality of test data from the normal image data and the abnormal image data;
Diagnose each test data with the classifier, calculate the correct diagnosis rate of the test data,
Creating the discriminator, extracting the test data, and calculating the correct diagnosis rate are repeated by changing the number of the teacher data to be selected,
Based on the correct diagnosis rate calculated for each teacher data having a different number, the optimum number of the teacher data is determined,
With the number of the teacher data as the optimum number, the creation of the discriminator, the extraction of the test data, and the calculation of the correct diagnosis rate are repeated again while changing the selected teacher data,
On the basis of the plurality of the diagnostic accuracy obtained by repeated again, imaging the fertilized egg you and determines the optimal classifier from a plurality of the classifiers obtained by the repetition of the re program. An image diagnostic method for a fertilized egg for causing a computer to perform image diagnosis of a fertilized egg using a machine learning method,
A step of taking in image data of a fertilized egg into a storage means;
An analysis step of analyzing the image data;
An output step of outputting an analysis result by the analysis means,
The fertilized egg image data includes image data whose pregnancy prognosis is known,
The image data for which the pregnancy prognosis is known is composed of normal image data that has resulted in the acquisition of a live child, and abnormal image data of chromosomal abnormalities,
In the analysis step,
Selecting a plurality of teacher data from the normal image data and the abnormal image data;
Creating a classifier from the teacher data;
Extracting a plurality of test data from the normal image data and the abnormal image data;
Diagnose each test data with the classifier, calculate the correct diagnosis rate of the test data,
Creating the discriminator, extracting the test data, and calculating the correct diagnosis rate are repeated by changing the number of the teacher data to be selected,
Based on the correct diagnosis rate calculated for each teacher data having a different number, the optimum number of the teacher data is determined,
With the number of the teacher data as the optimum number, the creation of the discriminator, the extraction of the test data, and the calculation of the correct diagnosis rate are repeated again while changing the selected teacher data,
On the basis of the plurality of the diagnostic accuracy obtained by repeated again, imaging the fertilized egg you and determines the optimal classifier from a plurality of the classifiers obtained by the repetition of the re Method.

Images