CN112102291A - Method for obtaining reference data of middle-of-surface defect target by anatomical feature point matching - Google Patents

Abstract

The invention discloses a method for acquiring reference data of a middle-of-plane defect target by anatomical feature point matching, which comprises the following steps: selecting normal craniofacial CT data from all CT data of an imaging department as data in a normal Chinese craniofacial three-dimensional shape database; for all models which meet the database inclusion standard and have complete craniomaxillofacial three-dimensional morphological data, manually marking 70 anatomical mark points of the craniomaxillofacial and recording the original three-dimensional coordinates Pi (xi, yi, zi); calculating the weights of all the feature points through an algorithm, and then taking the product of the corresponding coordinate distances and the weights of all the feature points as corresponding errors of the feature points; for all included patients, after their preoperative CT data were obtained, the displaced fracture block was segmented and removed, and the remaining cranial anatomical landmark points were marked. The invention improves the matching algorithm, verifies the precision of the method through defect simulation experiments so as to obtain a more accurate acquisition method of the target data in the middle of the face, expands the capacity of a database, and can be applied to clinic after verification of the precision through the simulation experiments.

Description

Method for obtaining reference data of middle-of-surface defect target by anatomical feature point matching

Technical Field

The invention relates to a method for acquiring reference data of a middle-of-face defect target, in particular to a method for acquiring reference data of a middle-of-face defect target by matching anatomical feature points.

Background

The middle part of the face mainly comprises structures such as zygomatic arches of bilateral cheekbones, eye sockets, bilateral frontal-nasal orbital sieves and the like, soft and hard tissues are complicated to dissect, bone shapes are irregular, bone bones are thin, once fracture and defect are caused by trauma, tumor and the like, accurate reduction is often difficult, and secondary face deformity and functional disorders such as diplopia, mouth opening limitation and the like are caused. The surgical treatment of the fracture and the defect in the middle of the face mainly aims at recovering the three-dimensional appearance and correcting the dysfunction, the surgical method is fracture reduction and skeleton three-dimensional appearance reconstruction, and the traditional surgical method is often unstable in effect due to weak bone segments, loss of anatomical landmark points, complex three-dimensional shape and the like, so that good facial three-dimensional appearance recovery and facial symmetry effects cannot be achieved.

Comprehensive application of digital surgical techniques, represented by digital preoperative planning and surgical navigation, has proven to significantly improve the therapeutic effect and has gradually become a common aid for maxillofacial surgery. The digital surgical technology assists the jaw face fracture and defect reconstruction, preoperative design is firstly carried out in digital surgical software, and then navigation operation is carried out, and all preoperative design preconditions are that target reference data of fracture reduction and restoration reconstruction are acquired, namely, a virtual reconstruction target of fracture reduction and restoration reconstruction is determined. For unilateral central fractures and defects, the mirror image technique is a commonly used and accepted method of target reference data acquisition. For bilateral or mid-midline fractures and defects, no mature preoperative design scheme exists at present due to the lack of healthy lateral data as a reference, and students consider that bilateral or mid-midline fractures are not suitable for navigation technology. Therefore, how to obtain the digital design target reference data by other means is a clinical problem to be solved when the mirroring technology is not available.

There are two methods for acquiring target reference data besides the mirror image technology: 1. the method comprises the steps of establishing a normal Chinese database, screening a normal human skull closest to the skull of a patient in the database by using an anatomical feature point registration method to serve as reference data, and assisting reconstruction. Marc et al refer to the remaining healthy skull and generate a normal skull by a specific computer algorithm. The first method does not aim at the fracture and defect of the middle face, the weight of the feature point data of different areas is not considered, and optimization is needed. The second method still depends on mid-facial marker points such as nasion points, bilateral ear points, bilateral zygomatic frontal suture points and the like, is not suitable for marker point deletion, and the average skull adopted is caucasian ethnic, has fewer samples and is not suitable for Chinese people.

Disclosure of Invention

The invention aims to provide a method for acquiring reference data of a middle-of-face defect target by anatomical feature point matching, which improves a matching algorithm, verifies the precision of the method through a defect simulation experiment to acquire a more accurate middle-of-face target data acquisition method, expands the capacity of a database, and verifies that the precision can be applied to clinic through the simulation experiment to solve the problems in the background technology.

In order to achieve the purpose, the invention provides the following technical scheme:

a method for acquiring reference data of a middle-of-surface defect target by matching anatomical feature points comprises the following steps:

s1: creation of a database

The method comprises the steps of adopting MySQL as a database management system, applying a Microsoft Visual Studio development tool, developing a craniomaxillofacial surgery database application system and a model registration method module, and selecting normal human craniomaxillofacial CT data from all CT data of an imaging department as data in a normal Chinese craniomaxillofacial three-dimensional form database;

s2: feature point collection

For all models which meet the database inclusion standard and have complete craniomaxillofacial three-dimensional morphological data, marking 70 anatomical marking points of the craniomaxillofacial and recording the original three-dimensional coordinates Pi (xi, yi, zi) thereof by manual operation, and the specific operation is as follows: opening a three-dimensional model in self-study software, then, righting the head, adjusting the orbital-ear plane to be parallel to the horizontal plane, adjusting the middle sagittal plane to be vertical to the screen plane, sequentially marking 70 characteristic points of the maxillofacial region according to software graphic representation, and recording the original three-dimensional coordinates Pi (xi, yi and zi) of all the characteristic points by the software;

s3: feature point matching algorithm

Calculating weights of all feature points through an algorithm, namely the importance of all feature points to different defects, and then taking the product of the corresponding coordinate distances of all feature points and the weights as the corresponding errors of the feature points, thereby obtaining a more accurate result, considering that the orbital-ear plane of a patient with fracture and defect in the middle of the cross midline often shifts or lacks, adopting an orbital-ear plane coordinate system for the normalization of normal Chinese people in a database, and matching through a rigid registration algorithm for a patient model, thereby avoiding errors generated in the normalization process;

s4: feature point matching process

After obtaining preoperative CT data of all incorporated patients, performing three-dimensional reconstruction in ProPlan CMF3.0 software, dividing and removing displaced fracture blocks, introducing the rest undelivered skull into self-study software, marking the rest skull anatomical landmark points, and matching to obtain the most similar skull as a reference for defect reconstruction and fracture reduction;

s5: simulation experiment

The method comprises the steps of simulating a normal Chinese craniomaxillofacial three-dimensional model outside a database to cause mid-midline defect, searching a skull most similar to a defect model in the database through a characteristic point matching method, performing 3D chromatographic analysis on a target area of the tested complete skull and the most similar skull in Geomagic Control 2014, and evaluating the feasibility and reliability of the method.

Further, the inclusion criteria of S1: a) chinese aged more than or equal to 16 years old; b) the CT scanning range is from supraorbital to inframandibular; c) no obvious organic lesions of bone tissues and soft tissues; d) no serious developmental deformity; e) no identifiable hard tissue surgical history on CT; f) occlusion of Anshi class; g) the tooth arrangement is basically normal; h) the number of the missing teeth is less than or equal to 2, the missing tooth area is discontinuous, and the number of the crown restorations is less than or equal to 2; exclusion criteria: i) the age is less than 16 years; j) bone tissue, soft tissue organic lesions or visible hard tissue surgical history; k) malformation of maxillofacial development; 1) malocclusion; data from DICOM (digital Imaging and Communications in medicine) of normal Chinese spiral CT meeting exclusion criteria were screened for 519 cases, where 245 males, 274 females, 41.57 + -15.00 average male ages, 37.41 + -12.92 average female ages, and no significant difference between men and women ages (P < 0.05), and were imported into ProPlan CMF3.0 (BrainLAB, Feldkirchen, Germany) to reconstruct 3D skull, with three-dimensional reconstruction parameters: the window width is 500Hu, the window level is 100Hu, the threshold value is 226-3071Hu, the cervical vertebra and data below the cervical vertebra are removed, a skull model is reconstructed, the skull model is exported into an STL format three-dimensional model file and is imported into self-research database software, and basic information such as the name, the sex, the age, the case number and the like of a patient is recorded.

Further, S3 further includes the following steps:

s31: normalization of feature point coordinates

Because the body positions of patients can not be guaranteed to be completely consistent when the spiral CT is scanned, the three-dimensional coordinates of the same surface characteristic point can change in a three-dimensional space due to head position change, in order to eliminate errors caused by the head position, the recorded three-dimensional coordinates of the characteristic point need to be normalized, namely, the original three-dimensional coordinates Pi (xi, yi, zi) are converted into standard three-dimensional coordinates P' i (xi, yi, zi) which are consistent relative to the model;

s32: determination of feature point weights

Assigning different weights to all the feature points;

s33: and realizing a matching algorithm.

Further, S33 further includes the following steps:

s331: normalization of defect models to database models

When normal Chinese feature point data is input, all feature points exist, so that 5 feature points necessary in the normalization process exist, namely a left infraorbital point (OrL), a left ear point (PoL), a right ear point (PoR), a butterfly saddle central point (S) and a nasion point (N);

s332: calculating a similarity function

After normalization processing is completed, the feature point weight values of the defect type are called, similarity functions of the defect model i and all models j in the database are respectively calculated, the similarity function values are sorted from large to small, the data base skull with the largest numerical value is the most similar to the defect model, and the software returns the skull serial number with the largest similarity function value.

Further, S5 further includes the following steps:

s51: inclusion criteria

Search 10 volunteers, 5 male and 5 female, with inclusion criteria similar to normal database inclusion criteria: a) chinese people 18-60 years old; b) no obvious organic lesions of bone tissues and soft tissues; c) no serious developmental deformity; d) no history of surgery on hard tissues of maxillofacial region; e) occlusion of Anshi class; f) the tooth arrangement is basically normal; g) no missing tooth and crown prosthesis; exclusion criteria: h) bone tissue, soft tissue organic lesions or visible hard tissue surgical history; i) malformation of maxillofacial development; j) malocclusion; the volunteers who met the above inclusion and exclusion criteria were scanned for helical CT with the scan parameters: layer thickness 1.25mm, 16 rows of spiral CT, 3D skull reconstructed by importing DICOM format CT data into ProPlan CMF3.0 (brain lab, Feldkirchen, Germany) and cervical spine image removed, parameters reconstructed: window width 500Hu, window level 100Hu, threshold 226-;

s52: constructing a defect model;

s53: model matching

Respectively marking the defect model with the residual anatomical landmark points and then matching the defect model with the model in the database by a characteristic point matching method to obtain the most similar skull;

further, S52 includes the following steps:

s521: construction of orbital cribration (NOE) defect model

In ProPlan, taking the incisional traces on the bilateral orbits, taking four points of the infraorbital hole on the bilateral orbits as vertexes, drawing a curve section with the Rounding factor being 5, the width being 140mm, the thickness being 0.01mm, drawing a coronal section through the center point of the saddle, the width being 100mm, the length being 100mm, and the thickness being 0.01mm, and dividing the skull by the two sections to obtain an NOE region and an NOE defect model;

s522: construction of bilateral zygomatic bone zygomatic arch region defect model

In ProPlan, an osteotomy plane is drawn from bilateral zygomatic maxillary suture, with width of 50mm, length of 35mm, and thickness of 0.01mm, an osteotomy plane is drawn from bilateral zygomatic pedicle, with width of 20mm, length of 20mm, and thickness of 0.01mm, an L-shaped osteotomy plane is drawn from bilateral zygomatic frontal suture, with width of 35mm, angle of 120 mm, and thickness of 0.01mm, and the skull is segmented with these three sections to obtain bilateral zygomatic arch region and bilateral zygomatic arch region defect model.

Compared with the prior art, the invention has the beneficial effects that: the invention provides a method for acquiring reference data of a facial middle defect target by anatomical feature point matching, which is characterized in that 519 cases of normal national cranio-maxillofacial three-dimensional morphological data are acquired, different regional feature point calculation weights are set aiming at a facial middle defect, a matching algorithm is improved, the precision of the method is verified through a defect simulation experiment so as to acquire a more accurate facial middle target data acquisition method.

Drawings

FIG. 1 is a schematic view of a feature point entry process of the present invention;

FIG. 2 is a schematic diagram of a normalized coordinate system of the present invention;

FIG. 3 is a schematic diagram of feature point weights according to the present invention;

FIG. 4 is a schematic representation of the acquisition of the NOE defect model of the present invention;

FIG. 5 is a schematic diagram of the acquisition of a bilateral zygomatic arch defect model of the present invention;

FIG. 6 is a diagram of a 3D chromatographic process according to the invention;

fig. 7 is a schematic diagram of 3D deviation analysis of preoperative planning and postoperative CT of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

A method for acquiring reference data of a middle-of-surface defect target by matching anatomical feature points comprises the following steps:

step 1: creation of a database

MySQL is adopted as a database management system, a Microsoft Visual Studio development tool is applied to develop a craniomaxillofacial surgery database application system and a model registration method module, and normal human craniomaxillofacial CT data are selected from all CT data of an imaging department to be used as data in a normal Chinese craniomaxillofacial three-dimensional form database.

Inclusion criteria were: a) chinese aged more than or equal to 16 years old; b) the CT scanning range is from supraorbital to inframandibular; c) no obvious organic lesions of bone tissues and soft tissues; d) no serious developmental deformity; e) no identifiable hard tissue surgical history on CT; f) occlusion of Anshi class; g) the tooth arrangement is basically normal; h) the number of the missing teeth is less than or equal to 2, the missing tooth area is discontinuous, and the number of the crown restorations is less than or equal to 2; exclusion criteria: i) the age is less than 16 years; j) bone tissue, soft tissue organic lesions or visible hard tissue surgical history; k) malformation of maxillofacial development; 1) malocclusion; data from DICOM (digital Imaging and Communications in medicine) of normal Chinese spiral CT meeting exclusion criteria were screened for 519 cases, where 245 males, 274 females, 41.57 + -15.00 average male ages, 37.41 + -12.92 average female ages, and no significant difference between men and women ages (P < 0.05), and were imported into ProPlan CMF3.0 (BrainLAB, Feldkirchen, Germany) to reconstruct 3D skull, with three-dimensional reconstruction parameters: the window width is 500Hu, the window level is 100Hu, the threshold value is 226-3071Hu, the cervical vertebra and data below the cervical vertebra are removed, a skull model is reconstructed, the skull model is exported into an STL format three-dimensional model file and is imported into self-research database software, and basic information such as the name, the sex, the age, the case number and the like of a patient is recorded.

Step 2: feature point collection

For all models which meet the database inclusion standard and have complete craniomaxillofacial three-dimensional morphological data (stl format), marking 70 anatomical mark points of the craniomaxillofacial and recording original three-dimensional coordinates Pi (xi, yi, zi) thereof by manpower, and the concrete operations are as follows: opening a three-dimensional model in self-study software, then, righting the head, adjusting the plane of the orbitals to be parallel to the horizontal plane, adjusting the plane of the orbitals to be vertical to the plane of a screen, sequentially marking 70 characteristic points of the maxillofacial area according to the software diagram, and recording the original three-dimensional coordinates Pi (xi, yi, zi) of all the characteristic points by the software (as shown in figure 1); and all the characteristic points are subjected to consistency inspection between groups and in groups, most of the characteristic points have good repeatability, the error is within 2mm, and the repeatability of part of the characteristic points is poor.

And step 3: feature point matching algorithm

Calculating the weights of all the feature points through an algorithm, namely calculating the importance of all the feature points to different defects, and then taking the product of the corresponding coordinate distances of all the feature points and the weights as the corresponding errors of the feature points, thereby obtaining a more accurate result, considering that the orbital-ear plane of a patient with mid-midline fracture and defect often shifts or lacks, adopting an orbital-ear plane coordinate system for the normalization of normal Chinese people in a database, and matching through a rigid registration algorithm for a patient model, thereby avoiding the errors generated in the normalization process, and further comprising the following steps:

3.1 normalization of feature Point coordinates

Because the body positions of patients can not be guaranteed to be completely consistent when the spiral CT is scanned, the three-dimensional coordinates of the same surface characteristic point can change in a three-dimensional space due to head position change, in order to eliminate errors caused by the head position, the recorded three-dimensional coordinates of the characteristic point need to be normalized, namely, the original three-dimensional coordinates Pi (xi, yi, zi) are converted into standard three-dimensional coordinates P' i (xi, yi, zi) which are consistent relative to the model; in the software of the present invention, all three-dimensional coordinates will be converted to standard coordinates based on the FH-S-N coordinate system (fig. 2), as follows:

after all the feature points are marked, an XOY plane is formed by a left infraorbital point OrL, left ear points PoL and right ear points PoR, a butterfly saddle point S is used as an XOY plane normal line to be a Z axis, an intersection point of the Z axis and the XOY plane is used as an origin O, the XOZ plane is obtained by passing the Z axis and a nasion point N, an intersection line of the XOY and the XOZ plane is used as an X axis, a straight line which passes the origin O and is vertical to the X axis and the Y axis is used as a Y axis, and all original three-dimensional coordinates are converted into standard three-dimensional coordinates P' i (xi, yi and zi).

3.2 determination of feature Point weights

The reasons for giving different weights to all feature points are three:

a) the influence of the same characteristic point on defects of different areas due to different areas of the same characteristic point has different sizes, for example, infraorbital points are obviously more important to the infraorbital defect than the inframandibular defect;

b) the different characteristic points have different influences on the same defect, such as different influences on the defect of the orbit cribriform area by the nasal root point and the front edge point of the large hole of the occiput;

c) the errors of manually selected points of the feature points at different parts are different, and the study of the foreigners shows that the errors of most feature points are less than 2mm, and the errors of a few feature points even can reach more than 5 mm. The larger the setpoint error, the more unstable this feature is, and the less important it should be.

Therefore, the importance of different characteristic points to the defect of different areas is determined through the algorithm, namely the weights of different characteristic points have important significance for improving the accuracy of the matching algorithm.

And randomly selecting 100 normal persons from the database as samples for calculating the weight of the feature points, wherein 50 persons are male and 50 persons are female, and processing the 100 models and the standard three-dimensional coordinates. The method comprises the following specific steps:

defining:

f_i，k: the kth feature of the model i comprises the coordinates of the feature points and the distance between any two feature points

D(f_i，k，f_j，k): k-th characteristic difference of model i and model j

Wk: weight of kth feature

The similarity values of models i and j are:

solving an equation:

when the similarity function F is equal to 1, the model i is most similar to the model j; when F is 0, the two are most dissimilar.

The algorithm continuously iterates through a computer to try each feature point weight value, so that the similarity function of the model i and the model i is maximum, the similarity function of the model i and other models is as small as possible, and the relatively optimal feature weight value is found.

Classifying the craniomaxillary facial feature points according to different regional defects, and respectively calculating the weight values of the feature points under the zygomatic arch and nasal orbital sieve defect models of bilateral zygomatic bones according to the method.

S33: and realizing a matching algorithm.

3.3.1 normalization of Defect model to database model

When normal Chinese feature point data is recorded, all feature points exist, so that 5 feature points necessary for the normalization process exist, namely a left infraorbital point (OrL), a left ear point (PoL), a right ear point (PoR), a butterfly saddle central point (S) and a nasion point (N). On the other hand, in the defect model, if any one of the 5 feature points is shifted or missing, the normalization process cannot be performed, and the matching cannot be performed.

For the model with the regional defect, the characteristic point coordinates of the model are matched with the model coordinates in the database through an iterative closest point algorithm (ICP), so that errors possibly generated in the normalization process are eliminated. The principle is (formula 1.3):

for a point set p formed by all feature points of the model i, j, the number of the feature points is m and the number of the feature points is n, a rotation matrix is defined as R, a translation matrix is defined as t, and f (R, t) represents the error between the point set p and the point set q after the point set p is subjected to matrix transformation (R, t). Then when f (R, t) is the minimum, model i matches best with model j.

And when the f (R, t) minimum value is obtained through calculation, the model i and the model j finish normalization processing.

S332: calculating a similarity function

After normalization processing is completed, the feature point weight values of the defect types are called, similarity functions (formula 1.4) of the defect model i and all models j in the database are respectively calculated, the similarity function values are sorted from large to small, and the data base skull with the largest numerical value is the data base skull most similar to the defect model. The software will return the skull number with the largest similarity function value.

And 4, step 4: feature point matching process

After obtaining preoperative CT data of all incorporated patients, performing three-dimensional reconstruction in ProPlan CMF3.0 software, dividing and removing displaced fracture blocks, guiding the remaining healthy undelivered skull into self-study software, marking remaining skull anatomical landmark points, and matching to obtain the most similar skull as a reference for defect reconstruction and fracture reduction;

and 5: simulation experiment

Simulating a normal Chinese craniomaxillofacial three-dimensional model outside a database to cause mid-midline defect, searching a skull most similar to a defect model in the database by a characteristic point matching method, performing 3D chromatographic analysis on a target region of the tested complete skull and the most similar skull in Geomicic Control 2014, and evaluating the feasibility and reliability of the method, wherein the method further comprises the following steps of:

5.1 inclusion criteria

Search 10 volunteers, 5 male and 5 female, with inclusion criteria similar to normal database inclusion criteria: a) chinese people 18-60 years old; b) no obvious organic lesions of bone tissues and soft tissues; c) no serious developmental deformity; d) no history of surgery on hard tissues of maxillofacial region; e) occlusion of Anshi class; f) the tooth arrangement is basically normal; g) no missing tooth and crown prosthesis; exclusion criteria: h) bone tissue, soft tissue organic lesions or visible hard tissue surgical history; i) malformation of maxillofacial development; j) malocclusion; the volunteers who met the above inclusion and exclusion criteria were scanned for helical CT with the scan parameters: layer thickness 1.25mm, 16 rows of spiral CT, 3D skull reconstructed by importing DICOM format CT data into ProPlan CMF3.0 (brain lab, Feldkirchen, Germany) and cervical spine image removed, parameters reconstructed: window width 500Hu, window level 100Hu, threshold 226-;

5.2 Defect model construction

5.2.1 construction of model for orbital cribration (NOE) defects

In ProPlan, using bilateral supraorbital incisions, four points of bilateral infraorbital foramen as vertexes, Rounding factor is 5, width is 140mm, and thickness is 0.01mm, curve section is drawn, and coronal section is drawn through the center point of the butterfly saddle, width is 100mm, length is 100mm, and thickness is 0.01mm, and the skull is divided by these two sections to obtain NOE region and NOE defect model (as shown in fig. 4).

5.2.2 construction of bilateral zygomatic bone zygomatic arch region defect model

In ProPlan, an osteotomy plane is drawn from bilateral zygomatic maxillary suture, with width of 50mm, length of 35mm, and thickness of 0.01mm, an osteotomy plane is drawn from bilateral zygomatic pedicle, with width of 20mm, length of 20mm, and thickness of 0.01mm, an L-shaped osteotomy plane is drawn from bilateral zygomatic frontal suture, with width of 35mm, angle of 120 mm, and thickness of 0.01mm, and the skull is divided by these three sections to obtain bilateral zygomatic arch region and bilateral zygomatic arch region defect model (see fig. 5).

5.3 model matching

And respectively matching the defect model with the model in the database after marking the residual anatomical landmark points by a characteristic point matching method to obtain the most similar skull.

5.4 evaluation method of results

And simultaneously introducing the complete skull model of the simulation defect model volunteer and the most similar skull obtained by matching into the Geomagic Control 2014 for optimal fitting alignment. A simulated defect area is selected on the skull of a volunteer, the NOE defect is a surface area from the upper incisional track of bilateral orbit to the inferior orbital foramen, the zygomatic arch defect of the bilateral zygomatic bone is a seam from the upper to the frontal suture of the bilateral zygomatic bone, a seam from the lower interior to the upper jaw suture of the bilateral zygomatic bone, and then the NOE defect is a surface area from the upper incisional track of the bilateral orbit to the inferior orbital foramen. A 3D comparison of the selected areas is then made, the type of deviation being 3D deviation, colour segment 21, maximum critical value 10mm, maximum nominal value 1mm, minimum nominal value-1 mm, minimum critical value-10 mm (see figure 6).

The following results are obtained after the experiment:

a. database feature point weight calculation result

By the method, the weight calculation of the characteristic points and the line distances is carried out on the NOE defect and the bilateral zygomatic arch defect. The larger the absolute value of the weight value is, the more important the characteristic represented by the weight value is, and the weight value is irrelevant to the positive or negative of the weight value. Wherein, the first five NOE defect weights are respectively characterized by a right zygomatic frontal suture point (FMOR), a value of-1.893350, a left supraorbital margin point (ORSL), a value of 1.711749, a right infraorbital margin point (OrR), a value of-1.381132, a right infraorbital point (ORR), a value of 1.326182, a right supraorbital margin point (ORSR), and a value of 1.248897. The characteristic points of the first five weighted bilateral zygomatic bone zygomatic arch defects are the upper edge middle point (ORSR) of the right orbit, the value is 2.090637, the lower orbital foramen point (FIOR), the value is 2.059806, the lower orbital foramen point (FioL), the value is-1.359771, the lower orbital foramen point (FIOL), the value is-1.328252, the lateral point of the right condyle process (CLR), and the value is 1.319312.

b. Simulation experiment results

b1.NOE Defect matching results

In NOE defect simulation experiments, 5 volunteers in both men and women were matched as shown in Table 1.1. And (3) performing 3D deviation analysis after the skull with the most similar matching result is optimally matched and aligned with the complete skull of the volunteer in Geomagic. The mean standard deviation of the NOE region was 1.77 + -0.38 mm, with the standard deviation of the NOE region from the reference skull in male volunteers between 1.1484 mm-2.1389 mm, and the mean of 1.65 + -0.41 mm. The NOE region of female volunteers has standard deviation of 1.4054 mm-2.3204 mm from the reference skull, and the average value is 1.89 +/-0.29 mm. There were no significant differences between gender (P ═ 0.35 > 0.05) (table 1.2).

TABLE 1.1 basic information on NOE-deficient volunteers and matched craniums

TABLE 1.2 NOE regional Standard deviation

b2. Bilateral zygomatic bone zygomatic arch defect

In the bilateral zygomatic bone and zygomatic arch defect simulation experiment, the matching results of 5 male and female volunteers are shown in the table (Table 1.3). And (3) performing 3D deviation analysis after the skull with the most similar matching result is optimally matched and aligned with the complete skull of the volunteer in Geomagic. The mean standard deviation of the zygomatic arch area of the bilateral zygomatic bones was 2.21. + -. 0.52 mm. Wherein the standard deviation of the zygomatic arch area of the zygomatic bone of the male volunteer and the reference skull is between 1.2397mm and 2.9485mm, and the average value is 2.20 +/-0.59 mm. The standard deviation of the zygomatic arch area of the cheekbone of the female volunteers and the reference skull is 1.7551 mm-2.8992 mm, and the average value is 2.21 +/-0.43 mm. There were no significant differences between gender (P ═ 0.99 > 0.05) (table 1.4).

TABLE 1.3 bilateral zygomatic arch defect volunteers and basic information of matched skull

TABLE 1.4 Standard deviations of the zygomatic arch area of bilateral cheekbones

Finally, discussing, the invention collects 519 three-dimensional shape data of normal Chinese adults, improves the matching algorithm, and from the simulation experiment result, the mean standard deviation of the method for NOE simulation defect volunteer defect area is 1.77 +/-0.38 mm, and the mean standard deviation for double-side zygomatic bone zygomatic arch simulation defect volunteer defect area is 2.21 +/-0.52 mm. This result compares to the accepted 2mm error for unilateral fractures or defects, although this approach has a 3D deviation of more than 2mm from the volunteer's original skull in the bilateral zygomatic arch area, it is clinically desirable to restore bilateral symmetry in patients with bilateral simultaneous fractures and defects, thereby reducing facial deformities rather than completely restoring to pre-injury levels. Therefore, the method can meet the requirement of preliminary clinical application.

Of course, there are still some problems with this method. Firstly, the data entry workload is huge, for each model, manual entry of 70 anatomical landmark points is required, the workload is always accompanied with the whole data expansion process and cannot be reduced, and huge time and energy are required to be consumed; secondly, the operation result of the method is basically acceptable in the NOE region, and the error within 2mm of the zygomatic arch region of the bilateral zygomatic bone is not completely achieved compared with the error in the mirror image technology.

The following description of the method for matching anatomical feature points of defect target data in the mid-midline through the clinical application example of the invention is made:

4 patients with fracture accompanied with defect on the middle part of the double sides of the mid-line are collected. 3 men and 1 woman; mean age 28.8 ± 8.8 years; the injury causes are traffic accident injuries; the average time from injury to repair is 102.0 +/-99.8 days; the related parts and the defect parts of the fracture are shown in the table 1.5, and the diagnosis time is 4 months to 4 years. All clinical patients received spiral CT examinations under the same parameters as the above database samples and obtained their DICOM format CT data.

TABLE 1.5 patient basic data

Course of treatment of cases

In 4 cases, 1 case is frontal bone defect, 2 cases are bilateral maxillary bone defect, 1 case is frontal, nasal orbital sieve, bilateral zygomatic bone zygomatic arch and skull base large-scale defect. Because the defect position and the fracture condition are different, the treatment and the repair modes are different, and the treatment and the repair modes are respectively introduced according to the defect position, and the treatment and the repair modes comprise the following steps:

(1) forehead-nose orbital sieve defect

Case 1 is a frontal-nasal orbit sieve defect patient, a 17-year-old male, a frontal large-scale defect caused by traffic accident injury before 9 months, and spiral CT examination shows that the upper boundary of the frontal bone is to the nasal bone tissue defect, the upper edges of both lateral orbits are partially defective, and the right zygomatic bone zygomatic arch is old fracture.

Preoperative design: obtaining the most similar reference data to the patient according to the characteristic point matching method: acquiring a three-dimensional model of the skull of the patient, segmenting the fracture displacement part, marking the residual characteristic points, and matching the most similar skull in a database as reference data. Importing patient DICOM format CT data into ProPlan Cmf3.0 software, performing three-dimensional reconstruction, segmenting fracture bone blocks, aligning a reference data STL model with a patient three-dimensional image coordinate plane, overlapping as much as possible, and simulating reduction of right zygomatic bone zygomatic arch fracture. After the fracture is reset, the defect range is clear, a reference model is cut along the defect edge to obtain a defect area reconstruction template, and an individualized titanium restoration body is selected as a repairing and reconstructing mode because the defect range is mainly frontal bone.

Preoperative preparation: 3D printing of the skull of the patient is completed before the operation, and the personalized titanium stent prosthesis for the forehead defect is manufactured.

The operation process comprises the following steps: under the guidance of navigation, the coronary valve is accessed, the old fracture of the zygomatic arch of the right zygomatic bone is reduced and fixed, the personalized titanium prosthesis is placed in the defect area, the position of the 3D titanium mesh is determined under the guidance of navigation, then two 5cm x 1.5cm skull external plates are placed on the upper edge of the orbit to repair the upper edge of the orbit according to the defect range of the upper edge of the orbit of the two sides, and the personalized titanium prosthesis repairs the frontal defect. The prosthesis is fixed with the zygomatic bones on both sides and the titanium mesh at the same time. After the operation, the wound surface is closed.

(2) Maxillary defect

Case 2 and case 3 are both bilateral maxillary defects, and case 2 is taken as an example. Case 2 was a 41 year old male with mid-facial fracture and bilateral maxillary defects due to car accidents before 2 months. The visual inspection of the body can be seen with double vision, left eyeball invagination, intraoral multi-tooth loss, oral-nasal cavity fistula and wearing of an autogenous cutting cannula. CT examination shows that the maxilla on both sides has large-scale defects and old fracture of the left zygomatic bone.

Preoperative design: obtaining the most similar reference data to the patient according to the characteristic point matching method: acquiring a three-dimensional model of the skull of the patient, segmenting the fracture displacement part, marking the residual characteristic points, and matching the most similar skull in a database as reference data. Importing patient DICOM format CT data into ProPlan Cmf3.0, performing three-dimensional reconstruction, segmenting fracture bone blocks, and aligning the STL model of reference data with the three-dimensional image of the patient to be overlapped as much as possible. Firstly, simulating and resetting the zygomatic arch fracture of the left zygomatic bone, and then determining the maxillary defect range. Designing a three-section type vascularized peroneal muscle skin flap to repair the maxillary bone defect according to the maxillary bone defect range: and importing the CT data of the right fibula of the patient, and designing the length and the position of the three-section fibula according to the reference model and the position and the shape of the mandible of the patient so as to fuse the three-section fibula with the alveolar process.

Preoperative preparation: 3D prints patient's skull, fibula template before the art and is convenient for the fibula moulding.

The operation process comprises the following steps: the operation adopts navigation guide operation, the half-coronary approach exposes the fracture of the zygomatic arch of the left zygomatic bone, and the zygomatic arch of the left zygomatic bone is reset and fixed under the guidance of navigation. The intraoral upper jaw vestibular sulcus incision is opened and the maxillary bone is damaged, the right fibula flap is taken, the moulding right fibula skin flap designed before the operation is referred, according to the designed position before the operation, a 2.0mm small-sized titanium plate system is used for fixing at the broken ends of the cheekbones at two sides under the guidance of navigation, and the bone grafting and the internal fixation at two sides of the piriform hole are used for recovering the nasal force column. An incision is designed 2cm below the left mandible, and the left artery and the posterior mandibular vein are separated and exposed. The peroneal muscle skin flap blood vessel is respectively anastomosed with the left artery and the left mandible posterior vein after passing through the left ascending branch inner tunnel. After the operation, the wound surface is closed. The upper jaw implant is implanted 6 months later, 2 cheekbone implants are worn, and 4 common implants are worn. The maxillary dentition repair is completed after one year.

Case 3 was also bilateral maxillary defects, accompanied by right zygomatic fracture, full-thickness upper lip defect and perioral scar. Similarly, the fibula flap is used for repairing the maxillary bone defect, and the tongue flap is used for repairing the soft tissue defect of the upper lip, so that the implantation repair is not completed.

(3) Large range defect of upper part of face

Case 4 is male, 32 years old, severe craniomaxillofacial trauma caused by traffic accident injury before 70 days, first aid in local hospitals, removal of bilateral eyeballs due to eyeball rupture, debridement of the middle and upper part of the face and treatment of craniocerebral trauma, restoration of posterior consciousness, stable general condition, and large-area bone and soft tissue defects of the middle and upper part of the face. The patient has preoperative examination of forehead collapse, bilateral eyeball loss, right eyelid edema, left eyelid loss, nose and back collapse, upper jaw bone loosening and obvious displacement, and the patient can not bite. Preoperative CT showed extensive bone and soft tissue defects of the skull base, forehead, nasal orbital sieve, right zygomatic arch, left periorbital and maxilla, the maxillary strut was completely lost, the maxilla "floated" in the middle of the face, and the right mandibular angle showed fracture line and fixed object, but the reduction fixation was not good.

Preoperative design: if the anatomical landmarks in the middle of the face of the patient are missing and shifted too much, the reference data most similar to the patient is obtained according to the characteristic point matching method: acquiring a three-dimensional model of the skull of the patient, segmenting the fracture displacement part, marking the residual characteristic points, and matching the most similar skull in a database as reference data. Importing patient DICOM format CT data into ProPlan Cmf3.0, performing three-dimensional reconstruction, segmenting fracture bone blocks, aligning a reference data STL model with a patient three-dimensional image coordinate plane, overlapping as much as possible, resetting bilateral zygomatic bone zygomatic arch and maxillary fracture according to a reference data contour and occlusion, and determining a bone defect range. According to different defect areas, the defect range is divided into 3 areas of frontal nasal orbital sieve, right zygomatic maxilla and left periorbital. Selecting different repairing modes according to different regional defect shapes and soft and hard tissue quantity: the ilium of the vascularized ilium skin flap is used for repairing the zygomatic maxilla on the right side, the personalized pre-bent titanium mesh is used for repairing the frontal-nasal orbit sieve, the ilium free bone graft is used for repairing the periorbital defect on the left side, and the skin island and the muscle carried by the ilium flap are used for repairing the soft tissue defect under the left orbit and the anterior skull.

Preoperative preparation: preoperative 3D printing of a skull model and an ilium template of a patient and pre-bending of the personalized titanium net.

The operation process comprises the following steps: since neurosurgery involves the cranial crown, no navigation is performed during the operation. During operation, the operation is performed by the scar of the trauma at the middle part of the original surface, the neurosurgeon firstly separates the cranial base dura mater and the skin subcutaneous soft tissue, and the damaged dura mater is repaired by the free skull flap. Then, the fracture of the right lower jaw corner is reduced by incision, the maxilla is loosened, and the jaw is ligated. Two-section ilium valve is prepared according to the template, one section is 4.2cm multiplied by 2.8cm for repairing the lower edge of the right orbit, and the other section is 4.4cm multiplied by 2.8cm for repairing the outer wall of the right orbit. The iliac flap soft tissue skin island is dominated by the terminal crossing branch of the deep rotary iliac artery, and a 6cm x 8cm soft tissue skin island is prepared, so as to avoid unsmooth vein reflux of the skin island and simultaneously reserve the superficial rotary iliac vein for supplying the skin island reflux. The method is characterized in that a personalized pre-bent titanium mesh is used for repairing the frontal-nasal orbital sieve defect, the ilium valve is shaped according to a template to repair the zygomatic arch defect of the right side zygomatic bone, the deep artery and vein of the rotary ilium is anastomosed with the superficial temporal artery and vein of the right side, the superficial vein of the rotary ilium is anastomosed with the superficial temporal vein of the left side, the muscle of the ilium valve is filled with the dead space below the titanium mesh, and the skin island is used for repairing the defect of the soft tissue around the left side orbit. Fixing the maxilla and the ilium valve, and finally repairing the small-range bone defect around the left orbit by using the ilium free bone grafting to complete the operation and close the wound. The patient is treated with conventional anti-inflammatory supportive therapy.

(4) Postoperative facial symmetry evaluation

Inputting a DICOM format file of postoperative CT data of a patient into iPlan CMF3.0 software, extracting a bone tissue part in the CT data by adopting a default bone tissue threshold (HU (226-3071)) to perform three-dimensional reconstruction, and taking a plane which is perpendicular to an orbital-ear plane and passes through a sphenoid saddle central point, a nasion point and an anterior nasal spine point as a median sagittal plane. And selecting the axial CT fault with the largest zygomatic arch area from the horizontal view as a zygomatic arch observation layer. Measuring the distance from the origin of the coordinate system on the fault to the most protruded point on the surface of the zygomatic bone as the zygomatic bone protrusion degree by taking the origin of the coordinate system on the fault as a central point, and calculating the difference value of the zygomatic bone protrusion degrees on two sides; and measuring the maximum distance from the midsagittal plane to the zygomatic arches on both sides, taking the maximum distance as the width of the zygomatic arch, and calculating the width difference of the zygomatic arches on both sides.

(5) Evaluation of surgical accuracy

Importing patient DICOM format postoperative CT data into ProPlan CMF3.0, removing mandible and cervical vertebra, exporting to STL model, and importing to Geomagic Control 2014 software together with preoperative design STL file. Taking preoperative design as reference, aligning postoperative skull by best fitting, and performing 3D deviation analysis, wherein the color segment 21 is 10mm in maximum critical value, 1mm in maximum nominal value, minus 1mm in minimum nominal value and minus 10mm in minimum critical value.

The result of the method for matching anatomical feature points of the defect target data in the middle of the mid-span line surface is as follows: the 4 patients smoothly complete the operation according to the preoperative operation plan, the diagnosis time is 4 months to 4 years, no double vision is caused after the operation, and the facial appearance is satisfied. Case 1 completed frontal reconstruction, the postoperative patient was satisfied with the appearance, no obvious long-term complications; case 2 the maxillary bone defect patient completed maxillary reconstruction, soft tissue malformation modification and maxillary dentition repair with implant-supported denture, and the patient was satisfied with both facial appearance and occlusion function; case 3 the maxillary bone defect patient completed maxillary bone reconstruction, and subsequently, soft tissue malformation trimming and implant-supported denture maxillary dentition repair were performed. The patient of case 4 was reviewed for 4 months after the operation to recover the basic facial appearance, the occlusion function was good, the local titanium mesh on the left orbit was exposed, about 8mm × 5mm, and then the scalp island-shaped flap was closed to expose the titanium mesh, and the wound was well healed. And subsequently, the nose appearance repair and bilateral periorbital soft tissue malformation repair are carried out. All 4 patients had at least one zygomatic fracture and underwent incisional reduction internal fixation, and the difference in zygomatic crest degree and the difference in zygomatic width were measured after the operation as shown in Table 1.6 below.

TABLE 1.6 evaluation of post-operative symmetry

The average difference value of the zygomatic.

The first 3 patients adopt the navigation operation, and most of the deviation ranges before and after the operation are less than 2 mm. The standard deviation of the zygomatic arch area of the right zygomatic bone of the 4 th patient is 3.47 mm; standard deviation of orbital sieving area 5.48 mm; maxillary standard deviation 5.91 mm. The results of the three-dimensional chromatographic analysis showed that the surgical precision was comparable in the first three patients, and the maxilla position was shifted posteriorly and inferiorly from the preoperative design in the 4 th patient, as shown in fig. 7.

The key to digital pre-surgical planning is the acquisition of reference data. In the digital surgical design of unilateral middle fracture or defect, on the premise of defaulting to the symmetry of the bilateral parts of a patient, the method for obtaining the reference data of the normal position of the affected side by using the midsagittal mirror image on the healthy side is widely applied and generally regarded as a reliable method.

According to the clinical preliminary application of the feature point matching method, although the deviation within 2mm is obtained in the NOE region, and the 3D deviation between the zygomatic arch region of the bilateral zygomatic bones and the original skull of a volunteer is 2-3 mm, for a patient with bilateral fracture and defect, the clinical hope is to recover bilateral symmetry of the patient, so that the facial deformity is reduced, and the patient does not necessarily completely recover to the level before injury. Therefore, the method can meet the requirement of preliminary clinical application.

Evaluation of surgical error: by performing 3D deviation analysis on the postoperative skull and preoperative design, the error of the operation area is generally within 2mm on the premise of navigation application. For the cases that the clinical conditions are not allowed and navigation cannot be applied, although errors can be reduced as much as possible by means of a preoperative 3D printing head die, a pre-bent titanium mesh, a guide plate applied in the operation and the like, the surgical errors are still higher than the cases applying navigation due to the fact that fracture and defect areas are large, uncertain factors in the operation are more, and the surgical errors can reach 5-6 mm. This error requires both more sophisticated preoperative preparation, such as printing of a preoperative personalized guide plate and more full consideration of the positioning of smaller fracture pieces, and more advanced intraoperative guidance procedures, providing more convenient and accurate intraoperative guidance methods in situations where existing navigation systems cannot be applied.

Evaluation of post-operative symmetry: all 5 cases basically recover normal chewing functions after the operation is completed, and for the cases involving the zygomatic arches of the two sides, the errors of the two sides are within 2mm by measuring the widths and zygomatic arches of the zygomatic arches after the operation, and the facial contours are basically symmetrical. Bilateral apophysis is related to both bony orbital volume and the volume of orbital contents, and therefore, most patients still require fat filling after completion of orbital wall reconstruction to correct orbital content deficiencies. All patients had no diplopia for 3 months after surgery.

Through preliminary clinical application, the feature point matching algorithm has feasibility, can provide reference for clinical work, and particularly can remarkably improve the accuracy of clinical operation under the participation of intraoperative navigation guidance. However, at the same time, the method still has some problems to be solved: a) 70 facial feature points are mostly displaced and lost in complex mid-facial fracture with defect cases, especially for case 4, only 24 feature points which are not displaced remain, and even if different weights are given to the feature points by the above method, the accuracy of the method is uncertain because the number of the mark points is too small; b) the data entry workload is huge, and the operation is complex.

In an attempt to solve the above problems, the dependence on limited facial feature points is reduced, a more accurate solution can still be made to complex cases, the data entry workload is reduced, and rigid registration is a feasible method for database retrieval to obtain reference data. Unlike the feature point matching method, rigid registration was originally proposed by Besl et al 1992, and the basic principle is to determine its corresponding set of near points from a set of measurement points, then calculate a new set of near points by the methods of Faugera and Hebert, and perform iterative calculations until the objective function value formed by the sum of the squares of the residuals is unchanged. By utilizing the principle to perform matching operation on coordinates of all points of the three-dimensional model, the process of manually inputting the characteristic points can be eliminated, and the input workload of the database and the input error of the characteristic points are reduced. Of course, the amount and time of calculation are also increased.

On the basis of other methods, the invention is improved as follows: the database capacity is enlarged, and the matching algorithm is improved. Regarding the data volume, 519 three-dimensional shape data of normal Chinese adults are collected at present, and along with continuous accumulation, a more complete normal Chinese craniomaxillofacial three-dimensional shape database is expected to be established.

For the matching method, the matching method by feature point matching has two advantages: a) the clinical operation habit is fit. In clinical operation, more than 2 anatomical landmark points on a bone block are in place, the bone block can be basically considered to be in place, and the main parameters influencing the face shape are only characteristic anatomical landmark points. b) The amount of calculation is reduced. Different weights are given to different feature points, so that the influence caused by feature point selection errors is reduced, and meanwhile, different weight values are given to different defect conditions so as to improve the accuracy of the method.

With the gradual expansion of the database of normal people, the feature point matching method can remarkably reduce the operation amount without sacrificing too much accuracy. And with the establishment of subsequent databases and the continuous increase of sample size, the stable and reliable matching method is continuously improved, the precision of the method is continuously improved, and the method can be expected to have good growth performance and future prospect.

From the simulation results, the mean standard deviation of the method for the defect area of NOE simulated defective volunteers is 1.77 +/-0.38 mm, and the mean standard deviation for the defect area of bilateral zygomatic arch simulated defective volunteers is 2.21 +/-0.52 mm. This result compares to the accepted 2mm error for unilateral fractures or defects, although this approach has a 3D deviation of more than 2mm from the volunteer's original skull in the bilateral zygomatic arch area, it is clinically desirable to restore bilateral symmetry in patients with bilateral simultaneous fractures and defects, thereby reducing facial deformities rather than completely restoring to pre-injury levels. Therefore, the method can meet the requirement of preliminary clinical application.

Of course, there are still some problems with this method. Firstly, the data entry workload is huge, for each model, manual entry of 70 anatomical landmark points is required, the workload is always accompanied with the whole data expansion process and cannot be reduced, and huge time and energy are required to be consumed; secondly, the operation result of the method is basically acceptable in the NOE region, and the error within 2mm of the zygomatic arch region of the bilateral zygomatic bone is not completely achieved compared with the error in the mirror image technology.

Finally, the conclusion is drawn that for the fracture and defect in the middle of the mid-midline, the method improves the characteristic point matching algorithm based on the database, expands the capacity of the database, and can be clinically applied through the verification of simulation experiments.

In summary, the method for acquiring reference data of the facial middle defect target by anatomical feature point matching provided by the invention comprises the steps of acquiring 519 three-dimensional morphological data of normal national cranio-maxillofacial, setting different regional feature point calculation weights aiming at the facial middle defect, improving a matching algorithm, verifying the precision of the method through a defect simulation experiment so as to acquire a more accurate facial middle target data acquisition method, improving a database-based feature point matching algorithm for mid-midline facial fractures and defects, expanding the database capacity, and verifying the precision through the simulation experiment to be clinically applicable.

The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be considered to be within the technical scope of the present invention, and the technical solutions and the inventive concepts thereof according to the present invention should be equivalent or changed within the scope of the present invention.

Claims (6)

1. A method for acquiring reference data of a middle defect target by matching anatomical feature points is characterized by comprising the following steps:

s1: creation of a database

The method comprises the steps of adopting MySQL as a database management system, applying a Microsoft Visual Studio development tool, developing a craniomaxillofacial surgery database application system and a model registration method module, and selecting normal human craniomaxillofacial CT data from all CT data of an imaging department as data in a normal Chinese craniomaxillofacial three-dimensional form database;

s2: feature point collection

For all models which meet the database inclusion standard and have complete craniomaxillofacial three-dimensional morphological data, marking 70 anatomical marking points of the craniomaxillofacial and recording the original three-dimensional coordinates Pi (xi, yi, zi) thereof by manual operation, and the specific operation is as follows: opening a three-dimensional model in self-study software, then, righting the head, adjusting the orbital-ear plane to be parallel to the horizontal plane, adjusting the middle sagittal plane to be vertical to the screen plane, sequentially marking 70 characteristic points of the maxillofacial region according to software graphic representation, and recording the original three-dimensional coordinates Pi (xi, yi and zi) of all the characteristic points by the software;

s3: feature point matching algorithm

Calculating weights of all feature points through an algorithm, namely the importance of all feature points to different defects, and then taking the product of the corresponding coordinate distances of all feature points and the weights as the corresponding errors of the feature points, thereby obtaining a more accurate result, considering that the orbital-ear plane of a patient with fracture and defect in the middle of the cross midline often shifts or lacks, adopting an orbital-ear plane coordinate system for the normalization of normal Chinese people in a database, and matching through a rigid registration algorithm for a patient model, thereby avoiding errors generated in the normalization process;

s4: feature point matching process

After obtaining preoperative CT data of all incorporated patients, performing three-dimensional reconstruction in ProPlan CMF3.0 software, dividing and removing displaced fracture blocks, introducing the rest undelivered skull into self-study software, marking the rest skull anatomical landmark points, and matching to obtain the most similar skull as a reference for defect reconstruction and fracture reduction;

s5: simulation experiment

The method comprises the steps of simulating a normal Chinese craniomaxillofacial three-dimensional model outside a database to cause mid-midline defect, searching a skull most similar to a defect model in the database through a characteristic point matching method, performing 3D chromatographic analysis on a target area of the tested complete skull and the most similar skull in Geomagic Control 2014, and evaluating the feasibility and reliability of the method.

2. The method for acquiring the reference data of the mid-plane defect target by matching the anatomical feature points as claimed in claim 1, wherein the inclusion criterion of S1 is as follows: a) chinese aged more than or equal to 16 years old; b) the CT scanning range is from supraorbital to inframandibular; c) no obvious organic lesions of bone tissues and soft tissues; d) no serious developmental deformity; e) no identifiable hard tissue surgical history on CT; f) occlusion of Anshi class; g) the tooth arrangement is basically normal; h) the number of the missing teeth is less than or equal to 2, the missing tooth area is discontinuous, and the number of the crown restorations is less than or equal to 2; exclusion criteria: i) the age is less than 16 years; j) bone tissue, soft tissue organic lesions or visible hard tissue surgical history; k) malformation of maxillofacial development; 1) malocclusion; data from DICOM (digital Imaging and Communications in medicine) of normal Chinese spiral CT meeting exclusion criteria were screened for 519 cases, where 245 males, 274 females, 41.57 + -15.00 average male ages, 37.41 + -12.92 average female ages, and no significant difference between men and women ages (P < 0.05), and were imported into ProPlan CMF3.0 (BrainLAB, Feldkirchen, Germany) to reconstruct 3D skull, with three-dimensional reconstruction parameters: the window width is 500Hu, the window level is 100Hu, the threshold value is 226-3071Hu, the cervical vertebra and data below the cervical vertebra are removed, a skull model is reconstructed, the skull model is exported into an STL format three-dimensional model file and is imported into self-research database software, and basic information such as the name, the sex, the age, the case number and the like of a patient is recorded.

3. The method for acquiring the reference data of the mid-plane defect target by matching the anatomical feature points as claimed in claim 1, wherein the step S3 further comprises the steps of:

s31: normalization of feature point coordinates

Because the body positions of patients can not be guaranteed to be completely consistent when the spiral CT is scanned, the three-dimensional coordinates of the same surface characteristic point can change in a three-dimensional space due to head position change, in order to eliminate errors caused by the head position, the recorded three-dimensional coordinates of the characteristic point need to be normalized, namely, the original three-dimensional coordinates Pi (xi, yi, zi) are converted into standard three-dimensional coordinates P' i (xi, yi, zi) which are consistent relative to the model;

s32: determination of feature point weights

Assigning different weights to all the feature points;

s33: and realizing a matching algorithm.

4. The method for acquiring the reference data of the mid-plane defect target by matching the anatomical feature points as claimed in claim 3, wherein S33 further comprises the following steps:

s331: normalization of defect models to database models

When normal Chinese feature point data is input, all feature points exist, so that 5 feature points necessary in the normalization process exist, namely a left infraorbital point (OrL), a left ear point (PoL), a right ear point (PoR), a butterfly saddle central point (S) and a nasion point (N);

s332: calculating a similarity function

After normalization processing is completed, the feature point weight values of the defect type are called, similarity functions of the defect model i and all models j in the database are respectively calculated, the similarity function values are sorted from large to small, the data base skull with the largest numerical value is the most similar to the defect model, and the software returns the skull serial number with the largest similarity function value.

5. The method for acquiring the reference data of the mid-plane defect target by matching the anatomical feature points as claimed in claim 3, wherein S5 further comprises the following steps:

s51: inclusion criteria

Search 10 volunteers, 5 male and 5 female, with inclusion criteria similar to normal database inclusion criteria: a) chinese people 18-60 years old; b) no obvious organic lesions of bone tissues and soft tissues; c) no serious developmental deformity; d) no history of surgery on hard tissues of maxillofacial region; e) occlusion of Anshi class; f) the tooth arrangement is basically normal; g) no missing tooth and crown prosthesis; exclusion criteria: h) bone tissue, soft tissue organic lesions or visible hard tissue surgical history; i) malformation of maxillofacial development; j) malocclusion; the volunteers who met the above inclusion and exclusion criteria were scanned for helical CT with the scan parameters: layer thickness 1.25mm, 16 rows of spiral CT, 3D skull reconstructed by importing DICOM format CT data into ProPlan CMF3.0 (brain lab, Feldkirchen, Germany) and cervical spine image removed, parameters reconstructed: window width 500Hu, window level 100Hu, threshold 226-;

s52: constructing a defect model;

s53: model matching

And respectively matching the defect model with the model in the database after marking the residual anatomical landmark points by a characteristic point matching method to obtain the most similar skull.

6. The method for acquiring the reference data of the mid-plane defect target by matching the anatomical feature points as claimed in claim 5, wherein S52 comprises the following steps:

s521: construction of orbital cribration (NOE) defect model

In ProPlan, taking the incisional traces on the bilateral orbits, taking four points of the infraorbital hole on the bilateral orbits as vertexes, drawing a curve section with the Rounding factor being 5, the width being 140mm, the thickness being 0.01mm, drawing a coronal section through the center point of the saddle, the width being 100mm, the length being 100mm, and the thickness being 0.01mm, and dividing the skull by the two sections to obtain an NOE region and an NOE defect model;

s522: construction of bilateral zygomatic bone zygomatic arch region defect model

In ProPlan, an osteotomy plane is drawn from bilateral zygomatic maxillary suture, with width of 50mm, length of 35mm, and thickness of 0.01mm, an osteotomy plane is drawn from bilateral zygomatic pedicle, with width of 20mm, length of 20mm, and thickness of 0.01mm, an L-shaped osteotomy plane is drawn from bilateral zygomatic frontal suture, with width of 35mm, angle of 120 mm, and thickness of 0.01mm, and the skull is segmented with these three sections to obtain bilateral zygomatic arch region and bilateral zygomatic arch region defect model.

Images