CN107545578A

CN107545578A - Femoral head region segmentation method, device and equipment in CT images

Info

Publication number: CN107545578A
Application number: CN201710743390.5A
Authority: CN
Inventors: 刘朋鑫; 李丙生
Original assignee: Neusoft Medical Systems Co Ltd
Current assignee: Neusoft Medical Systems Co Ltd
Priority date: 2017-08-25
Filing date: 2017-08-25
Publication date: 2018-01-05
Anticipated expiration: 2037-08-25
Also published as: CN107545578B

Abstract

The invention discloses femoral head region segmentation method, device and equipment in a kind of CT images, methods described includes obtaining subject femur head region CT sequence images；Positioned by the femur head region in every piece image to the CT sequence images, build femoral head bounding box and neck of femur bounding box；Multiple sampled points are obtained in the femoral head bounding box and the neck of femur bounding box based on space radiation collimation method；Two-dimentional sampled images are generated according to the multiple sampled point；The cortex of bone external boundary of the femur head region is determined on the two-dimentional sampled images.The present invention can realize the cortex of bone external boundary of femur head region in accurate Ground Split CT sequence images, and the implementation for schemes such as subsequent artefacts' joint replacements provides foundation.

Description

Femoral head region segmentation method, device and equipment in CT images

Technical field

The present invention relates to femoral head region segmentation method, dress in technical field of image processing, more particularly to a kind of CT images Put and equipment.

Background technology

Artificial joint replacement is one of important means of hip joint disease treatment.Before prosthetic replacement is carried out, need Select the artificial thigh bone to match with patient itself joint.Existing technical scheme is the region to femoral head to neck of femur After carrying out feature extraction, data mining is carried out according to the feature of extraction, to estimate the feature before patient femur's lesion, Jin Erxuan Go out the artificial thigh bone of matching, ensure the progress of operation.In order to extract femoral head exactly to the feature in the region of neck of femur, need The region of femoral head to neck of femur is split exactly.

The continuous development of CT technologies makes it be widely used in clinical medical many fields, occurs crowd therewith The partitioning algorithm of more CT images.And due to femoral head in hip joint sequence C T images and acetabular bone line of demarcation unobvious, common CT figures As partitioning algorithm can not exactly in segmentation figure picture femoral head to the region of neck of femur profile (i.e. cortex of bone external boundary), Thus the implementation of the schemes such as artificial joint replacement is influenceed.

The content of the invention

In view of this, on the present invention proposes in a kind of CT images that femoral head region segmentation method, device and equipment are to solve State technical problem.

In order to achieve the above object, the technical solution adopted in the present invention is：

First aspect according to embodiments of the present invention, it is proposed that femoral head region segmentation method in a kind of CT images, including：

Obtain subject femur head region CT sequence images；Wherein, the femur head region includes femoral head to neck of femur Region；

Positioned by the femur head region in every piece image to the CT sequence images, build femoral head Bounding box and neck of femur bounding box；Wherein, the femoral head bounding box is used to surround femur head model to be built, the femur Neck bounding box is used to surround neck of femur model, and the femoral head bounding box is connected with one end of neck of femur bounding box；

Multiple sampled points are obtained in the femoral head bounding box and the neck of femur bounding box based on space radiation collimation method；

Two-dimentional sampled images are generated according to the multiple sampled point；

The cortex of bone external boundary of the femur head region is determined on the two-dimentional sampled images.

Second aspect according to embodiments of the present invention, it is proposed that femoral head region segmentation device in a kind of CT images, including：

Image collection module, for obtaining subject femur head region CT sequence images；Wherein, the femur head region bag Femoral head is included to the region of neck of femur；

Bounding box builds module, for passing through the femur head region in every piece image to the CT sequence images Positioned, build femoral head bounding box and neck of femur bounding box；Wherein, the femoral head bounding box is to be built for surrounding Femur head model, the neck of femur bounding box are used to surround neck of femur model, the femoral head bounding box and neck of femur bounding box One end be connected；

Sampled point acquisition module, for being surrounded based on space radiation collimation method in the femoral head bounding box and the neck of femur Multiple sampled points are obtained in box；

Image generation module, for generating two-dimentional sampled images according to the multiple sampled point；

Border determining module, outside the cortex of bone for determining the femur head region on the two-dimentional sampled images Boundary.

The third aspect according to embodiments of the present invention, it is proposed that a kind of electronic equipment, the electronic equipment include：

Processor；

It is configured as storing the memory of processor-executable instruction；

Wherein, the processor is configured as：

Fourth aspect according to embodiments of the present invention, it is proposed that a kind of computer-readable recording medium, be stored thereon with meter Calculation machine program, the program are realized when being processed by the processor：

Compared with prior art, femoral head region segmentation method in CT images of the invention, by obtaining subject stock Bone region CT sequence images, and determined by the femur head region in every piece image to the CT sequence images Position, femoral head bounding box and neck of femur bounding box are built, based on space radiation collimation method in the femoral head bounding box and the stock Multiple sampled points are obtained in bone neck bounding box, and two-dimentional sampled images are generated according to the multiple sampled point, and then described two The cortex of bone external boundary of the femur head region is determined on dimension sampled images, it is possible to achieve stock in accurate Ground Split CT sequence images The cortex of bone external boundary of bone region, the implementation for schemes such as subsequent artefacts' joint replacements provide foundation.

Brief description of the drawings

Fig. 1 shows the flow of femoral head region segmentation method in the CT images according to one example embodiment of the present invention Figure；

Fig. 2 shows how to build femoral head bounding box and neck of femur encirclement according to one example embodiment of the present invention The schematic flow sheet of box；

Fig. 3 shows the schematic flow sheet for how obtaining multiple sampled points according to one example embodiment of the present invention；

Fig. 4 shows the cortex of bone external boundary for how determining femur head region according to one example embodiment of the present invention Schematic flow sheet；

Fig. 5 shows the access path for how obtaining total connection Least-cost according to one example embodiment of the present invention Schematic flow sheet；

Fig. 6 shows the structure of femoral head region segmentation device in the CT images according to one example embodiment of the present invention Block diagram；

Fig. 7 shows the knot of femoral head region segmentation device in CT images in accordance with an alternative illustrative embodiment of the present invention Structure block diagram；

Fig. 8 shows the structured flowchart of the electronic equipment according to one example embodiment of the present invention；

Fig. 9 shows the stream of femoral head region segmentation method in CT images in accordance with an alternative illustrative embodiment of the present invention Cheng Tu；

Figure 10 A show the schematic diagram of the femur head region according to one example embodiment of the present invention；

Figure 10 B show that rectangular coordinate system builds schematic diagram in the CT images according to one example embodiment of the present invention；

Figure 11 shows the knot of the femoral head bounding box and neck of femur bounding box according to one example embodiment of the present invention Structure schematic diagram；

Figure 12 A show the structural representation of the triangular plate according to one example embodiment of the present invention；

Figure 12 B are shown to be shown according to the femur head model of one example embodiment of the present invention and the structure of neck of femur model It is intended to；

Figure 12 C show the structural representation of the femur head model before the optimization according to one example embodiment of the present invention Figure；

Figure 12 D show the structural representation of the femur head model after the optimization according to one example embodiment of the present invention Figure；

Figure 13 A to Figure 13 C show the schematic diagram of the two-dimentional sampled images according to one example embodiment of the present invention；

Figure 14 shows multiple important parameters in being performed the operation according to the femoral head replacement of one example embodiment of the present invention Schematic diagram；

Figure 15 is shown outside the cortex of bone for how determining femur head region according to one example embodiment of the present invention The flow chart on boundary.

Embodiment

Below with reference to embodiment shown in the drawings, the present invention will be described in detail.But these embodiments are simultaneously The present invention is not limited, structure that one of ordinary skill in the art is made according to these embodiments, method or functionally Conversion is all contained in protection scope of the present invention.

It is only merely for the purpose of description specific embodiment in terminology used in the present invention, and is not intended to be limiting the present invention. It is also intended in " one kind " of the singulative of the invention with used in appended claims, " described " and "the" including majority Form, unless context clearly shows that other implications.It is also understood that term "and/or" used herein refers to and wrapped Containing the associated list items purpose of one or more, any or all may be combined.

It will be appreciated that though various structures, but these structures may be described using term first, second etc. in the present invention It should not necessarily be limited by these terms.These terms are only used for same type of structure being distinguished from each other out.

Fig. 1 shows the flow of femoral head region segmentation method in the CT images according to one example embodiment of the present invention Figure；As shown in figure 1, this method comprises the following steps S101-S105：

S101：Obtain subject femur head region CT sequence images.

In one embodiment, the subject gathered by equipment such as Medical CT machines (computed tomography machine) is obtained Femur head region CT sequence images.

Wherein, femur head region includes femoral head to the region of neck of femur.

S102：Positioned, built by the femur head region in every piece image to the CT sequence images Femoral head bounding box and neck of femur bounding box.

In one embodiment, the femur head region in every piece image of CT sequence images is positioned, obtains femur The positioning result of head region, and femoral head bounding box and neck of femur bounding box are built according to the positioning result.

Wherein, femoral head bounding box is used to surround femur head model to be built, and neck of femur bounding box is used to surround femur Eck mould type.

In one embodiment, because femoral head is connected with neck of femur, thus the femoral head bounding box and neck of femur bag built The one end for enclosing box is also connected.

S103：Multiple adopt is obtained in the femoral head bounding box and the neck of femur bounding box based on space radiation collimation method Sampling point；

In one embodiment, obtained respectively in femoral head bounding box and neck of femur bounding box based on space radiation collimation method Multiple sampled points, because femoral head bounding box and neck of femur bounding box surround femur head model and neck of femur model respectively, thus Sampled point on surface of the multiple sampled points obtained necessarily including femur head model and neck of femur model, i.e. femur head region Sampled point corresponding to cortex of bone external boundary.

S104：Two-dimentional sampled images are generated according to the multiple sampled point；

In one embodiment, according to multiple sampled points of acquisition, arranged according to corresponding acquisition order, to generate two Tie up sampled images.

S105：The cortex of bone external boundary of the femur head region is determined on the two-dimentional sampled images.

In one embodiment, due to including multiple sampled points of above-mentioned collection on two-dimentional sampled images, i.e., comprising femoral head Sampled point corresponding to the cortex of bone external boundary in region, thus pre-set image processing method (such as critical path method (CPM)) can be used, The cortex of bone external boundary of femur head region is determined on the two-dimentional sampled images.

Femoral head region segmentation method in the CT images of this implementation, by obtaining subject femur head region CT sequence charts Picture, and positioned by the femur head region in every piece image to the CT sequence images, build femoral head bag Box and neck of femur bounding box are enclosed, is obtained based on space radiation collimation method in the femoral head bounding box and the neck of femur bounding box Multiple sampled points, and two-dimentional sampled images are generated according to the multiple sampled point, and then determined on the two-dimentional sampled images The cortex of bone external boundary of the femur head region, it is possible to achieve the cortex of bone of femur head region in accurate Ground Split CT sequence images External boundary, the implementation for schemes such as subsequent artefacts' joint replacements provide foundation.

Fig. 2 shows how to build femoral head bounding box and neck of femur encirclement according to one example embodiment of the present invention The schematic flow sheet of box；The present embodiment is on the basis of embodiment illustrated in fig. 1, how to build femoral head bounding box and neck of femur It is illustrative exemplified by bounding box.As shown in Fig. 2 pass through each width to the CT sequence images in above-mentioned steps S102 The femur head region in image is positioned, and is built femoral head bounding box and neck of femur bounding box, can be included following step Rapid S201-S204：

S201：The femur head region is positioned in every piece image, it is described per piece image to determine In the central point of femoral head, the center line of radius and neck of femur, radius.

Because unique similar round is presented in femoral head in CT images, thus in one embodiment, similar round can be based on Recognizer positions to the femur head region (i.e. femoral head and neck of femur) in every piece image, to determine each width respectively The central point of femoral head in image, the center line of radius and neck of femur, radius.

S202：The central point of femoral head in every piece image, the center line of radius and neck of femur, half Footpath, estimate center line, the radius of the central point of the femur head model, radius and the neck of femur model.

In one embodiment, the central point of the femoral head in every piece image of determination, radius and neck of femur Center line, radius, the central point of femur head model, the center line of radius and neck of femur model, radius can be estimated, Determine the central point of femur head model, the center line of radius and neck of femur model, the substantially numerical value of radius.

S203：According to central point, the radius of the femur head model determine the femoral head bounding box central point, half Footpath, and the femoral head bounding box is built according to the central point of the femoral head bounding box, radius；Wherein, the femoral head bag The central point for enclosing box overlaps with the central point of the femur head model, and the radius of the femoral head bounding box is more than the femoral head The radius of model.

, can be according to stock because the femoral head bounding box of structure is used to surround femur head model, thus in one embodiment Central point, the radius of bone model determine the central point of femoral head bounding box, radius.

In one embodiment, the central point of femoral head bounding box can be made to be overlapped with the central point of femur head model, femur The radius of head bounding box is more than the radius of femur head model, such as can make the radius of femoral head bounding box for femur head model 2 times of radius, and then femoral head bounding box can be built according to central point, the radius of femoral head bounding box.

S204：According to center line, the radius of the neck of femur model determine the neck of femur bounding box center line, half Footpath, and the neck of femur bounding box is built according to the center line of the neck of femur bounding box, radius；Wherein, the neck of femur bag The center line for enclosing box overlaps with the center line of the neck of femur model, and the radius of the neck of femur bounding box is more than the neck of femur The radius of model.

, can be according to stock because the neck of femur bounding box of structure is used to surround neck of femur model, thus in one embodiment Center line, the radius of bone eck mould type determine the center line of neck of femur bounding box, radius.

In one embodiment, the center line of neck of femur bounding box can be made to be overlapped with the center line of neck of femur model, femur The radius of neck bounding box is more than the radius of neck of femur model, such as can make the radius and femoral head bounding box of neck of femur bounding box Radius it is identical.

As shown from the above technical solution, the present embodiment realizes each width to CT sequence images by step S201-S204 The femur head region in image is positioned, to build femoral head bounding box and neck of femur bounding box according to positioning result, Femoral head bounding box and neck of femur bounding box can be built exactly, be advantageous to improve femur head region in follow-up progress CT images The accuracy of segmentation.

Fig. 3 shows the schematic flow sheet for how obtaining multiple sampled points according to one example embodiment of the present invention； The present embodiment is illustrative exemplified by how obtaining multiple sampled points on the basis of embodiment illustrated in fig. 2.Such as Fig. 3 It is shown, obtained in step S103 in the femoral head bounding box and the neck of femur bounding box based on space radiation collimation method multiple Sampled point, it may comprise steps of S301-S303：

S301：On the center line of the neck of femur bounding box, and the femoral head that is located at of the center line surrounds On extended line inside box, multiple points are determined；

In one embodiment, in the center line and the extended line of the center line of neck of femur bounding box, (extended line is located at Inside bounding box) on, multiple points are determined with default sampling step length.

S302：On the cross section belonging to every bit in the multiple point, the point centered on the every bit, interval is in advance If angle draws a plurality of radiation to space；Wherein, center line and institute of the cross section perpendicular to the neck of femur bounding box State extended line.

In one embodiment, for the every bit in above-mentioned multiple points, surrounded in the presence of one perpendicular to neck of femur The center line of box and the cross section of extended line.Point is spaced predetermined angle on the cross section corresponding to it centered on every bit A plurality of radiation is drawn to space.Such as 2 ° of extraction radiation are spaced on each cross section, then 180 radiation can be drawn altogether Line.

S303：According to the lead direction of every radiation in a plurality of radiation, gather successively on a plurality of radiation, position Multiple sampled points inside the femoral head bounding box and neck of femur bounding box.

In one embodiment, according to the lead direction of every radiation in a plurality of radiation (i.e. by central point pointing space Direction), gathered successively on a plurality of radiation, in femoral head bounding box and neck of femur bounding box according to default sampling step length Each sampled point in portion.

As shown from the above technical solution, realized by step S301-S303 based on space radiation collimation method in the femoral head Multiple sampled points are obtained in bounding box and the neck of femur bounding box, multiple sampled points can be obtained exactly, be advantageous to improve The follow-up accuracy for carrying out femoral head region segmentation in CT images.

On this basis, two-dimentional sampled images are generated according to the multiple sampled point in step S104, can included：

Multiple sampled points are arranged in order according to the lead direction and extraction order of corresponding radiation, to generate State two-dimentional sampled images.

In one embodiment, can by the multiple sampled point, corresponding to same radiation sampled point according to corresponding The lead direction of radiation is arranged in order from left to right；And the extraction order by the sampled point of arrangement according to corresponding radiation It is arranged in order from the top down, to generate two-dimentional sampled images.

Fig. 4 shows the cortex of bone external boundary for how determining femur head region according to one example embodiment of the present invention Schematic flow sheet；The present embodiment is on the basis of embodiment illustrated in fig. 1, how to determine outside the cortex of bone of femur head region It is illustrative exemplified by boundary.As shown in figure 4, the femur Head Section is determined in step S105 on the two-dimentional sampled images The cortex of bone external boundary in domain, may comprise steps of S401-S402：

S401：According to the connection cost of each sampled point and neighbor point on the two-dimentional sampled images, total connection cost is obtained Minimum access path.

In one embodiment, on 2d, from top to bottom determine respectively on each sampled point and preset direction (such as it is left Under, just under, bottom right) neighbor point connection cost, and then determine the total connection cost of each bar by the access path under upper, with It is determined that always connect the access path of Least-cost.

In one embodiment, shown in the calculation formula such as following formula (1) of the connection cost：

Cost=w₁*A+w₂*B+w₃*C (1)

Wherein, Cost is the cost that 2 points (i.e. sampled point and a neighbor point) is connected, and A is 2 points of laterally two neighborhood gray scale differences Sum of value, B are the difference of 2 gradients, and C is 2 points of distance, w₁、w₂And w₃Respectively A, B and C weight.

S402：The cortex of bone external boundary is determined according to the acquisition result of the access path.

In one embodiment, however, it is determined that an access path under always connecting cost most, then can be according on the access path Each sampled point determine the cortex of bone external boundary of femur head region.

As shown from the above technical solution, by obtaining the access path of total connection Least-cost, and then according to the connection The acquisition result in path determines the cortex of bone external boundary, it is possible to achieve outside the cortex of bone for accurately determining femur head region Boundary, be advantageous to improve the follow-up accuracy for carrying out femoral head region segmentation in CT images.

Fig. 5 shows the access path for how obtaining total connection Least-cost according to one example embodiment of the present invention Schematic flow sheet；The present embodiment is on the basis of embodiment illustrated in fig. 4, how to obtain the link road of total connection Least-cost It is illustrative exemplified by footpath.As shown in figure 5, in step S401 according to each sampled point on the two-dimentional sampled images with it is neighbouring The connection cost of point, the access path of total connection Least-cost is obtained, may comprise steps of S501-S504：

S501：Initialization one is used for the priority query for storing the sampled point；Wherein, respectively adopted in the priority query The priority of sampling point is determined by the size of corresponding connection cost；

In one embodiment, the priority query for storing sampled point is initialized, and is the two-dimentional sampled images Each sampled point assigns a mark；Wherein, the species of the mark includes：Frozen (point of initialization), Faraway are (also The point not being judged), Narrow (point of the existing cost value judged)；

In one embodiment, the priority of each sampled point cost value according to corresponding to the sampled point in the priority query Size determine that the smaller then priority of cost value is higher.

For example, in the first row that the two-dimentional sampled images are stored in initial phase, the priority query All sampled points, and assign " Frozen " mark for the point of the first row in the two-dimentional sampled images, assigned for remaining point " Faraway " is identified.

S502：The maximum sampled point of priority is taken out from the priority query, calculate the sampled point of taking-up with it is multiple The cost value of neighbor point, assign the cost value of calculating to corresponding neighbor point respectively；

In one embodiment, judge whether current queue is empty：If so, terminate to calculate；If it is not, it is maximum then to eject priority Sampled point, calculate in the sampled point and the two-dimentional sampled images (lower-left, just lower and bottom right) three neighbor points respectively Cost value, and assign the cost value of calculating to corresponding neighbor point respectively；

In one embodiment, first sampled point ejected in queue is gray scale in the first row of the two-dimentional sampled images The maximum sampled point of Grad.

On this basis, the access path of the sampled point and the neighbor point of cost value minimum of ejection is recorded, and judges cost Whether the minimum neighbor point of value is located at image last column：If so, terminate to calculate.

S503：The species of the minimum neighbor point of the cost value is judged, according to the renewal of the judged result of the species The cost value of the minimum neighbor point of cost value, and the minimum neighbor point of the cost value is added into the priority query；

In one embodiment, the mark of the minimum neighbor point of the cost value is judged：

If the neighbor point is identified as Faraway, the cost value for assigning the neighbor point is：The sampled point of the ejection is worked as The mark of the neighbor point is set to Narrow by preceding cost value plus the sampled point of the ejection and the connection cost of the neighbor point, The neighbor point is added into the queue again；

If the mark of the neighbor point is not Faraway (being Narrow), the generation that the neighbor point currently calculates is judged Whether value is less than the existing cost of the neighbor point：

If so, the existing cost for then changing the neighbor point is the cost currently calculated of the neighbor point, and this is neighbouring Point adds the queue.

S504：Repeat described taken out from the priority query described in the maximum sampled point of priority to described incite somebody to action The step of minimum neighbor point of cost value adds the priority query, until meeting preset termination condition, by current cost It is worth access path of the access path corresponding to minimum sampled point as total connection Least-cost.

In one embodiment, when meeting preset termination condition, terminate to calculate, and the minimum sampled point of cost value is corresponding Access path be defined as the shortest path.

As shown from the above technical solution, the access path of total connection Least-cost is obtained by step S501-S504, can be with The access path of total connection Least-cost is obtained exactly, and then can be accurately determined cortex of bone external boundary accordingly, is advantageous to Improve the follow-up accuracy for carrying out femoral head region segmentation in CT images.

On the basis of the various embodiments described above, the femur Head Section is determined in step S105 on the two-dimentional sampled images After the cortex of bone external boundary in domain, this method can also include：

Pair determine the cortex of bone external boundary optimize.

Due to inner side and the acetabular bone of femoral head lean on it is very near, and because by CT machines, precision itself is influenceed, stock Border more unobvious between the inner side of bone and acetabular bone, the femur can not be accurately determined with above-mentioned shortest-path method Head and neck of femur external boundary, thus need to be optimized.

In one embodiment, the above-mentioned process optimized to cortex of bone external boundary may comprise steps of：

S601：The femoral head model vertices are extracted to song between the edge of neck of femur model, remote femur head model (warp, the direction of latitude two) curve that rate changes greatly, adjusts the bent curvature of a curve so that and its Curvature varying is uniformly gentle, and accordingly Ground adjusts the adjacent bent curvature of a curve of the curve；

S602：Determine that the larger curve of Curvature varying corresponds to the position range of acetabular bone, in the external boundary that currently determines and right One Local Extremum is set between the acetabular bone answered, shortest path side is based on according to the Local Extremum again in the subrange The external boundary point that method calculates, such as the external boundary for providing newly to determine must be between the extreme point and acetabular bones, and other are not corresponding The boundary point of extreme point take advantage of a situation the new boundary point of connection so that the Curvature varying of the larger curve of the Curvature varying reduces.

As shown from the above technical solution, cortex of bone external boundary is optimized by step S601-S602, CT can be overcome The influence of machine precision itself, improve the accuracy for the cortex of bone external boundary for determining femur head region.

Fig. 6 shows the structure of femoral head region segmentation device in the CT images according to one example embodiment of the present invention Block diagram；As shown in fig. 6, the device include image collection module 610, bounding box structure module 620, sampled point acquisition module 630, Image generation module 640 and border determining module 650, wherein：

Image collection module 610, for obtaining subject femur head region CT sequence images；Wherein, the femur Head Section Domain includes femoral head to the region of neck of femur；

Bounding box builds module 620, for passing through the femoral head in every piece image to the CT sequence images Region is positioned, and builds femoral head bounding box and neck of femur bounding box；Wherein, the femoral head bounding box treats structure for encirclement The femur head model built, the neck of femur bounding box are used to surround neck of femur model, the femoral head bounding box and neck of femur bag The one end for enclosing box is connected；

Sampled point acquisition module 630, for based on space radiation collimation method in the femoral head bounding box and the neck of femur Multiple sampled points are obtained in bounding box；

Image generation module 640, for generating two-dimentional sampled images according to the multiple sampled point；

Border determining module 650, outside the cortex of bone for determining the femur head region on the two-dimentional sampled images Border.

As shown from the above technical solution, the present embodiment is by obtaining subject femur head region CT sequence images, and passes through The femur head region in every piece image of the CT sequence images is positioned, builds femoral head bounding box and femur Neck bounding box, multiple samplings are obtained in the femoral head bounding box and the neck of femur bounding box based on space radiation collimation method Point, and two-dimentional sampled images are generated according to the multiple sampled point, and then determine the femur on the two-dimentional sampled images The cortex of bone external boundary of head region, it is possible to achieve the cortex of bone external boundary of femur head region in accurate Ground Split CT sequence images, Implementation for schemes such as subsequent artefacts' joint replacements provides foundation.

Fig. 7 shows the knot of femoral head region segmentation device in CT images in accordance with an alternative illustrative embodiment of the present invention Structure block diagram；Wherein, image collection module 710, bounding box structure module 720, sampled point acquisition module 730, image generation module 740 and border determining module 750 and the image collection module 610 in embodiment illustrated in fig. 6, bounding box structure module 620, adopt The function phase of sampling point acquisition module 630, image generation module 640 and border determining module 650 is same, herein without repeating. As shown in fig. 7, bounding box builds module 720, can include：

Zone location unit 721, for being positioned in every piece image to the femur head region, to determine The central point of femoral head, the center line of radius and neck of femur, radius in every piece image；

Parameter estimation unit 722, central point, radius and femur for the femoral head in every piece image Center line, the radius of neck, estimate the central point of the femur head model, the center line of radius and the neck of femur model, half Footpath；

First construction unit 723, for determining that the femoral head surrounds according to central point, the radius of the femur head model Central point, the radius of box, and the femoral head bounding box is built according to the central point of the femoral head bounding box, radius；Wherein, The central point of the femoral head bounding box overlaps with the central point of the femur head model, and the radius of the femoral head bounding box is big In the radius of the femur head model；

Second construction unit 724, for determining that the neck of femur surrounds according to center line, the radius of the neck of femur model Center line, the radius of box, and the neck of femur bounding box is built according to the center line of the neck of femur bounding box, radius；Wherein, The center line of the neck of femur bounding box overlaps with the center line of the neck of femur model, and the radius of the neck of femur bounding box is big In the radius of the neck of femur model.

In one embodiment, sampled point acquisition module 730, can include：

First determining unit 731, for the center line in the neck of femur bounding box, and the center line is located at On extended line inside the femoral head bounding box, multiple points are determined；

Radiation draws unit 732, for the cross section belonging to the every bit in the multiple point, with described each Point centered on point, interval predetermined angle draw a plurality of radiation to space；Wherein, the cross section is perpendicular to the neck of femur bag Enclose the center line of box and the extended line；

Sampled point collecting unit 733, for the lead direction according to every radiation in a plurality of radiation, adopt successively Collect multiple sampled points on a plurality of radiation, inside the femoral head bounding box and neck of femur bounding box.

In one embodiment, image generation module 740, can include：

Image generation unit 741, for by lead direction of the multiple sampled point according to corresponding radiation and extraction Order is arranged in order, to generate the two-dimentional sampled images.

In one embodiment, border determining module 750, can include：

Path acquiring unit 751, for the connection cost according to each sampled point and neighbor point on the two-dimentional sampled images, Obtain the access path of total connection Least-cost；

Border acquiring unit 752, for determining the cortex of bone external boundary according to the acquisition result of the access path.

Cost=w₁*A+w₂*B+w₃*C (1)

Wherein, Cost is the cost of 2 points of connections, and A is the sum of 2 points of horizontal two neighborhood gray scale difference values, and B is 2 gradients Difference, C are 2 points of distance, w₁、w₂And w₃Respectively A, B and C weight.

In one embodiment, path acquiring unit 751 can be also used for：

Initialization one is used for the priority query for storing the sampled point；Wherein, each sampled point in the priority query Priority by it is corresponding connection cost size determine；

The maximum sampled point of priority is taken out from the priority query, calculates the sampled point of taking-up and multiple neighbor points Cost value, assign the cost value of calculating to corresponding neighbor point respectively；

Judge the species of the minimum neighbor point of the cost value, the cost value is updated according to the judged result of the species The cost value of minimum neighbor point, and the minimum neighbor point of the cost value is added into the priority query；

The sampled point that priority maximum is taken out from the priority query is repeated to described by the cost The step of minimum neighbor point of value adds the priority query, until meeting preset termination condition, by current cost value most Access path of the access path corresponding to small sampled point as total connection Least-cost.

In one embodiment, the device can also include：

Border optimization module 760, optimized for pair cortex of bone external boundary determined.

For device embodiment, because it corresponds essentially to embodiment of the method, so related part is real referring to method Apply the part explanation of example.Device embodiment described above is only schematical, wherein described be used as separating component The unit of explanation can be or may not be physically separate, can be as the part that unit is shown or can also It is not physical location, you can with positioned at a place, or can also be distributed on multiple NEs.Can be according to reality Need to select some or all of module therein to realize the purpose of the present invention program.Those of ordinary skill in the art are not paying In the case of going out creative work, you can to understand and implement.

The embodiment of femoral head region segmentation device can be applied on network devices in CT images of the present invention.Device is implemented Example can be realized by software, can also be realized by way of hardware or software and hardware combining.Exemplified by implemented in software, as Device on one logical meaning, it is by corresponding computer journey in nonvolatile memory by the processor of equipment where it Sequence instruction reads what operation in internal memory was formed.For hardware view, as shown in figure 8, being femur in the CT images of the present invention A kind of hardware structure diagram of electronic equipment where head region segmenting device, except the processor shown in Fig. 8, network interface, internal memory And outside nonvolatile memory, the equipment in embodiment where device can also generally include other hardware, such as it is responsible for place Manage forwarding chip of message etc.；The equipment is also possible to be distributed equipment from hardware configuration, may include multiple Interface card, to carry out the extension of Message processing in hardware view.

The embodiment of the present invention additionally provides a kind of computer-readable recording medium, is stored thereon with computer program, the journey Following task processing method is realized when sequence is processed by the processor：

The cortex of bone external boundary of the femur head region is determined on the two-dimentional sampled images.Below with a specific implementation Example illustrates the present invention, but is not used in and limits the scope of the invention.

Fig. 9 shows the stream of femoral head region segmentation method in CT images in accordance with an alternative illustrative embodiment of the present invention Cheng Tu.Figure 10 A show the schematic diagram of the femur head region according to one example embodiment of the present invention；Figure 10 B show root Schematic diagram is built according to rectangular coordinate system in the CT images of one example embodiment of the present invention；Figure 11 is shown according to the present invention's The femoral head bounding box of one exemplary embodiment and the structural representation of neck of femur bounding box；Figure 12 A are shown according to the present invention An exemplary embodiment triangular plate structural representation；Figure 12 B are shown according to one example embodiment of the present invention The structural representation of femur head model and neck of femur model.

As shown in figure 9, femoral head region segmentation method in the CT images of the present embodiment, comprises the following steps：

S901：The subject near end of thighbone CT sequence images that Medical CT machine is gathered are obtained, to every in the CT sequence images Femur head region (including femoral head to the region of neck of femur, referring to the a-quadrant shown in Figure 10 A) in piece image is determined Position, the central point of femoral head, the center line of radius and the neck of femur, radius in determining respectively per piece image, and then determine Center line, the radius of the central point of femur head model to be built, radius and neck of femur model to be built；

In one embodiment, because unique similar round is presented in femoral head in CT images, thus similar round can be based on Recognizer positions to the femoral head in the CT images；Similarly, because rectangle (cylinder is presented in neck of femur in CT images Section), thus neck of femur can be positioned based on corresponding image recognition algorithm.On this basis, according to each width In image the central point of femoral head, the center line of radius and neck of femur, radius determine the central point of femur head model, radius with And center line, the radius of neck of femur model.In one embodiment, the central point of femur head model can be located at neck of femur model On the extended line of center line.

S902：According to the central point of femur head model, radius structure femoral head bounding box, and according in neck of femur model Heart point, radius structure neck of femur bounding box (referring to Figure 11)；Wherein, one end phase of femoral head bounding box and neck of femur bounding box Even.

In one embodiment, the central point of femoral head bounding box overlaps with the central point of femur head model, and femoral head surrounds The radius of box is more than the radius (such as can be 2 times of the radius of femoral head model) of femur head model, so that femoral head surrounds Femur head model is enclosed in inside by box；The center line of neck of femur bounding box overlaps with the center line of neck of femur model, neck of femur The radius of bounding box is more than radius (the follow-up radius for calculating, can setting neck of femur bounding box for convenience of neck of femur model It is identical with the radius of femoral head bounding box) so that neck of femur model is enclosed in inside by neck of femur bounding box.

S903：Multiple adopt is obtained in the femoral head bounding box and the neck of femur bounding box based on space radiation collimation method Sampling point；

In one embodiment, using the center line of neck of femur bounding box as z-axis, using the central point of femoral head bounding box as origin Build xyz rectangular coordinate systems (being different from the human body coordinate system XYZ shown in Figure 10 B lower lefts)；In z-axis, walked with default sampling Grow it is on the summit of femoral head to neck of femur, away from femoral head summit edge between determine multiple points, respectively in the plurality of point It is where middle every bit, parallel in the plane of x/y plane, it is pre- using the point as the central point that radiation is drawn to space, interval If angle draws a plurality of radiation；Gather and stored on a plurality of radiation, positioned at femur successively according to the lead direction of radiation Multiple sampled points inside head bounding box or neck of femur bounding box；

For example, in one plane, o'clock a radiation is drawn every 2 ° centered on a point, then on this plane altogether 180 radiation (totally 360 °) can be drawn, are gathered successively on this radiation, positioned at stock along the lead direction of every radiation Sampled point inside bone bounding box or neck of femur bounding box, whole sampled points of determination are stored according to sampling order.

S904：By multiple sampled points of storage respectively according to preset order (such as according to sequence counter-clockwise) and neighbor point group Into triangular plate (referring to Figure 12 A), postmenstruation, which renders, obtains femur head model and neck of femur model ((referring to Figure 12 B)；

S905：Two-dimentional sampled images are generated according to the multiple sampled point；

Figure 13 A to Figure 13 C show the schematic diagram of the two-dimentional sampled images according to one example embodiment of the present invention；Such as Shown in Figure 13 A to Figure 13 C, in one embodiment, by above-mentioned multiple sampled points, pressed corresponding to the sampled point of same radiation It is arranged in order from left to right according to the lead direction of corresponding radiation；And the drawing according to corresponding radiation by the sampled point of arrangement Go out order to be arranged in order from the top down, to determine two-dimentional sampled images (referring to Figure 13 A)；

Wherein, a line sampled point belonging to the sampled point correspondence on each radiation on two-dimentional sampled images.

S906：Determine the cortex of bone external boundary of femur head region (referring to the vertical side in Figure 13 B on two-dimentional sampled images Upward White curves)；

In one embodiment, shortest-path method can be based on, the femur Head Section is determined on above-mentioned two-dimentional sampled images The cortex of bone external boundary in domain.

Wherein, femoral head and neck of femur external boundary are the border for distinguishing the inside of femoral head and neck of femur with outside, The process for finding border is the process in the path (shortest path) for finding a sampled point connection Least-cost, and above-mentioned cost is to work as The connection cost of the preceding adjacent point of sampled point, its calculation formula such as following formula (1)：

Cost=w₁*A+w₂*B+w₃*C (1)

Wherein, Cost is the cost of 2 points of connections, and A is the sum of 2 points of horizontal two neighborhood gray scale difference values, and B is 2 gradients Difference, C are 2 points of distance, w₁、w₂And w₃Respectively A, B and C weight (can rule of thumb set)；

In one embodiment, A value=(gray scale of the neighbor point of gray scale-right side one of the first from left neighbor point)+(the second from left neighbor point The gray scale of the neighbor point of gray scale-right side two) (2)

Further, on the basis of the cortex of bone external boundary to femur head region is determined, the method for the present embodiment is also It can include：

S907：Based on growth method outside region, lesser trochanter is split in every piece image to femur proximal part pre-determined distance Region (referring to the C regions shown in Figure 10 A)；

Existing region growth method is that seed point is selected in target area, is then increased with certain threshold range, The region finally given i.e. final goal region.And it is that (i.e. lesser trochanter is pre- to femur proximal part in target area in the present embodiment If the region of distance) the selected seed point in outside, increased with certain threshold range, obtain first area, the first area is not As target area, but it is target area that the first area is integrally subtracted with image.

On this basis, the method for the present embodiment can also include：

S908：Split greater trochanter in every piece image to the region between lesser trochanter (referring to the B areas shown in Figure 10 A Domain)；

Specifically, increase by the upward high density of Seed Layer of the C regions the superiors, using a-quadrant as left border, with higher than one As the condition of bone common threshold increased, obtain most of bone outer surface, continue to increase as seed point using the point of acquisition, End condition is obtained using local form and threshold value, completes segmentation.

In one embodiment, greater trochanter is obtained to the integral outer wheel in the region between lesser trochanter with higher predetermined threshold value Exterior feature, then carry out region growth downwards since greater trochanter part；

Wherein, due in the CT images of levels skeletal form profile it is close, thus increased using layer inner region, interlayer wheel The method of exterior feature limitation carries out region growth, and the region of growth is handled based on morphology principle, to determine more accurately Edge.

S909：Based on image segmentation result, important parameter is measured；

In one embodiment, after the trizonal segmentation of A, B, C is completed, important parameter can be measured.With Illustrated exemplified by 3 important parameters in femoral head replacement operation.3 important parameters include：The mean radius of femoral head, Eccentric throw and collodiaphyseal angle.

Figure 14 shows multiple important parameters in being performed the operation according to the femoral head replacement of one example embodiment of the present invention Schematic diagram；As shown in figure 14, straight line a is the center line of femoral shaft, and straight line b is the center line of neck of femur, and point O is in femoral head Heart point, point Q are subpoints of the O in straight line a, and point P is straight line a, b intersection point, and line segment l is femoral head radius, and line segment m is bias Away from angle n is collodiaphyseal angle.

Wherein, the computational methods of femoral head mean radius be segmentation optimize femoral head after, calculate each boundary point with The average value of the distance sum in the center of circle is as femoral head mean radius l.

Eccentric throw refers to that late-segmental collapse point to the vertical range m of femoral shaft center line, is successively counted when splitting C regions Central point is calculated, cluster obtains the center line a that overall center line is femoral shaft；

Collodiaphyseal angle refers to neck of femur center line b and femoral shaft center line a angle n.

Figure 15 is shown outside the cortex of bone for how determining femur head region according to one example embodiment of the present invention The flow chart on boundary；As shown in figure 15, above-mentioned steps S906 is determined on two-dimentional sampled images outside the cortex of bone of femur head region Boundary, it can include as follows：

S961：The priority query for storing sampled point is initialized, and is each sampling of above-mentioned two-dimentional sampled images Point assigns a mark；Wherein, the species of described mark includes：Frozen (point of initialization), Faraway (also not by The point of judgement), Narrow (point of the existing cost value judged)；

In one embodiment, in priority query the priority of each sampled point according to corresponding to the sampled point cost value it is big Small determination, the smaller then priority of cost value are higher.

All sampled points in the first row of above-mentioned two-dimentional sampled images are stored in initial phase, priority query, And " Frozen " mark is assigned for the point of the first row in two-dimentional sampled images, assign " Faraway " mark for remaining point.

S962：Judge whether current queue is empty：

If so, terminate to calculate；

If it is not, the maximum sampled point of priority is then ejected, calculate respectively in the sampled point and two-dimentional sampled images (lower-left, Just lower and bottom right) three neighbor points cost value, and assign the cost value of calculating to corresponding neighbor point respectively；

In one embodiment, first sampled point ejected in queue is shade of gray in the first row of two-dimentional sampled images It is worth maximum sampled point.

S963：The access path of the sampled point and the neighbor point of cost value minimum of ejection is recorded, and judges cost value minimum Neighbor point whether be located at image last column：

If so, terminate to calculate；

If it is not, judge the mark of the minimum neighbor point of the cost value：

S964：If the neighbor point is identified as Faraway, the cost value for assigning the neighbor point is：The sampling of the ejection The mark of the neighbor point is arranged to by the current cost value of point plus the sampled point of the ejection and the connection cost of the neighbor point Narrow, and the neighbor point is added into above-mentioned priority query, return to step S962；

S965：If the mark of the neighbor point is not Faraway (being Narrow), judge that the neighbor point currently calculates Cost value whether be less than the neighbor point existing cost：

If so, the existing cost for then changing the neighbor point is the cost currently calculated of the neighbor point, and this is neighbouring Point adds above-mentioned priority query, return to step S962.

S966：After terminating to calculate, access path corresponding to the minimum sampled point of cost value is defined as shortest path；

S967：Pair determine shortest path optimize；

Due to inner side and the acetabular bone of femoral head lean on it is very near, and because by CT machines, precision itself is influenceed, stock Border more unobvious between the inner side of bone and acetabular bone, femur Head Section can not be accurately determined with above-mentioned shortest-path method The external boundary in domain.Figure 12 C show that the structure of the femur head model before the optimization according to one example embodiment of the present invention is shown It is intended to.As indicated in fig. 12 c, if it is determined that external boundary it is inaccurate, then obvious " pit " occurs in model surface, that is, occurs bent The curve that rate changes greatly, thus need to optimize the external boundary.

In one embodiment, above-mentioned optimization process can include one of following two step or whole：

S9671：Femoral head model vertices to curvature between the edge of neck of femur model, remote femur head model is extracted to become Change larger curve, adjust the bent curvature of a curve so that its Curvature varying is uniformly gentle, and it is adjacent correspondingly to adjust the curve Bent curvature of a curve；

S9672：Determine that the larger curve of Curvature varying corresponds to the position range of acetabular bone, in the external boundary that currently determines and right One Local Extremum is set between the acetabular bone answered, shortest path side is based on according to the Local Extremum again in the subrange The external boundary point that method calculates, such as the external boundary for providing newly to determine must be between the extreme point and acetabular bones, and other are not corresponding The boundary point of extreme point take advantage of a situation the new boundary point (White curves that final result is shown in Figure 13 C) of connection so that Curvature varying The Curvature varying of larger curve reduces.

Figure 12 D show the structural representation of the femur head model after the optimization according to one example embodiment of the present invention Figure；As indicated in fig. 12d, after optimization, " pit " of model surface is filled and led up, i.e., the bent curvature of a curve at " pit " place becomes It is small.

Compared with prior art, femoral head region segmentation method, device and electronic equipment in CT images of the invention, lead to Acquisition subject femur head region CT sequence images are crossed, and pass through the femur head region in every piece image to CT sequence images Positioned, femoral head bounding box and neck of femur bounding box are built, based on space radiation collimation method in femoral head bounding box and femur Multiple sampled points are obtained in neck bounding box, and two-dimentional sampled images are generated according to multiple sampled points, and then in two-dimentional sampled images The upper cortex of bone external boundary for determining femur head region, it is possible to achieve the bone skin of femur head region in accurate Ground Split CT sequence images Matter external boundary, the implementation for schemes such as subsequent artefacts' joint replacements provide foundation.

Those skilled in the art will readily occur to the present invention its after considering specification and putting into practice invention disclosed herein Its embodiment.The application be intended to the present invention any modification, purposes or adaptations, these modifications, purposes or Person's adaptations follow the general principle of the present invention and including undocumented common knowledges in the art of the invention Or conventional techniques.Description and embodiments are considered only as exemplary, and true scope and spirit of the invention are by the application Claim point out.

It should be appreciated that the invention is not limited in the precision architecture for being described above and being shown in the drawings, and And various modifications and changes can be being carried out without departing from the scope.The scope of the present invention is only limited by appended claim.

Claims

A kind of 1. femoral head region segmentation method in CT images, it is characterised in that including：

Obtain subject femur head region CT sequence images；Wherein, the femur head region includes femoral head to the area of neck of femur Domain；

Positioned by the femur head region in every piece image to the CT sequence images, structure femoral head surrounds Box and neck of femur bounding box；Wherein, the femoral head bounding box is used to surround femur head model to be built, the neck of femur bag Enclose box to be used to surround neck of femur model, the femoral head bounding box is connected with one end of neck of femur bounding box；

Multiple sampled points are obtained in the femoral head bounding box and the neck of femur bounding box based on space radiation collimation method；

Two-dimentional sampled images are generated according to the multiple sampled point；

The cortex of bone external boundary of the femur head region is determined on the two-dimentional sampled images.
2. according to the method for claim 1, it is characterised in that every piece image by the CT sequence images In the femur head region positioned, build femoral head bounding box and neck of femur bounding box, including：

The femur head region is positioned in every piece image, to determine the femoral head in every piece image Central point, center line, the radius of radius and neck of femur；

According to the central point of femoral head, the center line of radius and neck of femur, the radius per in piece image, described in estimation Center line, the radius of the central point of femur head model, radius and the neck of femur model；

Central point, the radius of the femoral head bounding box are determined according to central point, the radius of the femur head model, and according to institute State the central point of femoral head bounding box, radius builds the femoral head bounding box；Wherein, the central point of the femoral head bounding box Overlapped with the central point of the femur head model, the radius of the femoral head bounding box is more than the radius of the femur head model；

Center line, the radius of the neck of femur bounding box are determined according to center line, the radius of the neck of femur model, and according to institute State the center line of neck of femur bounding box, radius builds the neck of femur bounding box；Wherein, the center line of the neck of femur bounding box Overlapped with the center line of the neck of femur model, the radius of the neck of femur bounding box is more than the radius of the neck of femur model.
3. according to the method for claim 2, it is characterised in that described to be surrounded based on space radiation collimation method in the femoral head Multiple sampled points are obtained in box and the neck of femur bounding box, including：

On the center line of the neck of femur bounding box, and the prolonging inside the femoral head bounding box of the center line On long line, multiple points are determined；

On the cross section belonging to every bit in the multiple point, the point centered on the every bit, interval predetermined angle to Draw a plurality of radiation in space；Wherein, center line and the extended line of the cross section perpendicular to the neck of femur bounding box；

According to the lead direction of every radiation in a plurality of radiation, gather successively on a plurality of radiation, positioned at the stock Multiple sampled points inside bone bounding box and neck of femur bounding box.
4. according to the method for claim 3, it is characterised in that described that two-dimentional sample graph is generated according to the multiple sampled point Picture, including：

The multiple sampled point is arranged in order according to the lead direction and extraction order of corresponding radiation, to generate State two-dimentional sampled images.
5. according to the method for claim 1, it is characterised in that described to determine the femur on the two-dimentional sampled images The cortex of bone external boundary of head region, including：

According to the connection cost of each sampled point and neighbor point on the two-dimentional sampled images, the connection of total connection Least-cost is obtained Path；

The cortex of bone external boundary is determined according to the acquisition result of the access path.
6. according to the method for claim 5, it is characterised in that shown in the calculation formula such as following formula (1) of the connection cost：

Cost=w₁*A+w₂*B+w₃*C (1)

Wherein, Cost is the costs of 2 points of connections, and A is sum of 2 points of horizontal two neighborhood gray scale difference values, and B is the difference of 2 gradients, C For 2 points of distance, w₁、w₂And w₃Respectively A, B and C weight.
7. according to the method for claim 5, it is characterised in that it is described according to each sampled point on the two-dimentional sampled images with The connection cost of neighbor point, the access path of total connection Least-cost is obtained, including：

Initialization one is used for the priority query for storing the sampled point；Wherein, in the priority query each sampled point it is excellent First level is determined by the size of corresponding connection cost；

The maximum sampled point of priority is taken out from the priority query, calculates the sampled point of taking-up and the generation of multiple neighbor points Value, assigns the cost value of calculating to corresponding neighbor point respectively；

Judge the species of the minimum neighbor point of the cost value, it is minimum to update the cost value according to the judged result of the species Neighbor point cost value, and the minimum neighbor point of the cost value is added into the priority query；

Repeat it is described taken out from the priority query the maximum sampled point of priority to it is described by the cost value most Small neighbor point adds the step of priority query, until meet preset termination condition, current cost value is minimum Access path of the access path corresponding to sampled point as total connection Least-cost.
8. according to the method for claim 5, it is characterised in that described to determine the femur on the two-dimentional sampled images After the cortex of bone external boundary of head region, methods described also includes：

Pair determine the cortex of bone external boundary optimize.
A kind of 9. femoral head region segmentation device in CT images, it is characterised in that including：

Image collection module, for obtaining subject femur head region CT sequence images；Wherein, the femur head region includes stock Bone is to the region of neck of femur；

Bounding box builds module, for being carried out by the femur head region in every piece image to the CT sequence images Positioning, build femoral head bounding box and neck of femur bounding box；Wherein, the femoral head bounding box is used to surround femur to be built Head model, the neck of femur bounding box are used to surround neck of femur model, and the one of the femoral head bounding box and neck of femur bounding box End is connected；

Sampled point acquisition module, for based on space radiation collimation method in the femoral head bounding box and the neck of femur bounding box Obtain multiple sampled points；

Image generation module, for generating two-dimentional sampled images according to the multiple sampled point；

Border determining module, for determining the cortex of bone external boundary of the femur head region on the two-dimentional sampled images.
10. device according to claim 9, it is characterised in that the bounding box builds module, including：

Zone location unit, it is described every to determine for being positioned in every piece image to the femur head region The central point of femoral head in piece image, the center line of radius and neck of femur, radius；

Parameter estimation unit, in the central point, radius and neck of femur of the femoral head in every piece image Heart line, radius, estimate center line, the radius of the central point of the femur head model, radius and the neck of femur model；

First construction unit, for being determined according to central point, the radius of the femur head model in the femoral head bounding box Heart point, radius, and the femoral head bounding box is built according to the central point of the femoral head bounding box, radius；Wherein, the stock The central point of bone bounding box overlaps with the central point of the femur head model, and the radius of the femoral head bounding box is more than described The radius of femur head model；

Second construction unit, for being determined according to center line, the radius of the neck of femur model in the neck of femur bounding box Heart line, radius, and the neck of femur bounding box is built according to the center line of the neck of femur bounding box, radius；Wherein, the stock The center line of bone neck bounding box overlaps with the center line of the neck of femur model, and the radius of the neck of femur bounding box is more than described The radius of neck of femur model.
11. device according to claim 10, it is characterised in that the sampled point acquisition module, including：

First determining unit, for the center line in the neck of femur bounding box, and the center line is located at the stock On extended line inside bone bounding box, multiple points are determined；

Radiation draws unit, for the cross section belonging to the every bit in the multiple point, using the every bit in Heart point, interval predetermined angle draw a plurality of radiation to space；Wherein, the cross section is perpendicular to the neck of femur bounding box Center line and the extended line；

Sampled point collecting unit, for the lead direction according to every radiation in a plurality of radiation, gather successively a plurality of Multiple sampled points on radiation, inside the femoral head bounding box and neck of femur bounding box.
12. device according to claim 11, it is characterised in that described image generation module, including：

Image generation unit, for the multiple sampled point to be carried out according to the lead direction and extraction order of corresponding radiation It is arranged in order, to generate the two-dimentional sampled images.
13. device according to claim 9, it is characterised in that the border determining module, including：

Path acquiring unit, for the connection cost according to each sampled point and neighbor point on the two-dimentional sampled images, obtain total Connect the access path of Least-cost；

Border acquiring unit, for determining the cortex of bone external boundary according to the acquisition result of the access path.
14. device according to claim 13, it is characterised in that calculation formula such as following formula (1) institute of the connection cost Show：

Cost=w₁*A+w₂*B+w₃*C (1)

Wherein, Cost is the costs of 2 points of connections, and A is sum of 2 points of horizontal two neighborhood gray scale difference values, and B is the difference of 2 gradients, C For 2 points of distance, w₁、w₂And w₃Respectively A, B and C weight.
15. device according to claim 13, it is characterised in that the path acquiring unit is used for：

Initialization one is used for the priority query for storing the sampled point；Wherein, in the priority query each sampled point it is excellent First level is determined by the size of corresponding connection cost；

The maximum sampled point of priority is taken out from the priority query, calculates the sampled point of taking-up and the generation of multiple neighbor points Value, assigns the cost value of calculating to corresponding neighbor point respectively；

Judge the species of the minimum neighbor point of the cost value, it is minimum to update the cost value according to the judged result of the species Neighbor point cost value, and the minimum neighbor point of the cost value is added into the priority query；

Repeat it is described taken out from the priority query the maximum sampled point of priority to it is described by the cost value most Small neighbor point adds the step of priority query, until meet preset termination condition, current cost value is minimum Access path of the access path corresponding to sampled point as total connection Least-cost.
16. device according to claim 13, it is characterised in that described device also includes：

Border optimization module, optimized for pair cortex of bone external boundary determined.
17. a kind of electronic equipment, it is characterised in that the electronic equipment includes：

Processor；

It is configured as storing the memory of processor-executable instruction；

Wherein, the processor is configured as：

Obtain subject femur head region CT sequence images；Wherein, the femur head region includes femoral head to the area of neck of femur Domain；

Positioned by the femur head region in every piece image to the CT sequence images, structure femoral head surrounds Box and neck of femur bounding box；Wherein, the femoral head bounding box is used to surround femur head model to be built, the neck of femur bag Enclose box to be used to surround neck of femur model, the femoral head bounding box is connected with one end of neck of femur bounding box；

Multiple sampled points are obtained in the femoral head bounding box and the neck of femur bounding box based on space radiation collimation method；

Two-dimentional sampled images are generated according to the multiple sampled point；

The cortex of bone external boundary of the femur head region is determined on the two-dimentional sampled images.
18. a kind of computer-readable recording medium, is stored thereon with computer program, it is characterised in that the program is by processor Realized during processing：

Obtain subject femur head region CT sequence images；Wherein, the femur head region includes femoral head to the area of neck of femur Domain；

Positioned by the femur head region in every piece image to the CT sequence images, structure femoral head surrounds Box and neck of femur bounding box；Wherein, the femoral head bounding box is used to surround femur head model to be built, the neck of femur bag Enclose box to be used to surround neck of femur model, the femoral head bounding box is connected with one end of neck of femur bounding box；

Multiple sampled points are obtained in the femoral head bounding box and the neck of femur bounding box based on space radiation collimation method；

Two-dimentional sampled images are generated according to the multiple sampled point；

The cortex of bone external boundary of the femur head region is determined on the two-dimentional sampled images.