CN101561937B - Big data size medical image three-dimensional interaction method based on common computer - Google Patents
Big data size medical image three-dimensional interaction method based on common computer Download PDFInfo
- Publication number
- CN101561937B CN101561937B CN2009100721381A CN200910072138A CN101561937B CN 101561937 B CN101561937 B CN 101561937B CN 2009100721381 A CN2009100721381 A CN 2009100721381A CN 200910072138 A CN200910072138 A CN 200910072138A CN 101561937 B CN101561937 B CN 101561937B
- Authority
- CN
- China
- Prior art keywords
- data
- precision
- data block
- medical image
- dimensional
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 50
- 230000003993 interaction Effects 0.000 title claims abstract description 14
- 238000013507 mapping Methods 0.000 claims abstract description 8
- 238000004364 calculation method Methods 0.000 claims abstract description 6
- 241001269238 Data Species 0.000 claims description 5
- 238000005520 cutting process Methods 0.000 claims description 4
- 230000011218 segmentation Effects 0.000 claims description 3
- 238000000547 structure data Methods 0.000 claims description 3
- 230000002452 interceptive effect Effects 0.000 abstract description 8
- 238000012800 visualization Methods 0.000 abstract description 7
- 238000000638 solvent extraction Methods 0.000 abstract 1
- 230000008569 process Effects 0.000 description 20
- 230000000007 visual effect Effects 0.000 description 8
- 230000000694 effects Effects 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 6
- 238000009877 rendering Methods 0.000 description 6
- 230000014509 gene expression Effects 0.000 description 5
- 238000011160 research Methods 0.000 description 4
- 238000007794 visualization technique Methods 0.000 description 4
- 230000007423 decrease Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000018109 developmental process Effects 0.000 description 3
- 238000002059 diagnostic imaging Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 230000008846 dynamic interplay Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000009412 basement excavation Methods 0.000 description 1
- 238000005266 casting Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000002405 diagnostic procedure Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 238000003780 insertion Methods 0.000 description 1
- 230000037431 insertion Effects 0.000 description 1
- 210000004185 liver Anatomy 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012856 packing Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000013341 scale-up Methods 0.000 description 1
- 238000002603 single-photon emission computed tomography Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Images
Landscapes
- Image Generation (AREA)
- Processing Or Creating Images (AREA)
Abstract
The invention provides a big data size medical image three-dimensional interaction method based on a common computer, and relates to the technical field of computer graphics and information visualization. The invention aims to solve the problem of the interaction of reconstructed big data size three-dimensional models. The method is realized by the following steps that: a medical image three-dimensional data body is partitioned into blocks in multi-precision, and the partitioned data is stored by an octree structure; the multi-precision data blocks undergo texture mapping, namely, a mapping relation of the multi-partitioning data and an image texture buffer of a microcomputer graphic card is established; a group of tangent planes which are equidistant and parallel to each other are selected, a plurality of sections are generated when the group of tangent planes are intersected with the medical image three-dimensional data body, the data blocks positioned on the sections are determined , and the precisions of the data blocks are orderly selected in the mode of traversing the data blocks with the octree structure; the interactive drawing is realized, namely, the sections formed by the data blocks are sent to the image texture buffer of the microcomputer graphic card, and the microcomputer graphic card finishes the interpolation calculation of the gaps among the sections, so that a three-dimensional image is drawn.
Description
Technical field
The present invention relates to computer graphics and information visualization technical field.Be specifically related to the three-dimensional interaction method of medical image.
Background technology
Along with progress of science and technology; Modern medical imaging device function is more and more stronger; Various medical imaging devices have become the important tool that the medical personnel draws diagnosis basis, like CT, MR, PET, SPET, DSA or the like, all can generate anatomic image clearly to the different causes of disease.At present, the medical imaging device of main flow all can provide bidimensional image sequence equidistant in a large number, spatial alignment.Along with the continuous progress of image technology, the data volume that image documentation equipment generates is also increasing, begins to the development of G rank from initial tens, up to a hundred million, and this has also caused it is analyzed the increase day by day of required workload.Draw diagnostic result accurately, the doctor need rely on its professional knowledge and subjective judgement, three-dimensional position, shape, the size of design focus in human body from two-dimensional images, thus caused the not directly perceived of whole process.In order to make diagnostic procedure more directly perceived, scholars have transferred to three-dimensional aspect with sight line, hope the tomographic sequence of continuous two dimension is gone out three-dimensional effect through computer reconstruction, and this technology is referred to as the scientific visualization technology.Visualization technique can directly be operated the people to corporality information in the three-dimensional picture world, directly exchange with computing machine.This technology is unified the mode of the strength of people and machine nature with a kind of intuition, and this revolutionary variation will greatly improve people's work efficient undoubtedly.
With regard to the mode of three-dimensional visualization, be divided into two kinds at present substantially: based on the iso-surface patch method of contour surface with based on the object plotting method of volume elements.The iso-surface patch method is in 3 d data field, to extract the contour surface that the surface geometry pel that is complementary with threshold value is spliced by the given threshold value of user; And then by traditional calculating machine graphics technology realization pattern drafting, like the Marching cubes algorithm of classics.The volume drawing rule is directly to generate final image by three-dimensional data, like ray-casting, splatting and shear-warp scheduling algorithm.The reconstruction speed of iso-surface patch is very fast, on common computer, basically can realize mutual drafting, but to comprise information less owing to it, can't reflect whole characteristics of three-dimensional data.Though contain much information and volume drawing comprises, the calculated amount in drawing process also is far longer than the iso-surface patch method, therefore when data volume acquired a certain degree, the realization that relies on software algorithm to be difficult to faster merely was visual.
1993; Cullip and Neumann utilize the RealityEngine graphics workstation of SGI company to realize hardware-accelerated volume drawing first; For a new direction has been pointed out in the research of three-dimensional visualization; Visualization technique begins to be quickened to change to hardware-accelerated direction by software, and the 3 D medical data normally are formed by stacking two-dimensional image sequence, after the volume data that will form after above-mentioned hardware-accelerated technology will superpose is loaded into the graphical textures buffer memory of computer display card; Utilize graphics card hardware to accomplish complicated interpolation calculation, be redeveloped into three-dimensional model.Behind twentieth century, the quality that various image documentation equipments generate image is also progressively increasing, and high-quality information acquiring technology also provides data source true to nature for visualization technique.Also, cause the surge of medical image data amount, to its three-dimensional visualization more and more difficult that also becomes just because of the lifting of data acquisition precision and picture quality.The big data quantity three-dimensional visualization realizes that the main cause of difficulty is memory capacity and the bandwidth that computer hardware is limited, and therefore in order to realize visual to big data size medical image, method is to improve the hardware performance of computing machine the most intuitively.But at first the growth rate of data is considerably beyond the growth of hardware performance; Secondly, high performance hardware must be accompanied by very expensive price, and this also is a very big obstacle for this technology popularizing of each application.Along with medical field gradually to " becoming more meticulous " development, visual requirement no longer is confined to the aspect of " seeing ", also need in " seeing ", can carry out alternately with reconstruction model, this also is the fundamental purpose of visual research.At present; Most big data quantity visual research still concentrates on the three-dimensional reconstruction aspect that how to realize big data quantity; Promptly how can demonstrate final three-dimensional model more quickly, also not about how the big data quantity three-dimensional model after rebuilding being carried out mutual research.Among the present invention alleged three-dimension interaction promptly be meant to accomplished become 3-D view from the tomographic sequence of two dimension through computer reconstruction after, again to the rotation or the translation of this 3-D view.Mutual through to this 3-D view makes the medical personnel can obtain the angle or the position of its 3-D view that needs.
Summary of the invention
The purpose of this invention is to provide a kind of big data size medical image three-dimensional interaction method, the big data quantity three-dimensional model after rebuilding is carried out mutual problem to solve based on common computer.
The present invention realizes through following step: one, many precision piecemeal of medical image 3-D data volume: the 3-D data volume that is specially medical image carries out repeatedly equal portions piecemeal by precision; Wherein high-precision equal portions piecemeal is the segmentation once more on each blocks of data basis that the equal portions piecemeal that hangs down precision is produced, and the data behind the piecemeal are stored with octree structure;
Two, multiple precision number is according to the texture of piece: be specially the mapping relations between the graphical textures buffer memory of setting up many block datas and microcomputer video card, thereby realize representing with the variation of texture the precision variation of data block;
Three, selected one group of equidistance and parallel tangent plane utilize should the group tangent plane to intersect with the medical image 3-D data volume to produce a plurality of sections, confirm to be positioned in the section data block and through mode that the octree structure data block the is traveled through precision of selected data piece successively; The accuracy selection of data block is so that the max cap. that the overall texture dimensions sum of all data blocks can be held less than microcomputer video card buffer memory in its texture precision maximum and a plurality of sections;
Four, mutual realization of drawing: send into the graphical textures buffer memory of microcomputer video card to the section of forming by data block, accomplish the interpolation calculation in the gap of respectively cutting into slices by the microcomputer video card, thus the drawing three-dimensional image.
The visual fundamental purpose of medical 3 D is in order to make the medical personnel to carry out alternately with medical data more easily, and the development of large scale integrated circuit also makes the function of computer graphical hardware and processing power more and more stronger.The present invention then is a texture function of utilizing present common computer graphic hardware to have, in conjunction with effective data structure and dispatching algorithm, on common computer, realizes the dynamic interaction to big data quantity three-dimensional reconstruction model.And the test to invention is on the microcomputer of present terminal configuration, to carry out, and concrete parameter is: Pentium IV 3.0G, and internal memory 1GB, video card is for supporting the Radeon X600 of three-D grain, video memory size 256MB.The present invention carries out piecemeal with the 3 D medical image set by precision, and sets up octree structure and come it is managed.In reciprocal process, data are written into the graphical textures buffer memory by different level according to the strategy that is designed, thereby break through the restriction of common computer graphical textures buffer memory and bandwidth, realize that successively many precision of big data size medical image are drawn with mutual.This method has been carried out effective the combination with the common computer texture features with dispatching algorithm, when taking into account rendering quality and interactivity, effectively reduces the demand of big data quantity visualization technique to hardware.To the visual problem to hardware performance, capacity high request of big data size medical image, the present invention accomplishes the Three-Dimensional Dynamic of the big data size medical image collection that surpasses the restriction of computer display card buffer memory capacity mutual with common computer.Ardware feature through abundant excavation graphics card; Realize a kind of rational graphical textures buffer scheduling method in conjunction with tree form data structure and structure Block Sort chain table algorithm; Under the prerequisite that guarantees certain three-dimensional reconstruction quality, on common computer, realize mutual to the Three-Dimensional Dynamic of big data size medical image collection.
Description of drawings
Fig. 1 is a deblocking and the synoptic diagram of setting up Octree in the embodiment one; Fig. 2 is the texture primitive synoptic diagram of one-dimensional data; A represent pixel wherein, c represents shared region.Fig. 3 to Fig. 6 is the accuracy selection process 4 step synoptic diagram of a certain section in the data of piecemeal; Fig. 7 is the synoptic diagram of embodiment two, because the variation of slice position must cause the again variation of data block in the graphical textures buffer memory when mutual, the data block chained list then is to set up according to the difference of this variation; Fig. 8 to Figure 12 is the synoptic diagram that embodiment five confirms to be positioned at a kind of method of the data block in the section; Figure 13 to 18 is embodiment six experimental result synoptic diagram.Wherein Figure 13 and Figure 14 represent the mutual effect of spine medical data, and the size of the Octree leaf node of being set up among Figure 13 and Figure 14 is followed successively by 64
3With 32
3The mutual effect of Figure 15 and 16 expression liver medical datas, the size of the Octree leaf node of being set up among Figure 15 and 16 is followed successively by 128
3With 64
3The mutual effect of Figure 17 and the high-precision head medicine data of 18 expressions, the size of the Octree leaf node of being set up among Figure 17 and 18 is followed successively by 64
3With 256
3
Embodiment
Embodiment one: this embodiment is realized by following step: one, many precision piecemeal of medical image 3-D data volume: the 3-D data volume that is specially medical image carries out repeatedly equal portions piecemeal by precision; Wherein high-precision equal portions piecemeal is the segmentation once more on each blocks of data basis that the equal portions piecemeal that hangs down precision is produced; Thereby make divided data piece size in the graphical textures buffer memory of microcomputer video card, to load and unload; Data behind the piecemeal are stored with octree structure, and in this octree structure, realize traversal data block; The size of the data block that data block comprises of same precision and size are uniformly, and each node of Octree data block in the corresponding three-dimensional medical image all.The medical 3 D data normally are formed by stacking according to order a plurality of two dimensional images; Therefore between per two two dimensional images, there is seam; The present usual hardware speed technology of this seam is handled; The volume data that processing procedure forms after needing to superpose is loaded into the graphical textures buffer memory, utilizes graphics card hardware to accomplish complicated interpolation calculation again, thereby is reconstructed into the medical image 3-D data volume.Owing to carry out the restriction of transfer bandwidth and microcomputer video memory size between computer display card and computer bus, for the volume data of big data quantity, can't once it all be loaded in the video memory, this is to cause one of render speed main reasons for decrease.
In order to guarantee the rationality of video memory loading data in the drawing process, the volume data piecemeal with big utilizes octree structure to store.As shown in Figure 1, whole volume data is divided into the piece of appropriate size, and wherein the piece of minimum dimension constitutes the leaf node of Octree, and per 8 adjacent unidimensional pieces then constitute its father node, are used for representing more other piece of coarse grade.By that analogy, final root node is whole initial body data.The root node is here represented minimum precision, and leaf node is then represented full accuracy.
When utilizing Octree dividing data piece and it is carried out access,, sometimes need ignore father node and directly begin to divide from its child node because the amount of memory that computing machine can be divided is limited.For example in 32 machines commonly used; The memory block that once can mark off is 2GB to the maximum, for the data greater than 2G, just can't distribute the required internal memory of root node and father node; In this case; Just need begin to divide, and father node only keeps as addressing usefulness from a certain child node, and actual storage data (being gray scale or color) not.
Two, multiple precision number is according to the texture of piece: be specially the mapping relations between the graphical textures buffer memory of setting up many block datas and microcomputer video card, thereby realize representing with the variation of texture the precision variation of data block; The precision variation of data block is to represent how much confirming of data bit with texture primitive in the graphic hardware.
For the data block of different accuracy grade in the Octree, represent with the texture of different accuracy grade.At first, according to the evolution of texture mapping technology, begin from the method for expressing of the simplest one dimension texture.Initial texture mapping technology is that a certain pixel of expression is the Interpolation Process at center, suppose that one section one-dimensional data is to be combined by m adjacent pixels, but then its linear mapping in the hardware texture function is the value on the interval [1/2m, (2m-1)/2m].Texture for each data segment is represented, in the graphical textures buffer memory, all is mapped as [0,1] interval value, and the texture of each data segment representes to be referred to as texture tile unit (pixel) like this.Because the neighbouring relations between pixel must some be shared between per two sheet units, the shared region that promptly indicates among Fig. 2, this also is the reason of texture function interval less than [0,1].Utilize texture sub sampling technology, can in the three-D grain function of common computer graphic hardware, realize equally.Here the grade of precision just refers to the scope of the volume data value that texture tile unit is represented, and the scope of the volume data that data block was represented when the texture precision was high is little, can reach relation one to one; Otherwise then the scope of the data value of each sheet unit expression is big, and the result who causes like this is: the interactive speed when the texture precision is low is fast, and the rendering quality of texture precision when high is high.
Three, selected one group of equidistance and parallel tangent plane utilize should the group tangent plane to intersect with the medical image 3-D data volume to produce a plurality of sections, confirm to be positioned in the section data block and through mode that the octree structure data block the is traveled through precision of selected data piece successively; The accuracy selection of data block is so that the max cap. that the overall texture dimensions sum of all data blocks can be held less than microcomputer video card buffer memory in its texture precision maximum and a plurality of sections; Said section can be the state in the face of computer screen after having accomplished this each mutual step, that is to say that this group tangent plane is as the criterion with the needs that meet the medical worker in the selection of the position of medical image 3-D data volume.
The accuracy selection principle of data block is: the data block that will be applied to different accuracy based on the volume drawing of texture; Crossing situation through judging tangent plane and data block is selected the different accuracy data block; When taking into account rendering quality and interactive speed, try to achieve the required data block of reconstruction model the best visual effect.In to the Octree ergodic process, use leaf node (being full accuracy) to represent data as far as possible, but because the video memory finite capacity, expressing possibility with full accuracy fully to exceed the video card capacity, reduces taking video memory so need to reduce precision.
One of main thought of the method for the volume drawing based on texture used in the present invention is exactly to represent body with cutting into slices.In this step, finding out the process that intersects the data block on the face is to ask for the process of section and volume data Bounding Box intersection point.Intersection point to ask for algorithm still identical with existing three-D grain object plotting method, but the form that intersects by original only with single whole body Bounding Box become with Octree in the different pieces of information piece intersect.
Utilize the tree that builds in the step 1, data are drawn.Different with previous methods is need ask for the intersection point of tangent plane and each data block Bounding Box, and be stored in the corresponding node data structure.This process is equivalent in this octree structure of the top-down insertion of tangent plane, obtains the piece that octree structure and this tangent plane intersect successively, and process such as Fig. 3 are to shown in Figure 6.Because length is limit, Fig. 3 to Fig. 6 only representes the part ergodic process of single tangent plane in Octree, and general steps is following:
Step 2. is judged the crossing situation of tangent plane and data block, if crossing, and this piece is not leaf node, and this data block is segmented once more.
Step 4. is then ended traversal if all back end of expression volume data are the capacity that the size of leaf node (being full accuracy) or these back end has reached the graphical textures buffer memory.
Wherein the size of leaf node then is provided with according to the difference of video card ardware feature and video memory size.For senior video card, the size of leaf node can be selected smaller; For the poor slightly video card of performance, then the size of leaf node should be selected bigger slightly.The piece that indicates shade that finally obtains among Fig. 6 is needed of this section of demonstration, and other piece all can be ignored after traversal, will its figure texture cache of packing in.In order to obtain high as far as possible image quality; What the traversal of the Octree behind the piecemeal was adopted is the method for depth-first traversal; In other words; This traversal method is exactly when guaranteeing interactive speed and considering hardware restriction itself, to represent volume data with higher precision grade as much as possible.
Four, mutual realization of drawing: send into the graphical textures buffer memory of microcomputer video card to the section of forming by data block successively, accomplish the interpolation calculation in each gap of cutting into slices by the microcomputer video card, thereby the drawing three-dimensional image is to accomplish alternately.The realization of this step can be adopted the pointed hardware-accelerated technology of background technology.
Embodiment two: this embodiment with the difference of embodiment one is: through step 3 Octree is traveled through; Can obtain the data block of different accuracy in section under the current three-dimensional position; And after the data after rebuilding were carried out interactive operation, the position of section can change.The present invention be directed to the bigger situation of data volume; On common computer, will all be written into the decline that microcomputer memory will inevitably cause whole rendering performance in batches, and select here to come before and after the maintenance interaction data block that changes to have taken place with chained list with changing the back relevant data block of section.What will show the tangent plane needs in the step 4 packagedly goes into chained list; Being called needs the piece chained list; With its with the graphical textures buffer memory in data block compare; Wherein being loaded but not belonged to needs the piece chained list, can constitute the future use block chained list, and remaining data block then constitutes in the graphical textures buffer memory needs heavily loaded piece chained list.Need compare with needing the piece chained list by heavily loaded piece chained list, the two different data block adds graphical textures buffer memory.At this moment have two kinds of situation:
1) the one, initiate block size can be with its direct adding less than the residual capacity of graphical textures buffer memory;
2) moreover be the residual capacity of initiate block size, at this moment need to add graphical textures data in buffer piece reduction precision greater than the graphical textures buffer memory.Promptly represent this place's texture block with the father node of Octree.Consider the drawing data that the present invention is directed to memory capacity much larger than the common computer graphic hardware, at this moment the size that need be written into the data block chained list need and draw between precision and compromise very likely greater than remaining texture cache in interactivity.
Because in the reciprocal process, texture variations should be continuous in theory, so need the data block portions of heavy duty still to reside in the graphical textures buffer memory.Therefore before being written into the graphical textures buffer memory to data block, detect the future use block chained list earlier, with this avoid data repeat be written into.Data block scheduling process when Fig. 4 provides dynamic interaction, the multiple precision number of volume drawing data block is set forth in step 3 according to the system of selection of piece, just repeats no more here.This mode can be accelerated reciprocal process.Be applicable in the repeatedly interactive operation.
Embodiment three: this embodiment with the difference of embodiment one is: the data block accuracy class of region of interest is higher than the data block away from region of interest in the accuracy selection of step 3 data block.In the implementation process of this embodiment; Serves as according to coming precision and speed are taken into account with the medical personnel to the region of interest of data; Consider the requirement of rendering quality; The data block that when mutual, need be written into is during greater than the graphical textures buffer memory; The reduction of precision begins to reduce precision at first from the texture block far away apart from region of interest, representes this data block with the precision grade of its adjacent upper level, octree structure is apart from region of interest data block farthest begin upwards to search for its father node at first.Like this, big data will be loaded into video memory effectively in batches and be drawn, and when guaranteeing the region of interest rendering quality, accelerate render speed.
Embodiment four: this embodiment with the difference of embodiment three is: it also comprises step 5, the 3-D view that obtains after step 4 is finished, and the data block accuracy class of raising region of interest reduces the accuracy class away from the data block of region of interest.Because in the drafting of medical volume data, for the area-of-interest that will observe, at first requiring has certain image quality, so just can be convenient to the doctor and obtain correct diagnostic result; Secondly, in order to make this method that certain popularization arranged, this method requires on common computer, it to be realized mutual, promptly requires certain mutual frame per second.It seems that the two be contradiction, realize this two requirements, must compromise to a certain degree be arranged to drawing process.Here we adopt the thought that reduces precision when mutual and improve precision when observing to accomplish this process.Because in the reciprocal process (when data are rotated) three-dimensional accuracy of data is required is not too high, and in the time need observing it after the mutual completion, can adopt higher precision to draw.So just, can reach the purpose of wanting through this process.
Embodiment five: this embodiment provides a kind of method of confirming to be positioned at the data block in the section in the step 3.Section equation E can be expressed as:
E(x，y，z)＝ax+by+cz (0-1)
Situation such as Fig. 8 after the two intersects.Make that AB is a certain limit on the volume data Bounding Box, if plane and AB have intersection point, then establishing intersection point is P, can be drawn by mathematical derivation:
Wherein t is a scale-up factor; Can know that by geometry arbitrary plane can be unique definite through the normal
on 1 N of plan meaning and this plane, then:
In data space; The normal in section
can through type (0-1) constant coefficient obtain, that is:
Can draw
equally as if
is not orthogonal to
and make
according to above condition, can
Draw as drawing a conclusion:
1. if
then limit
must have intersection point with the plane; And then can know; If P is put on the AB of limit in 0≤t≤1; If t＜0 or t＞1, then line segment AB and plane do not have intersection point, and promptly P is not on line segment AB.
2. if
then can also be divided into two kinds of situations is if
explains limit AB in the plane; If
plane and limit AB do not have intersection point.
As shown in Figure 8, the volume data Bounding Box has 8 summits, can form 12 limits.Can judge with above-mentioned conclusion for every limit, just can obtain all intersection points of plane and Bounding Box.If plane and cube intersect, then the intersection point number is 6 at most, and minimum is 3, and is extremely shown in Figure 12 like Fig. 9:
Embodiment six: be the validity of checking the inventive method; Under following configuration, three groups of volume datas are tested: operating system Windows XP; Processor is Pentium IV 3.0G, internal memory 1G, and video card is for supporting the Radeon X600 of three-D grain; Video memory size 256M, programmed environment VC6.0.Data size does not wait to several GB from tens MB, when data volume obviously surpasses the video memory capacity, draws and still can constantly carry out.Along with the increase of data volume, interactive speed can decrease, but under current configuration, still can reach interaction level, draws effect and lists in the table 1.In addition, under equal data volume situation, the data precision that reduces when mutual can effectively improve interactive speed.For outstanding piecemeal effect, we also begin the Bounding Box of Octree data block successively to draw out from root node, and effect is seen Figure 13 to 18.
The big data volume drawing result tabulation of table 1
Claims (5)
1. based on the big data size medical image three-dimensional interaction method of common computer; It is characterized in that it realizes through following step: one, many precision piecemeal of medical image 3-D data volume: the 3-D data volume that is specially medical image carries out repeatedly equal portions piecemeal by precision; Wherein high-precision equal portions piecemeal is the segmentation once more on each blocks of data basis that the equal portions piecemeal that hangs down precision is produced, and the data behind the piecemeal are stored with octree structure;
Two, multiple precision number is according to the texture of piece: be specially the mapping relations between the graphical textures buffer memory of setting up many block datas and microcomputer video card, thereby realize representing with the variation of texture the precision variation of data block;
Three, selected one group of equidistance and parallel tangent plane utilize should the group tangent plane to intersect with the medical image 3-D data volume to produce a plurality of sections, confirm to be positioned in the section data block and through mode that the octree structure data block the is traveled through precision of selected data piece successively; The accuracy selection of data block is so that the max cap. that the overall texture dimensions sum of all data blocks can be held less than microcomputer video card buffer memory in its texture precision maximum and a plurality of sections;
Four, mutual realization of drawing: send into the graphical textures buffer memory of microcomputer video card to the section of forming by data block, accomplish the interpolation calculation in the gap of respectively cutting into slices by the microcomputer video card, thus the drawing three-dimensional image.
2. the big data size medical image three-dimensional interaction method based on common computer according to claim 1; What it is characterized in that will showing in the step 4 tangent plane needs packagedly goes into chained list; To be called need the piece chained list, and itself and data block in the graphical textures buffer memory are compared, and wherein being loaded but not belonged to needs the piece chained list; Can constitute the future use block chained list, remaining data block then constitutes in the graphical textures buffer memory needs heavily loaded piece chained list; Need compare with needing the piece chained list by heavily loaded piece chained list, the two different data block adds graphical textures buffer memory.
3. the big data size medical image three-dimensional interaction method based on common computer according to claim 2 is characterized in that if newly add graphical textures data in buffer piece size less than the graphical textures buffer memory capacity, with its direct adding; If newly add graphical textures data in buffer piece size, will add graphical textures data in buffer piece and reduce precision greater than the graphical textures buffer memory capacity.
4. the big data size medical image three-dimensional interaction method based on common computer according to claim 1 is characterized in that the data block accuracy class of region of interest in the accuracy selection of step 3 data block is higher than the data block away from region of interest.
5. the big data size medical image three-dimensional interaction method based on common computer according to claim 4; It is characterized in that it also comprises step 5; To the 3-D view that obtains after the step 4 end; Improve the data block accuracy class of region of interest, reduce accuracy class away from the data block of region of interest.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2009100721381A CN101561937B (en) | 2009-05-27 | 2009-05-27 | Big data size medical image three-dimensional interaction method based on common computer |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2009100721381A CN101561937B (en) | 2009-05-27 | 2009-05-27 | Big data size medical image three-dimensional interaction method based on common computer |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101561937A CN101561937A (en) | 2009-10-21 |
| CN101561937B true CN101561937B (en) | 2012-05-09 |
Family
ID=41220723
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2009100721381A Active CN101561937B (en) | 2009-05-27 | 2009-05-27 | Big data size medical image three-dimensional interaction method based on common computer |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101561937B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663803B (en) * | 2012-04-13 | 2014-11-26 | 北京工业大学 | Simulation projection DRR generating method based on RayCasting improved algorithm |
| CN105912638A (en) * | 2016-04-08 | 2016-08-31 | 苏州中科图新网络科技有限公司 | Oblique photograph data storage and scheduling method and apparatus |
| JP6972647B2 (en) * | 2017-05-11 | 2021-11-24 | 富士フイルムビジネスイノベーション株式会社 | 3D shape data editing device and 3D shape data editing program |
| CN114305470A (en) * | 2017-11-10 | 2022-04-12 | 湖北锐世数字医学影像科技有限公司 | Coincidence event screening device for all-digital PET |
| CN109857741B (en) * | 2019-01-31 | 2021-04-06 | 重庆零壹空间航天科技有限公司 | Rocket telemetry data selection method and device |
| CN111105476B (en) * | 2019-12-24 | 2023-04-11 | 哈尔滨工业大学 | Three-dimensional reconstruction method for CT image based on Marching Cubes |
| CN112258661B (en) * | 2020-10-28 | 2023-12-26 | 江苏旗阵信息技术服务有限公司 | Power grid equipment model slicing method based on octree |
-
2009
- 2009-05-27 CN CN2009100721381A patent/CN101561937B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN101561937A (en) | 2009-10-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101561937B (en) | Big data size medical image three-dimensional interaction method based on common computer | |
| CN1716317B (en) | Sliding texture volume rendering | |
| CN104574263A (en) | Quick three-dimensional ultrasonic reconstruction and display method on basis of GPU (graphics processing unit) | |
| CN102096939B (en) | Medical mass data-oriented multi-resolution volume rendering method | |
| Tran et al. | A research on 3D model construction from 2D DICOM | |
| CN105825471A (en) | Unity-3D-based three-dimensional surface reconstruction and rendering method | |
| EP2826023B1 (en) | Systems and methods for creating a three-dimensional texture atlas | |
| CN101996415B (en) | Three-dimensional modeling method for eyeball | |
| Sobierajski et al. | A fast display method for volumetric data | |
| US7307630B2 (en) | Volume rendering apparatus and method | |
| Ciżnicki et al. | Efficient isosurface extraction using marching tetrahedra and histogram pyramids on multiple GPUs | |
| EP2266457B1 (en) | Intermediate image generating method, device, and program | |
| US6396494B1 (en) | Method for virtual clipping a three-dimensional graphics image | |
| CN1632834A (en) | Point reconstruction based very large-scale medical image 3D visualization method | |
| Kaufman et al. | A survey of architectures for volume rendering | |
| CN103745495A (en) | Medical volume data based volume rendering method | |
| CN103377480A (en) | Medical image imaging system | |
| US20120313942A1 (en) | System and method for digital volume processing with gpu accelerations | |
| Xue et al. | Efficient volume rendering methods for out-of-Core datasets by semi-adaptive partitioning | |
| Dai et al. | Volume-rendering-based interactive 3D measurement for quantitative analysis of 3D medical images | |
| CN102074051A (en) | Fast locating method for data bonding volume of volume rendering parallel projection | |
| Zhao et al. | High-performance and real-time volume rendering in CUDA | |
| CN106384377B (en) | Method and device for volume rendering of medical data | |
| CN110738719A (en) | Web3D model rendering method based on visual range hierarchical optimization | |
| Çalışkan et al. | Overview of Computer Graphics and algorithms |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20171120 Address after: Sonoyama street Longgang Avenue in Longgang District of Shenzhen City, Guangdong province 518116 No. 8288 Shenzhen Dayun town 13 software building 3 floor B District Patentee after: Shenzhen pulse medical robot Co., Ltd. Address before: 150001 Harbin, Nangang, West District, large straight street, No. 92 Patentee before: Harbin Institute of Technology |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20180629 Address after: 213023 Changzhou tower building, No. 213 ERON South Road, Zhong Lou District, Changzhou, Jiangsu 9731 Patentee after: Changzhou Mai Kang Yi medical robot Co., Ltd. Address before: 518116 Longgang road 8288, Yuen Shan Street, Longgang District, Shenzhen, Guangdong, 3, B, 13 building, Shenzhen Dayun software town. Patentee before: Shenzhen pulse medical robot Co., Ltd. |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20211222 Address after: 311500 Room 301, building 1, phase II, electronic equipment Industrial Park, 368 baiyunyuan East Road, Tonglu Economic Development Zone, Tonglu County, Hangzhou City, Zhejiang Province Patentee after: Hangzhou kangzhe Medical Instrument Co.,Ltd. Address before: 213023 Changzhou tower building, No. 213 ERON South Road, Zhong Lou District, Changzhou, Jiangsu 9731 Patentee before: Changzhou Mai Kang Yi medical robot Co.,Ltd. |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20220610 Address after: Room 701-6, building 3, No. 371, Mingxing Road, Xiaoshan District, Hangzhou, Zhejiang 311202 Patentee after: Hangzhou Weijing medical robot Co.,Ltd. Address before: 311500 Room 301, building 1, phase II, electronic equipment Industrial Park, 368 baiyunyuan East Road, Tonglu Economic Development Zone, Tonglu County, Hangzhou City, Zhejiang Province Patentee before: Hangzhou kangzhe Medical Instrument Co.,Ltd. |