WO2025192540A1 - Cranial correction effect estimation system and cranial correction effect estimation method - Google Patents

Abstract

[Problem] To provide a cranial correction effect estimation system and a cranial correction effect estimation method capable of providing an appropriate expectation value for a correction result of cranial correction. [Solution] A cranial correction effect estimation system includes a storage unit 70, a feature vector calculating unit 20, a feature vector processing unit 30 (distance calculating unit), and a display data processing unit 50. The storage unit 70 stores a plurality of pieces of time-series data including feature-related data corresponding to a plurality of cranial shape correction results in time series, and feature-related data of a subject, as subject data D17. The feature vector calculating unit 20 calculates a feature vector of the subject data D17 and a feature vector of the feature-related data, at a predetermined timing of the time-series data, in a predetermined feature space. The feature vector processing unit 30 calculates a distance between the feature vector of the subject data D17, and the feature vector of the feature-related data. The display data processing unit 50 visualizes the time-series data, including the feature-related data for which the distance has been calculated.

Description

Cranial correction effect estimation system and cranial correction effect estimation method

The present invention relates to a cranial correction effect estimation system and a cranial correction effect estimation method.

Conventionally, a system for estimating the effect of treatment on the body, including the head, is known, such as that described in Patent Document 1.

The system for estimating treatment effects described in Patent Document 1 stores information on treated patients (including, but not limited to, one or more of the following: images, location, size, progress, therapist age and gender, treatment method, treatment cost, treatment period, and post-healing images) for cases that have already been treated, and searches for information on patients seeking treatment, outputting predicted information such as effects and costs.

Patent No. 7386368

Incidentally, human infants, particularly those over about three months of age, can develop conditions such as plagiocephaly, dolichocephaly, or brachycephaly, which are conditions in which the head shape is distorted. In response to these conditions, a corrective method (hereinafter referred to as "helmet correction") is sometimes adopted in which a skull-correcting helmet is fitted to the infant's head for a specified period of time up to about 18 months of age, and the infant lives their daily life in that condition.

In such orthodontics, an explanation of the treatment results is often required. Conventionally, a treatment effect estimation system such as that described in Patent Document 1 is known that searches for treatment examples similar to the treatment target for birthmarks, blemishes, burn scars, wounds, cleft lip, etc., which are externally observable body targets for treatment.
On the other hand, for the specific symptom of skull deformation, there is a need for an explanation appropriate to the skull shape treatment process. In other words, because helmet correction involves gradually correcting a distorted head shape to the target shape, it is desirable to be able to explain how the skull shape changes as a result of the correction, including the treatment process through which the skull shape is corrected to the desired shape. In addition, especially when the patient is an infant, there is a strong desire to be able to provide parents and others making treatment decisions with detailed information on appropriate expectations for the results of the correction of the infant's skull shape, as a means of reassuring them.

The present invention was made in consideration of these circumstances, and its purpose is to provide a cranial correction effect estimation system and cranial correction effect estimation method that can provide appropriate expected values for the results of cranial correction.

A cranial correction effect estimation system that solves the above problem includes a storage unit that stores a plurality of time series data based on a plurality of cranial shape correction results, the time series data including feature-related data containing a plurality of features related to cranial shape in a time series, and stores the feature-related data of a subject as subject data; a subject feature vector calculation unit that calculates a feature vector of the subject data in a feature space based on selected features as one or more features selected from the plurality of features; and a subject feature vector calculation unit that calculates the feature-related data at a timing corresponding to the subject data from the time series of the time series data. and a resultant feature vector calculation unit that selects feature-related data and calculates a feature vector in the feature space for the selected feature-related data; a distance calculation unit that calculates the distance between the feature vector of the subject data and the feature vector of the feature-related data; and a display data processing unit that visibly displays the time-series data including the feature-related data for which the distance has been calculated, wherein the plurality of features include at least one of the subject's attributes of age in months, sex, and premature birth, and the subject's feature quantities of head circumference, anterior-posterior diameter, lateral diameter, volume ratio, brachycephalic ratio, frontal symmetry rate, occipital symmetry rate, CA, and CVAI.

A method for estimating the effect of cranial correction that solves the above problem includes the steps of: storing a plurality of time series data based on a plurality of cranial shape correction results in a memory unit, the time series data including feature-related data in a time series including a plurality of features related to cranial shape; storing the feature-related data of a subject as subject data; and calculating a feature vector of the subject data in a feature space based on selected features as one or more features selected from the plurality of features in a subject feature vector calculation unit; and calculating a feature vector of the subject data in a feature space based on selected features as one or more features selected from the plurality of features in a result feature vector calculation unit. The method includes a step of selecting the feature-related data and calculating a feature vector in the feature space for the selected feature-related data; a step of calculating the distance between the feature vector of the subject data and the feature vector of the feature-related data using a distance calculation unit; and a step of visibly displaying the time-series data including the feature-related data for which the distance has been calculated using a display data processing unit, wherein the plurality of features include at least one of the subject's attributes of age in months, sex, and premature birth, and the subject's feature quantities of head circumference, anterior-posterior diameter, lateral diameter, volume ratio, brachycephalic ratio, frontal symmetry rate, occipital symmetry rate, CA, and CVAI.

With this configuration or method, when the time-series data is displayed visually, it is possible to take into account the distance calculated between the feature vector of the subject data and the feature vector of the feature-related data included in the time-series data. Furthermore, rather than calculating the distance for each feature, i.e., by grouping, it is possible to calculate the distance as a feature vector in a feature space in which selected features are combined. By taking into account the distance between the subject data and the feature-related data of the time-series data, it becomes possible to compare the similarity between multiple cranial shape correction results and the subject data based on the distance.

In other words, it will be possible to extract and display appropriate cranial shape correction results so that parents and others can know the results of their infant's cranial shape correction when making treatment decisions. This will make it possible to provide appropriate expectations for the results of cranial correction.

In a preferred configuration, the distance calculation unit calculates the distance between the feature vector of the subject data and each of the feature vectors of the feature-related data, and extracts one or more feature vectors of the feature-related data that are closest to the feature vectors of the feature-related data from the calculated distances, and the display unit displays each of the corresponding time-series data that includes the extracted one or more feature-related data.

With this configuration, it is possible to extract cranial shape correction results that are highly similar to the subject data based on the close distance in feature space between the feature vector of the subject data and the feature-related data of the time-series data.

In a preferred configuration, the storage unit stores multiple sets of time-series data corresponding to multiple cranial shape correction results.

With this configuration, the results of cranial shape correction are stored for each correction case (past case), and each cranial shape correction result is extracted as a correction case (case). In other words, it becomes possible to predict the correction results based on the case.

In a preferred configuration, the storage unit stores one or more pieces of time-series data corresponding to one or more statistical processing results based on multiple cranial shape correction results.

With this configuration, the results of cranial shape correction are stored as the result of statistical processing of several correction cases (past cases), so the results of cranial shape correction can be extracted as statistical results. In other words, it becomes possible to estimate the results of correction based on statistical results.

In a preferred configuration, the feature space is clustered to reduce the dimension to a value lower than the dimension constituted by the plurality of features, the features are quantified as feature quantities, and the feature vector is projected onto the reduced-dimensional feature space.

With this configuration, the feature space can be defined based on the basis and contribution rate of the feature vectors that represent the distribution, which are calculated from the distribution of feature vectors in past cases using an algorithm such as PCA.

In a preferred configuration, the subject feature vector calculation unit calculates the feature amounts by setting weights separately for each of the multiple features, or the result feature vector calculation unit calculates the feature amounts by setting weights separately for each of the multiple features, or the distance calculation unit calculates the distance between the feature vector of the subject data and the feature vector of the feature-related data by weighting each feature.

With this configuration, weights can be applied to feature vectors, making it possible to calculate distances based on features that reflect the weights.

In a preferred configuration, the weights are assigned according to their contribution to the distribution, or according to the magnitude of their influence on the results of multiple cranial shape corrections.

With this configuration, similar cases can be extracted based on the distance to the feature vectors of past cases, which is calculated using weighting according to the contribution rate to the distribution in the feature space. Alternatively, similar cases can be extracted based on the distance to the feature vectors of past cases, which is calculated using weighting according to the magnitude of the impact on the results of multiple cranial shape corrections.

The present invention makes it possible to provide appropriate expectations regarding the results of cranial orthodontic treatment.

FIG. 1 is a block diagram showing the functional configuration of a cranial orthodontic effect estimation system and a cranial orthodontic effect estimation method according to an embodiment of the present invention. FIG. 2 is a schematic diagram showing an example of the skull shape of a subject according to the embodiment. FIG. 2 is a schematic diagram showing an example of a cranial correction helmet in the same embodiment. 5A and 5B are schematic diagrams showing data structures in the embodiment, in which FIG. 5A shows the structure of correction result data, and FIG. 5B shows the structure of subject data. FIG. 4 is a diagram showing an example of attribute data according to the embodiment. FIG. 10 is a diagram showing an example of result data according to the embodiment. FIG. 4 is a diagram showing an example of feature-related data in the embodiment. 10 is a flowchart showing the flow of a feature space processing step in the embodiment. 10 is a flowchart showing the flow of a feature vector processing step in the embodiment. 10 is a flowchart showing the flow of a display data processing step in the embodiment. Graphs showing the change in skull shape over time in this embodiment, where (a) is a graph showing the change in anterior-posterior diameter, (b) is a graph showing the change in lateral diameter, and (c) is a graph showing the change in head circumference. Graphs showing the change in cranial shape over time in the same embodiment, where (a) is a graph showing the change in brachycephalic ratio, (b) is a graph showing the change in frontal symmetry ratio, and (c) is a graph showing the change in occipital symmetry ratio. 6A and 6B are graphs showing the change in cranial shape over time in the same embodiment, where FIG. 6A shows the change in CA and FIG. 6B shows the change in CVAI. Tables showing the change in cranial shape over time in another embodiment, where (a) is a table showing data related to the characteristics of the subject, (b) is a table showing the average values of data accumulated at the time of examination for past subjects, and (c) is a table showing the average values of treatment results data for past subjects. 10 is a graph showing the relationship between the number of scans (elapsed time) and the graduation rate (percentage of patients who have completed treatment) in another embodiment. Graphs showing the mean and standard deviation of cranial shape over time in another embodiment, where (a) is a graph showing the relationship between the number of scans and head circumference, and (b) is a graph showing the relationship between the number of scans and brachycephalic rate. Graphs showing the mean and standard deviation of skull shape over time in another embodiment, where (a) is a graph showing the relationship between the number of scans and the frontal symmetry rate, and (b) is a graph showing the relationship between the number of scans and the occipital symmetry rate. Graphs showing the mean value and standard deviation of the cranial shape over time in another embodiment, where (a) is a graph showing the relationship between the number of scans and CA, and (b) is a graph showing the relationship between the number of scans and CVAI.

With reference to Figures 1 to 13, one embodiment of a cranial correction effect estimation system and cranial correction effect estimation method will be described.

As shown in Figure 1, the cranial correction effect estimation system 10 is a system that provides a corrected shape 121 (see Figure 2) as an estimated post-correction cranial shape for the input cranial shape 101 (see Figure 2) of the subject 100. In more detail, the cranial correction effect estimation system 10 provides an estimated correction result for the input cranial shape 101 of the subject 100 based on accumulated data (such as correction result data D16) on the correction results of multiple other subjects (hereinafter referred to as "past subjects") who are different from the subject 100 and have undergone cranial shape correction in the past. In other words, the subject 100 input here has an uncorrected cranial shape or has undergone incomplete cranial shape correction.

For example, the correction shape 121 (see Figure 2) provided by the cranial correction effect estimation system 10 is an example of a cranial shape estimated as a result of correcting the cranial shape 101 (see Figure 2) by having the subject 100 wear a cranial shape correction helmet 61 on their head.

The subject 100 is preferably an infant with a flexible skull that can be corrected, and more specifically, an infant between the ages of a newborn and approximately six months old. Note that the subject 100 may be referred to as a patient when undergoing medical treatment.

Referring to Figure 3, the cranial shape correction helmet 61 comprises an outer shell 62 made of a hard resin that is difficult to deform, and an inner lining 63 that has flexibility, such as elasticity, and is placed between the outer shell 62 and the skull of the subject 100. The cranial shape correction helmet 61 is provided with the outer shell 62 and inner lining 63 so that an appropriate force is transmitted from the cranial shape correction helmet 61 to the skull of the subject 100 (see Figure 1). Furthermore, in order to make the cranial shape correction helmet 61 lightweight and to reduce the burden on the subject 100 (see Figure 1), it is preferable that the outer shell 62 be made of a light material such as polystyrene foam, and the inner lining 63 be made of a light material such as sponge or urethane.

The outer shell 62 is preferably designed to reduce internal stuffiness, for example with many ventilation holes. The interior lining 63 is also preferably designed to absorb sweat and reduce dermatitis, and is also preferably designed to adjust the corrective force exerted on the skull by adjusting its thickness and elasticity, and is also preferably replaceable for maintaining cleanliness.

With reference to Figure 1, details of the cranial correction effect estimation system 10 and the equipment connected to the system will be described.

The cranial correction effect estimation system 10 has a main processing server 11. The cranial correction effect estimation system 10 may also have the main processing server 11 connected to a network NW, and a subject-side terminal 31, a diagnosis-side terminal 41, and a manufacturing device 51 connected via this network NW so that information can be exchanged between them. The main processing server 11, subject-side terminal 31, diagnosis-side terminal 41, and manufacturing device 51 connected via the network NW may be so close that at least two of them are bus-connected, or may be connected to each other via short-range communication. Any one of these two or more devices may be connected to the network NW.

Subject-side terminal 31 outputs cranial shape measurement information, which is information required to calculate the cranial shape 101 (see Figure 2) of subject 100. Subject-side terminal 31 is an information processing device including a camera for measuring subject 100, such as a digital camera, mobile phone, smartphone, tablet terminal, or small computer. Subject-side terminal 31 can obtain information about subject 100's cranial shape 101 (see Figure 2) by measuring subject 100, and can also transmit information via network NW.

The subject-side terminal 31 can acquire images or videos of the overall external shape of the subject's 100's skull as skull shape measurement information. For example, the subject-side terminal 31 should be able to capture images or videos of the entire circumference and top of the subject's 100 skull with the subject's skin visible. The subject-side terminal 31 may also be a device that can measure the skull shape of the subject 100 more accurately than images or videos, such as a 3D scanner using a laser.

The subject-side terminal 31 comprises an input unit 32 and an output unit 33. The input unit 32 is a unit for acquiring information from outside the subject-side terminal 31, and is a unit for inputting information from the network NW via a touch panel, camera, microphone, etc. The output unit 33 is a unit for outputting information from inside the subject-side terminal 31, and is a unit for outputting information to the network NW via an image display device, audio device, etc.

The diagnostic side terminal 41 is a terminal operated by the diagnostician 110, and the information displayed on the terminal can be shared with related parties such as the subject 100 and the subject's parents. The diagnostic side terminal 41 is an information processing device including an image display device and an instruction input device, such as a smartphone, tablet terminal, or small computer. The diagnostic side terminal 41 can display information about the subject 100 obtained from the main processing server 11 connected via the network NW on the image display device. The diagnostic side terminal 41 can also set adjustment parameters including adjustment amounts corresponding to the information about the subject 100 whose image is displayed, and may also be able to transmit the set adjustment parameters to the main processing server 11.

The diagnosis side terminal 41 is equipped with an input unit 42 and an output unit 43. The input unit 42 is a unit for acquiring information from outside the diagnosis side terminal 41, and includes a keyboard, mouse, camera, microphone, and other units for inputting information from the network NW. The output unit 43 is a unit for outputting information from inside the diagnosis side terminal 41, and includes an image display device, audio output device, and other units for outputting information to the network NW. The diagnosis side terminal 41 can also register, for example, shape changes to the outer shell 62 and interior lining 63 of the cranial shape correction helmet 61.

The subject-side terminal 31 and the diagnosis-side terminal 41 can display the subject data D17 included in the subject information from the main processing server 11, as well as the correction result data D16 as an example of the estimated skull shape after correction. The subject-side terminal 31 and the diagnosis-side terminal 41 should be able to display the subject data D17 and correction result data D16 from the main processing server 11 in a recognizable manner using graphs and tables in various formats.

Based on the information obtained via the diagnosis terminal 41, the diagnoser 110 understands the cranial shape of the subject 100, determines whether correction is necessary, determines whether the shape and correction power of the cranial shape correction helmet 61 are appropriate, and gives new instructions. For example, if cranial shape correction is provided as a medical treatment, the diagnoser 110 should be a medical professional such as a doctor.

The manufacturing device 51 is a device that manufactures the outer shell 62 of the cranial shape correction helmet 61 based on production data. The manufacturing device 51 manufactures the cranial shape correction helmet 61 made of foamed resin such as polystyrene foam, for example, using a 3D printer, resin molding, or cutting from raw material.

The manufacturing apparatus 51 is equipped with an input unit 52 and an output unit 53. The input unit 52 is a unit for acquiring information from outside the manufacturing apparatus 51, and includes a keyboard, mouse, touch panel, microphone, etc., and units for inputting information from the network NW. The output unit 53 is a unit for outputting information from inside the manufacturing apparatus 51, and includes a unit for outputting information to the network NW, such as an image display device or audio output device.

Next, we will explain the main processing server 11 that makes up the cranial correction effect estimation system 10.

The main processing server 11 is an information processing device such as a computer or server, and has an input unit 12, an output unit 13, an information processing unit 14, and a memory unit 70.

The input unit 12 is an input interface through which necessary information is input, such as from a keyboard, mouse, or touch panel, a camera or microphone, information input from other devices or servers via the network NW, or information input from an external storage device.

The output unit 13 is an output interface that outputs necessary information, such as to an image display device, character display device, or audio device, to other devices or servers via the network NW, or to an external storage device.

The information processing unit 14 is a part of the main processing server 11 that performs information processing and is made up of a computer device or the like. For example, the information processing unit 14 has a central processing unit, volatile memory, non-volatile memory, and an input/output interface. The input/output interface is capable of communicating with the input unit 12, output unit 13, memory unit 70, etc.

The storage unit 70 is a part that can store various information such as subject information, and is capable of reading, registering, and erasing stored information between it and the information processing unit 14. The storage unit 70 is a part that consists of internal storage, external storage, or a combination of these, and is composed of one or more of, for example, a hard disk, SSD, USB memory, etc. For example, the storage unit 70 may also include a cloud system that exchanges information by information communication via the network NW.

For example, the storage unit 70 stores attribute data D15, correction result data D16, and subject data D17. More specifically, the storage unit 70 stores data related to cranial correction accumulated for multiple past subjects C1 to Cn (n is an integer; see Figure 5, etc.). That is, the storage unit 70 stores attribute data D15 and correction result data D16 corresponding to multiple past subjects C1 to Cn, and subject data D17 corresponding to subject 100. The multiple past subjects C1 to Cn preferably include 500 or more, more preferably 1,000 or more, and even more preferably 2,000 or more.

Attribute data D15 is data that includes IDs corresponding to subject 100 and multiple past subjects C1 to Cn, as well as subject information associated with those IDs.

Referring to Figure 5, the attribute data D15 may include subject information associated with an ID, such as date of birth L01, gender L02, premature birth L03, treatment start date L04, age at treatment start L05, age at treatment end L06, treatment period L07, and number of hospital visits L08. For example, the IDs "C1", "C2", and "C3" represent different past subjects C1, C2, and C3, respectively.

The correction result data D16 shown in Figure 1 includes data showing the correction results accumulated for past subjects C1 to Cn.

As shown in FIG. 4(a), the correction result data D16 includes result data D20 and multiple time series data DC1, DC2, DC3, DC4, ..., DCn (n is an integer).

Referring to Figure 6, the result data D20 includes information associated with past subjects C1 to Cn by ID, including information related to head circumference, brachycephalic ratio, frontal symmetry ratio, occipital symmetry ratio, CA, and CVAI.

In more detail, the result data D20 may include an initial value L10, a final value L11, and a change amount L12 for head circumference, an initial value L20, a final value L21, and a change amount L22 for brachycephalic ratio, or an initial value L30, a final value L31, and a change amount L32 for frontal symmetry ratio. Furthermore, the result data D20 may include an initial value L40, a final value L41, and a change amount L42 for occipital symmetry ratio, an initial value L50, a final value L51, and a change amount L52 for CA, or an initial value L60, a final value L61, and a change amount L62 for CVAI.

The multiple time-series data DC1-DCn are based on multiple cranial shape correction results corresponding to multiple past subjects C1-Cn.

A single piece of time-series data DC1 is data that includes correction results corresponding to past subject C1, who is any one of multiple past subjects C1 to Cn. In other words, the storage unit 70 stores multiple pieces of time-series data DC1 to DCn corresponding to multiple cranial shape correction results.

Referring to Figure 4(a), the multiple time-series data DC1-DCn each include multiple feature-related data J1-Jm. Each of the feature-related data J1-Jm includes multiple features. For example, one time-series data DC1 includes feature-related data J1-Jm, including multiple features related to skull shape, in a time series from timing "1" to timing "m." For example, each timing may be a measurement time, such as a scan. Note that, because the number of measurements (scans) varies for each of the multiple past subjects C1-Cn, the number of feature-related data J1-Jm for each of the multiple time-series data DC1-DCn is arbitrary. Therefore, the value indicated by "m" may be different for each of the subjects C1-Cn, and the same applies below.

Each of the multiple feature-related data J1 to Jm is a set of data consisting of data obtained by measuring multiple features of the skull shape at different times. One piece of feature-related data (for example, feature-related data J1) is obtained for each measurement. In other words, a set of data measured for each measurement is accumulated as feature-related data J2, J3, J4, ..., Jm. In this way, the multiple feature-related data J1 to Jm are obtained as time-series data.

In more detail, the data from the first timing is feature-related data J1, the data from the second timing is feature-related data J2, and the data from the final (mth) timing is feature-related data Jm.

Referring to FIG. 7, each of the plurality of feature-related data J1 to Jm may include, as information associated with the above-mentioned ID, the measurement date K10, the anterior-posterior diameter K11, the lateral diameter K12, the head circumference K13, the brachycephalic ratio K14, the frontal symmetry ratio K15, the occipital symmetry ratio K16, CAK17, and CVAIK18.

The subject data D17 shown in Figure 1 includes feature-related data JH1 (see Figure 4(b)) of subject 100 obtained by measuring the skull shape of subject 100. The feature-related data JH1 (see Figure 4(b)) of subject 100 includes a plurality of measured features. For example, the feature-related data JH1 is data obtained by measuring the skull shape of subject 100 before cranial correction was performed.

The feature-related data JH1 shown in Figure 4(b) includes a set of information (see Figure 7) consisting of the same content as each of the multiple feature-related data J1 to Jm. In other words, the storage unit 70 stores the feature-related data JH1 of the subject 100 as subject data D17.

The main processing server 11 includes a subject information management unit 15, a correction result data processing unit 16, a subject data processing unit 17, and a feature space processing unit 18, all of which function through predetermined program processing in the information processing unit 14. The main processing server 11 also includes a feature vector calculation unit 20, a feature vector processing unit 30, a correction result extraction unit 40, and a display data processing unit 50, all of which function through predetermined program processing.

The subject information management unit 15 is a unit that manages information about the subject 100, including subject information that is information necessary for cranial shape correction of the subject 100, and manages, for example, information such as that recorded in a medical chart. The subject information management unit 15 also manages cranial shape measurement information for the cranial shape corresponding to the subject 100, subject data D17, various parameters, etc.

The subject information management unit 15 may manage information on multiple past subjects C1 to Cn, similar to the information management for subject 100. This various data is stored, for example, in the storage unit 70.

The correction result data processing unit 16 performs processing on the correction result data D16. For example, the correction result data processing unit 16 performs predetermined processing on multiple time series data DC1 to DCn corresponding to multiple past subjects C1 to Cn included in the correction result data D16.

The correction result data processing unit 16 may, for example, apply standardization processing to each item of feature-related data J1 to Jm. Standardization processing also makes it possible to favorably compare feature-related data J1 to Jm, which are made up of combinations of values with different physical dimensions.

At this time, the correction result data processing unit 16 may perform weight calculations and standardization processing based on the degree of influence on correction based on each item of feature-related data J1 to Jm included in one piece of time-series data DC1. The correction result data processing unit 16 may also perform weight calculations and standardization processing based on the degree of influence on correction based on each item of feature-related data J1 to Jm included in multiple pieces of time-series data DC1 to DCn.

For example, the correction result data processing unit 16 may calculate weights to be assigned according to the contribution rate to the distribution.

The subject data processing unit 17 performs predetermined processing on the subject data D17 in the same manner as the correction result data processing unit 16. For example, the subject data processing unit 17 may perform standardization processing on each item of the feature-related data JH1 included in the subject data D17. Standardization processing makes it possible to favorably compare subject data D17, which consists of a combination of values with different physical dimensions, with feature-related data J1 to Jm. The subject data processing unit 17 may also perform processing to reflect preset weights in the subject data D17.

The feature space processing unit 18 extracts multiple features (selected features) from the feature-related data J1-Jm processed by the correction result data processing unit 16. The feature space processing unit 18 also represents and processes the feature-related data J1-Jm and JH1 in a feature space expressed based on the multiple extracted features (selected features). The feature-related data J1-Jm are included in the correction result data D16, and the feature-related data JH1 is included in the subject data D17. In other words, the feature space processing unit 18 performs processing to define a feature space based on the selected features, which are one or more features selected from the multiple features.

The multiple features extracted for the feature space here include at least one of the multiple attributes and multiple feature quantities of subject 100 and multiple past subjects C1 to Cn.

The multiple attributes include at least one of the following: date of birth, age in months, gender, and premature birth. Note that age in months may be an estimate calculated, for example, by dividing age in days by "30.4" or "30.4375".

The multiple features include at least one of the subject's features: head circumference, anterior-posterior diameter, lateral diameter, volume ratio, brachycephalic ratio, frontal symmetry ratio, occipital symmetry ratio, CA (Cranial Asymmetry), and CVAI (Cranial Vault Asymmetry Index).

For example, it is preferable that the features extracted for the feature space have been quantified as feature amounts. In other words, it is preferable that the feature-related data J1-Jm and feature-related data JH1 from which features are extracted have been quantified. Note that the quantified feature-related data J1-Jm and feature-related data JH1 may be standardized.

Feature extraction from feature-related data J1 to Jm is used in machine learning and deep learning to extract useful information from the original data. Feature extraction is expected to reduce dimensionality, improve feature expression, simplify data processing, and improve performance. Feature extraction in machine learning and deep learning can be performed using well-known general methods, feature selection, feature transformation, etc. Furthermore, for example, feature extraction can also utilize rotation and dimensionality reduction in the feature space using principal component analysis and independent component analysis.

The feature vector calculation unit 20 expresses the feature-related data J1-Jm and the subject data D17 in feature vector format. In other words, the feature vector calculation unit 20 calculates feature vectors V1-Vm expressed in feature space from the feature-related data J1-Jm, and calculates a feature vector VH expressed in feature space from the subject data D17.

In other words, the feature vector calculation unit 20, which constitutes the subject feature vector calculation unit, calculates the feature vector VH of the feature-related data JH1 of the subject data D17 in feature space.

Furthermore, the feature vector calculation unit 20, which constitutes the result feature vector calculation unit, selects feature-related data J1 to Jm at the timing corresponding to the subject data D17 from the time series of time-series data DC1 to DCn, and calculates feature vectors V1 to Vm in feature space for this selected feature-related data J1 to Jm.

The feature vector processing unit 30, which constitutes the distance calculation unit, calculates the difference between the feature vector VH of the feature-related data JH1 of the subject data D17 and the feature vectors V1 to Vm of the feature-related data J1 to Jm of the time-series data DC1 to DCn.

The feature vector processing unit 30 calculates the difference, for example, the distance between the feature vector VH of the subject data D17 and the feature vectors V1-Vm of the feature-related data J1-Jm. In other words, the feature vector processing unit 30 calculates the distance between the feature vector VH of the subject data D17 and each of the feature vectors V1-Vm of the feature-related data J1-Jm corresponding to each of the time-series data DC1-DCn.

The correction result extraction unit 40, which constitutes the distance calculation unit, extracts feature vectors similar to the feature vector VH of the feature-related data JH1 of the subject data D17 from the feature vectors V1 to Vm of the feature-related data J1 to Jm of the time-series data DC1 to DCn.

In other words, the correction result extraction unit 40 extracts one or more feature vectors V1 to Vm of the feature-related data J1 to Jm that are closest to the feature vector VH of the subject data D17 from the calculated distances. The correction result extraction unit 40 may then narrow down the time-series data DC1 to DCn corresponding to the extracted feature vectors as similar cases to the subject data D17.

The feature vectors V1 to Vm of the feature-related data J1 to Jm that are close in distance to the feature vector VH of the subject data D17 may be selected when the ages are close in months, or may be selected when the ages are close in months even if the ages are not close. Then, the time-series data DC1 to DCn containing the selected feature-related data J1 to Jm may be considered to be similar cases corresponding to the feature-related data JH1 of the subject data D17.

The display data processing unit 50 can provide display data in a manner that allows comparison between the feature quantities of the subject data D17 and the feature quantities of the feature-related data J1-Jm of the time-series data DC1-DCn. For example, the display data processing unit 50 may display the change in monthly age for a predetermined feature quantity of the time-series data DC1-DCn, and also indicate a predetermined feature quantity of the subject data D17 in that display, thereby allowing comparison between the predetermined feature quantity of the subject data D17 and the predetermined feature quantity of the time-series data DC1-DCn.

In other words, the display data processing unit 50 performs processing to visually display the time-series data DC1 to DCn, which includes the feature-related data J1 to Jm for which distances have been calculated.

The display data processing unit 50 can output processed data that can be displayed visually from the output unit 13 to the output unit 33 of the subject-side terminal 31, the output unit 43 of the diagnosis-side terminal 41, etc. The display data processing unit 50 may also output processed data that can be displayed visually to the display unit of the output unit 13.

As a result, a display unit included in the output unit 13 or the like can display the corresponding time-series data DC1-DCn containing one or more extracted feature-related data J1-Jm.

An example of processing in the cranial correction effect estimation system 10 will be described with reference to Figures 8 to 10. The cranial correction effect estimation system 10 has a feature space processing step S180 (see Figure 8), a feature vector processing step S200 (see Figure 9), and a display data processing step S500 (see Figure 10).

First, referring to FIG. 8, the feature space processing step S180 is a step of performing processing to define a feature space based on one or more selected features selected from a plurality of features. The feature space processing step S180 is a step executed by the feature space processing unit 18. The feature space processing step S180 includes a feature extraction step (step S181 in FIG. 8), a feature-related data selection step (step S182 in FIG. 8), a preprocessing step (step S183 in FIG. 8), and a feature amount calculation step (step S184 in FIG. 8).

The feature extraction process (step S181 in FIG. 8) extracts multiple features (selected features) from the feature-related data J1 to Jm processed by the correction result data processing unit 16. As described above, known general methods, feature selection, feature transformation, etc. can be applied to feature extraction in machine learning and deep learning. Furthermore, for example, feature extraction may utilize rotation or dimensionality reduction in feature space using principal component analysis or independent component analysis. The multiple features may include attributes such as date of birth, age in months, sex, and premature birth, and features such as head circumference, anterior-posterior diameter, lateral diameter, volume ratio, brachycephalic ratio, frontal symmetry rate, occipital symmetry rate, CA, and CVAI.

The feature-related data selection process (step S182 in Figure 8) selects feature-related data J1 to Jm that need to be expressed in feature space from the time-series data DC1 to DCn.

The pre-processing step (step S183 in FIG. 8) may perform the specified processing required for expressing the feature-related data J1-Jm in feature space. For example, the feature-related data J1-Jm and feature-related data JH1 from which features are extracted may be subjected to quantification processing. The quantified feature-related data J1-Jm and feature-related data JH1 may also be subjected to standardization processing.

The feature calculation process (step S184 in Figure 8) calculates features corresponding to the data required for expression in feature space from the selected feature-related data J1 to Jm and feature-related data JH1.

Referring to FIG. 9, the feature vector processing step S200 is a step of extracting time series data DC1-DCn that is similar to feature-related data JH1 of subject data D17 from multiple past time series data DC1-DCn based on feature vectors. In other words, the feature vector processing step S200 enables the extraction of similar time series data DCS1-DCSp corresponding to similar subjects CS1-CSp (p is an integer ≦n) extracted from past subjects C1-Cn as similar cases.

The feature vector processing step S200 is a step executed by the feature vector calculation unit 20, feature vector processing unit 30, and correction result extraction unit 40.

The feature vector processing step S200 includes a subject feature vector calculation step (step S201 in Figure 9), a result feature vector calculation step (step S202 in Figure 9), a feature vector preprocessing step (step S203 in Figure 9), a feature vector comparison step (step S204 in Figure 9), and a correction result extraction step (step S205 in Figure 9).

In the subject feature vector calculation process (step S201 in Figure 9), the feature vector VH of the feature-related data JH1 of the subject data D17 is calculated. For example, the feature vector VH may be calculated based on feature amounts in which weights set individually for multiple features are reflected in the feature-related data JH1.

In the result feature vector calculation process (step S202 in FIG. 9), feature-related data J1-Jm at the timing corresponding to the subject data D17 is selected from the time-series data DC1-DCn. Also, in the result feature vector calculation process (step S202 in FIG. 9), feature vectors V1-Vm in feature space are calculated for this selected feature-related data J1-Jm. For example, feature vectors V1-Vm may be calculated based on feature amounts in which weights set individually for multiple features are reflected in the feature-related data J1-Jm.

The feature vector preprocessing step (step S203 in Figure 9) performs predetermined preprocessing on the feature vector VH of the feature-related data JH1 of the subject data D17 and the feature vectors V1 to Vm of the feature-related data J1 to Jm of the time-series data DC1 to DCn. For example, the weights set for each feature may be reflected in the feature vectors V1 to Vm and the feature vector VH.

The feature vector comparison step (step S204 in FIG. 9) calculates the difference, e.g., distance, between the feature vector VH of the feature-related data JH1 of the subject data D17 and the feature vectors V1-Vm of the feature-related data J1-Jm of the time-series data DC1-DCn. For example, the weight set for each feature may be reflected in the distance between the feature vector VH of the subject data D17 and the feature vectors V1-Vm of the feature-related data J1-Jm.

The correction result extraction process (step S205 in FIG. 9) extracts feature vectors similar to the feature vector VH of the feature-related data JH1 of the subject data D17 from the feature vectors V1-Vm of the feature-related data J1-Jm of the time-series data DC1-DCn. For example, a predetermined number of the extracted time-series data DC1-DCn may be extracted such that the distance between their corresponding feature vectors V1-Vm and the feature vector VH of the subject data D17 is short, or such that the distance falls within a predetermined range.

Furthermore, the extracted multiple time series data DC1 to DCn may be selected so that the distance between the feature vector VH of the subject data D17 and the feature vectors V1 to Vm is close when the two are of similar age in months, or may be selected so that the distance is close even if the two are not of similar age in months.

As a result, in the correction result extraction process (step S205 in Figure 9), similar time series data DCS1 to DCSp of similar subjects CS1 to CSp (p is an integer ≦ n) extracted from past subjects C1 to Cn can be extracted as similar cases.

Referring to Figure 10, the display data processing step S500 is a step of providing data that enables the subject data D17 and the extracted feature-related data J1-Jm of the time-series data DC1-DCn to be displayed in a manner that allows them to be compared. The display data processing step S500 is executed by the display data processing unit 50. The display data processing step S500 can provide data that enables the subject data D17 and the time-series data DC1-DCn to be displayed, for example, in a graph format with measured values on the vertical axis (see Figures 11-13).

In addition, the display data processing step S500 may provide data that can be displayed, for example, in table format (see Figure 14) for the subject data D17 and time-series data DC1 to DCn, or in graph format using standard deviation values (see Figures 15 to 18).

The display data processing step S500 includes a subject data display preparation step (step S510 in FIG. 10), a time-series data display preparation step (step S520 in FIG. 10), a display data generation step (step S530 in FIG. 10), and a display data output step (step S540 in FIG. 10).

In the subject data display preparation process (step S510 in Figure 10), data is prepared for the subject data D17 that can be displayed in graph format with the measured values at the measured age in months on the vertical axis.

In the time-series data display preparation process (step S520 in Figure 10), data is prepared that can display the change in measured values over time in months for multiple feature-related data J1-Jm corresponding to one or more time-series data DC1-DCn selected as similar cases in a graph format with the measured values on the vertical axis. In other words, in the extracted time-series data display preparation process, processing is performed to visually display the time-series data DC1-DCn that includes the feature-related data J1-Jm from which the distances have been calculated.

The display data generation process (step S530 in FIG. 10) generates data that displays the skull shape features of subject 100 and those of similar cases in a visually comparable manner. In the display data generation process (step S530 in FIG. 10), data is prepared that enables the subject data D17 and multiple measurement values contained in the time-series data DC1 to DCn to be displayed together in a graph showing each feature. In the display data generation process, the display range of the graph may be set to display all data, the display range of the graph may be restricted to a predetermined display range, or the display range of the graph may not be restricted at all. Furthermore, in the display data generation process, data required for a graph creation function provided in the diagnostic side terminal 41 or the like may be generated, or a graph for each feature may be generated as an image that can be displayed on the diagnostic side terminal 41 or the like.

The display data output processing step (step S540 in FIG. 10) can output data that displays the skull shape features of the subject 100 and the skull shape features of similar cases in a visually and comparative manner from the output unit 13 to the output unit 33 of the subject-side terminal 31 or the output unit 43 of the diagnosis-side terminal 41. The display data output processing step may also output the data that is displayed in a visually and comparative manner to the display unit of the output unit 13 of the main processing server 11.

With reference to Figures 11 to 13, an example of a display mode that visually and comparatively displays the skull shape features of subject 100 and those of similar cases will be described. Similar cases correspond to similar time-series data DCS1 to DCSp of similar subjects CS1 to CSp (p is an integer ≦ n) selected from past subjects C1 to Cn.

Figure 11(a) shows a graph of the anteroposterior diameter of the skull shape, showing the measured values of subject 100 and a time series of the measured values of similar subjects CS1 to CS10. More specifically, the change in anteroposterior diameter with age for each of similar subjects CS1 to CS10 is shown in line graphs P1 to P10, with subject 100's anteroposterior diameter indicated by a larger circle P0 at the age of measurement. Line graphs P1 to P10 are line graphs that connect small dots representing the change in anteroposterior diameter with age for each of similar subjects CS1 to CS10 at each measurement timing from approximately 5 to 12 months of age. In other words, by comparing the measured values of subject 100 with the time series of similar subjects CS1 to CS10, it is possible to estimate the correction effect for subject 100's anteroposterior diameter.

Figure 11(b) shows a graph of the time series of measurements of the cranial shape left-right diameter for subject 100 and similar subjects CS1 to CS10. More specifically, the change in left-right diameter with age for each similar subject CS1 to CS10 is shown in line graphs P1 to P10, and the left-right diameter for subject 100 is indicated by a large circle P0 at the measured age. This means that it is possible to estimate the correction effect for subject 100's left-right diameter.

Figure 11(c) shows a graph of the head circumference of subject 100 over time, as well as the measurements of similar subjects CS1 to CS10. More specifically, the change in head circumference over time for each of similar subjects CS1 to CS10 is shown as line graphs P1 to P10, and subject 100's head circumference is indicated by a large circle P0 at the age at which it was measured. This means that it is possible to estimate the correction effect for subject 100's head circumference.

Figure 12(a) shows a graph of the brachycephalic ratio of subject 100 over time, as well as the measurements of similar subjects CS1 to CS10. More specifically, the change in brachycephalic ratio over time for each of similar subjects CS1 to CS10 is shown in line graphs P1 to P10, and the brachycephalic ratio of subject 100 is indicated by a large circle P0 at the age of measurement. This means that it is possible to estimate the correction effect for subject 100's brachycephalic ratio.

Figure 12(b) shows a graph of the time series of measurements of the frontal symmetry rate of the skull shape for subject 100 and similar subjects CS1 to CS10. More specifically, the change in the frontal symmetry rate with age for each similar subject CS1 to CS10 is shown in line graphs P1 to P10, and the frontal symmetry rate of subject 100 is indicated by a large circle P0 at the measurement age. In other words, it is possible to estimate the correction effect for subject 100's frontal symmetry rate.

Figure 12(c) shows a graph of the time series of measurements of the occipital symmetry rate of the skull shape for subject 100 and similar subjects CS1 to CS10. More specifically, the change in the occipital symmetry rate with age for each similar subject CS1 to CS10 is shown in line graphs P1 to P10, and the occipital symmetry rate of subject 100 is indicated by a large circle P0 at the measured age. In other words, it is possible to estimate the correction effect for subject 100's occipital symmetry rate.

Figure 13(a) shows a graph of the time series of measurements of skull shape CA for subject 100 and similar subjects CS1 to CS10. More specifically, line graphs P1 to P10 show the change in CA with age for each similar subject CS1 to CS10, and subject 100's CA is indicated by a large circle P0 at the measured age. In other words, it is possible to estimate the correction effect for subject 100's CA.

Figure 13(b) shows a graph of the time series of measurements of skull shape CVAI for subject 100 and similar subjects CS1 to CS10. More specifically, line graphs P1 to P10 show the change in CVAI with age for each similar subject CS1 to CS10, and subject 100's CVAI is shown with a large circle P0 at the measured age. This means that it is possible to estimate the correction effect for subject 100's CVAI.

As a result, the cranial correction effect estimation system 10 of this embodiment can provide appropriate expectations for the results of cranial correction.

As explained above, the cranial correction effect estimation system and cranial correction effect estimation method according to this embodiment provide the following advantages.

(1) When the time-series data DC1 to DCn are visually displayed, it becomes possible to take into account the distance calculated between the feature vector VH of the subject data D17 and the feature vectors V1 to Vm of the feature-related data J1 to Jm included in the time-series data DC1 to DCn. Furthermore, rather than calculating the distance for each feature, i.e., by grouping, it becomes possible to calculate the distance as a feature vector in a feature space in which selected features are combined.

By taking into account the distance between the subject data D17 and the feature vectors V1 to Vm of the time series data DC1 to DCn, it becomes possible to compare the similarity between the subject data D17 and multiple cranial shape correction results, such as similar time series data DCS1 to DCSp, based on distance.

In other words, it will be possible to extract and display appropriate cranial shape correction results so that parents and others can know the results of their infant's cranial shape correction when making treatment decisions. This will make it possible to provide appropriate expectations for the results of cranial correction.

(2) Extraction of similar time-series data DCS1 to DCSp that are highly similar to subject data D17 is possible based on the close distance in feature space between the feature vector VH of subject data D17 and the feature-related data J1 to Jm of time-series data DC1 to DCn.

(3) Since the cranial shape correction results (time-series data DC1 to DCn) are stored for each correction case (past case), the cranial shape correction results are extracted as correction cases (cases). In other words, it becomes possible to predict the correction results based on the cases.

(4) Because the cranial shape correction results (time-series data DC1 to DCn) are stored as the result of statistical processing of several correction cases (past cases), the cranial shape correction results can be extracted as statistical results. In other words, it becomes possible to predict the correction results based on statistical results.

- Extraction of similar cases may be performed based on the distance to the feature vectors of past cases, calculated using weighting according to the contribution rate to the distribution in the feature space.

This allows similar cases to be extracted based on the distance to the feature vectors of past cases, which is calculated using weighting based on the contribution rate to the distribution in feature space.

- The feature vector calculation unit 20 (subject feature vector calculation unit or result feature vector calculation unit) may calculate feature amounts by setting separate weights for multiple features.

- The feature vector processing unit 30 (distance calculation unit) may calculate the distance between the feature vector VH of the subject data D17, which is obtained by assigning weights to each feature, and the feature vectors V1 to Vm of the feature-related data J1 to Jm.

- The correction result data processing unit 16 may, for example, perform a predetermined process such as reflecting a weight previously set for each item of the feature-related data J1 to Jm in the corresponding feature-related data J1 to Jm.

This allows weights to be applied to feature vectors, making it possible to calculate distances based on features that reflect the weights.

- The feature space processing unit 18 may cluster the feature space so as to reduce the dimension thereof below the dimension formed by the multiple features. In other words, in the feature extraction process (step S181 in FIG. 8), the feature space may be clustered so as to reduce the dimension thereof below the dimension formed by the multiple features. In this case, in the feature-related data selection process (step S182 in FIG. 8), necessary features may be selected from the feature-related data J1 to Jm so as to correspond to the feature space that has been clustered so as to reduce the dimension thereof below the dimension formed by the multiple features.

At this time, when the feature space has been reduced in dimension, the feature vector calculation unit 20 may calculate the feature vectors V1 to Vm, VH of the feature-related data J1 to Jm and feature-related data JH1 as projections onto the reduced-dimensional feature space. In other words, in the subject feature vector calculation step, when the feature space has been reduced in dimension, the feature vector VH of the feature-related data JH1 is calculated as a projection onto the reduced-dimensional feature space.

- As a new feature space, the basis and contribution rate of the feature vectors that represent the distribution may be determined using an algorithm such as PCA from the distribution of feature vectors in past cases, and the feature space may be defined based on the basis that represents the distribution. In this case, the feature vector may be projected (rotated) onto the newly defined feature space and used as the new feature vector.

This makes it possible to determine the basis and contribution rate of the feature vectors that represent the distribution from the distribution of feature vectors in past cases using algorithms such as PCA, and to define the feature space based on the basis that represents the above-mentioned distribution.

- The correction result extraction unit 40 may use a norm comparison to narrow down the extracted similar cases. For example, the correction result extraction unit 40 may calculate the average value (average feature) of each feature, calculate the L1 norm between the average feature and the feature of the measurement value obtained by 3D scanning at the age before treatment, and select the sequence with the smallest L1 norm or the sequence in ascending order to narrow down the extracted similar cases.

In other words, in the correction result extraction process (step S205 in Figure 9), norm comparison may be used to narrow down the extracted similar cases.

- Although an example has been given in which multiple feature vectors with close distances in feature space are selected, it is also possible to select one feature vector with the smallest distance.

- The time-series data DC1-DCn may be the result of processing one or more pieces of data. In other words, the storage unit 70 may store one or more pieces of time-series data DC1-DCn corresponding to statistical processing results obtained by processing one or more pieces of data based on multiple cranial shape correction results. In this case, the multiple past subjects C1-Cn and the time-series data DC1-DCn may not correspond one-to-one, or the numbers may differ. In such cases, the time-series data may be expressed as DC1-DCx (x is an integer).

- In the above embodiment, an example was given in which line graphs P1 to P10 were displayed for each of similar subjects CS1 to CS10, but this is not limiting. Average values may also be calculated from the feature amounts of the similar time-series data DCS1 to DCSp of similar subjects CS1 to CSp and displayed as numerical values. In other words, the skull shape feature amounts of subject 100 and the skull shape feature amounts of similar cases may be displayed in a tabular format so that they can be visually recognized and compared.

For example, Figure 14(a) may show the following features of subject 100: anterior-posterior diameter K11, lateral diameter K12, head circumference K13, brachycephalic ratio K14, frontal symmetry ratio K15, occipital symmetry ratio K16, CAK17, and CVAI K18. Furthermore, to facilitate comparison, age in days K20 and age in months K21 may be displayed. Furthermore, to facilitate understanding of the condition, the level K25 for the frontal symmetry ratio K15, the level K26 for the occipital symmetry ratio K16, the level K27 for CAK17, and the level K28 for CVAI K18 may be displayed as evaluations of each feature.

In addition, Figure 14(b) displays the average values and amplitudes calculated for each feature value at the time of examination for the 30 similar subjects CS1 to CSp who underwent examination for each of the following: anterior-posterior diameter K11, lateral diameter K12, head circumference K13, brachycephalic ratio K14, frontal symmetry ratio K15, occipital symmetry ratio K16, CAK17, and CVAIK18. Age in days K20 and age in months K21 may also be displayed.

Furthermore, Figure 14(c) displays the average values and amplitudes calculated for each feature of the similar time-series data DCS1 to DCSp of 30 similar subjects CS1 to CSp who have completed treatment at the start and end of correction for each of the anterior-posterior diameter K11, lateral diameter K12, head circumference K13, brachycephalic ratio K14, frontal symmetry rate K15, occipital symmetry rate K16, CAK17, and CVAIK18. Age in days K20 and age in months K21 may also be displayed. Furthermore, the average values and amplitudes of change between the start and end of correction for each of the anterior-posterior diameter K11, lateral diameter K12, head circumference K13, brachycephalic ratio K14, frontal symmetry rate K15, occipital symmetry rate K16, CAK17, and CVAIK18 may also be displayed, as well as the average values and amplitudes for the treatment period.

- In the above embodiment, line graphs P1 to P10 are used to display similar subjects CS1 to CS10, but this is not limiting. The average value and standard deviation at the scan timing of similar time-series data DCS1 to DCSp may be displayed in a graphical format that allows for visual comparison with the values at the time of measurement for subject 100. In other words, it is preferable that this graphical format also display the average value at graduation and the +1SD, -1SD, +2SD, and -2SD values for the standard deviation centered on that average value.

First, Figure 15 shows the relationship between the percentage of patients who have completed treatment and the timing of scans (number of scans). This makes it possible to estimate the number of scans that will be performed until treatment is completed.

In Figure 16(a), the head circumference of the skull shape is graphed, showing the measurements of subject 100 and the average values of the measurements of similar subjects CS1 to CSp for each scan. Specifically, the change in the average head circumference measurements of similar subjects CS1 to CSp over the number of scans is shown by line graph Q, and the head circumference of subject 100 at the first scan is indicated by a star P0. Line graph Q is a graph connecting points that show the average values for each scan timing of similar subjects CS1 to CSp for each measurement timing. Furthermore, for line graph Q, Qdp indicates a standard deviation of 1 SD, and Qdm indicates a standard deviation of -1 SD (this also applies to the graphs in Figures 16 to 18 below). Furthermore, the measurements at the end of treatment, the so-called graduation mean, are shown by graph QG, and for this graph QG, QGdp indicates a standard deviation of 1 SD, and QGdm indicates a standard deviation of -1 SD (this also applies to the graphs in Figures 16 to 18 below). In other words, by comparing the measurements of subject 100 with the time series of the average values (line graph Q) and standard deviations (Qdp, Qdm) of similar subjects CS1 to CSp, it becomes possible to estimate the correction effect on subject 100's head circumference.

In Figure 16(b), the line graph Q shows the change in the number of scans for the average brachycephalic ratio measured values of similar subjects CS1 to CSp, and the brachycephalic ratio of subject 100 at the first scan is indicated by a star P0. In other words, by comparing the measured values of subject 100 with the time series of the average values and standard deviations of similar subjects CS1 to CSp, it is possible to estimate the correction effect for subject 100's brachycephalic ratio.

In Figure 17(a), the change in the number of scans for the average measured values of the frontal symmetry rate of the skull shape for similar subjects CS1 to CSp is shown by line graph Q, and the frontal symmetry rate of subject 100 for the first scan is shown by a star P0. In other words, by comparing the measured values of subject 100 with the time series of the average values and standard deviations of similar subjects CS1 to CSp, it is possible to estimate the correction effect for subject 100's frontal symmetry rate.

In Figure 17(b), the change in the number of scans for the average measured value of the occipital symmetry rate of similar subjects CS1 to CSp for the skull shape is shown by line graph Q, and the occipital symmetry rate of subject 100 at the first scan is shown by a star P0. In other words, by comparing the measured value of subject 100 with the time series of the average value and standard deviation of similar subjects CS1 to CSp, it is possible to estimate the correction effect for subject 100's occipital symmetry rate.

In Figure 18(a), the change in the number of scans for the average CA measurement values of similar subjects CS1 to CSp for skull shape CA is shown by line graph Q, and the CA of subject 100 is indicated by a star P0 at the first scan. In other words, by comparing the measurement values of subject 100 with the time series of the average values and standard deviations of similar subjects CS1 to CSp, it is possible to estimate the correction effect for subject 100's CA.

In Figure 18(b), the change in the number of scans for the average CVAI measurement values of similar subjects CS1 to CSp for skull shape CVAI is shown by line graph Q, and the CVAI of subject 100 at the first scan is indicated by a star P0. In other words, by comparing the measurement values of subject 100 with the time series of the average values and standard deviations of similar subjects CS1 to CSp, it is possible to estimate the correction effect for subject 100's CVAI.

10 Cranial correction effect estimation system 11 Main processing server 12 Input unit 13 Output unit 14 Information processing unit 15 Subject information management unit 16 Correction result data processing unit 17 Subject data processing unit 18 Feature space processing unit 20 Feature vector calculation unit 30 Feature vector processing unit 31 Subject side terminal 32 Input unit 33 Output unit 40 Correction result extraction unit 41 Diagnosis side terminal 42 Input unit 43 Output unit 50 Display data processing unit 51 Manufacturing device 52 Input unit 53 Output unit 61 Cranial shape correction helmet 62 Outer shell 63 Interior 70 Memory unit 100 Subject 101 Cranial shape 110 Diagnologist 121 Correction shape C1 to Cn Subjects CS1 to CS10, CS1 to CSp Similar subjects D15 Attribute data D16 Correction result data D17 Subject data D20 Result data DC1 to DCn, DC1 to DCx Time series data DCS1 to DCSp Similar time series data J1 to Jm, JH1 Feature related data Jm Feature related data NW Network

Claims (8)

storing a plurality of time-series data based on a plurality of cranial shape correction results, the plurality of time-series data including, in time series, feature-related data including a plurality of features related to cranial shape;
a storage unit that stores the feature-related data of the subject as subject data;
a subject feature vector calculation unit that calculates a feature vector of the subject data in a feature space based on selected features as one or more features selected from the plurality of features;
a resultant feature vector calculation unit that selects the feature-related data at a timing corresponding to the subject data from the time series of the time-series data, and calculates a feature vector in the feature space for the selected feature-related data;
a distance calculation unit that calculates a distance between a feature vector of the subject data and a feature vector of the feature-related data;
a display data processing unit for visually displaying the time-series data including the feature-related data for which the distance has been calculated,
The plurality of features include at least one of the subject's attributes of age in months, sex, and premature birth, and the subject's feature quantities of head circumference, anterior-posterior diameter, lateral diameter, volume ratio, brachycephalic ratio, frontal symmetry rate, occipital symmetry rate, CA, and CVAI. the distance calculation unit calculates distances between the feature vector of the subject data and the feature vectors of the plurality of feature-related data, and extracts one or more feature vectors of the feature-related data that are closest to the feature vector of the subject data from among the calculated distances;
The cranial orthodontic effect estimation system according to claim 1 , wherein the display unit displays the time-series data corresponding to each of the extracted one or more feature-related data. The cranial correction effect estimation system according to claim 2 , wherein the storage unit stores a plurality of the time-series data corresponding to a plurality of cranial shape correction results. The cranial correction effect estimation system according to claim 2 , wherein the storage unit stores one or more pieces of the time-series data corresponding to one or more statistical processing results based on a plurality of cranial shape correction results. the feature space is clustered to reduce the dimension thereof to a dimension lower than that constituted by the plurality of features;
The cranial orthodontic effect estimation system according to claim 3 or 4, wherein the features are quantified as feature amounts, and the feature vector is a projection onto the reduced-dimensional feature space. 6. The cranial orthodontic effect estimation system according to claim 5, wherein the subject feature vector calculation unit calculates the feature amounts by setting weights for each of the plurality of features, or the result feature vector calculation unit calculates the feature amounts by setting weights for each of the plurality of features, or the distance calculation unit calculates the distance between the feature vector of the subject data and the feature vector of the feature-related data by weighting each feature. The weights are assigned according to the contribution rate to the distribution, or the weights are assigned according to the magnitude of the influence on the results of multiple cranial shape corrections.
The cranial correction effect estimation system according to claim 6. a storage unit stores a plurality of time-series data based on a plurality of cranial shape correction results, the time-series data including feature-related data in time series, the feature-related data including a plurality of features related to cranial shape; and the feature-related data of a subject is stored as subject data;
a step of calculating a feature vector of the subject data in a feature space based on selected features as one or more features selected from the plurality of features by a subject feature vector calculation unit;
a step of selecting, by a result feature vector calculation unit, the feature-related data at a timing corresponding to the subject data from the time series of the time-series data, and calculating a feature vector in the feature space for the selected feature-related data;
a step of calculating a distance between a feature vector of the subject data and a feature vector of the feature-related data by a distance calculation unit;
and a step of visually displaying the time-series data including the feature-related data for which the distance has been calculated by a display data processing unit,
The method for estimating the effect of cranial correction, characterized in that the plurality of features include at least one of the subject's attributes of age in months, sex, and premature birth, and the subject's features of head circumference, anterior-posterior diameter, lateral diameter, volume ratio, brachycephalic ratio, frontal symmetry rate, occipital symmetry rate, CA, and CVAI.