WO2012130132A1 - 脊柱椎体和椎间盘分割方法、装置、磁共振成像系统 - Google Patents

脊柱椎体和椎间盘分割方法、装置、磁共振成像系统 Download PDF

Info

Publication number
WO2012130132A1
WO2012130132A1 PCT/CN2012/073131 CN2012073131W WO2012130132A1 WO 2012130132 A1 WO2012130132 A1 WO 2012130132A1 CN 2012073131 W CN2012073131 W CN 2012073131W WO 2012130132 A1 WO2012130132 A1 WO 2012130132A1
Authority
WO
WIPO (PCT)
Prior art keywords
pusher
region
boundary
growth
point
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.)
Ceased
Application number
PCT/CN2012/073131
Other languages
English (en)
French (fr)
Inventor
刘炎
邓晓云
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Mindray Bio Medical Electronics Co Ltd
Original Assignee
Shenzhen Mindray Bio Medical Electronics Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shenzhen Mindray Bio Medical Electronics Co Ltd filed Critical Shenzhen Mindray Bio Medical Electronics Co Ltd
Publication of WO2012130132A1 publication Critical patent/WO2012130132A1/zh
Priority to US14/041,919 priority Critical patent/US9445744B2/en
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/05—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves
    • A61B5/055—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/70—Determining position or orientation of objects or cameras
    • G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/45—For evaluating or diagnosing the musculoskeletal system or teeth
    • A61B5/4538—Evaluating a particular part of the muscoloskeletal system or a particular medical condition
    • A61B5/4566—Evaluating the spine
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10072—Tomographic images
    • G06T2207/10088—Magnetic resonance imaging [MRI]
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20112—Image segmentation details
    • G06T2207/20156—Automatic seed setting
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30008—Bone
    • G06T2207/30012—Spine; Backbone
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30172—Centreline of tubular or elongated structure

Definitions

  • the present invention relates to a magnetic resonance imaging apparatus, and more particularly to a method and apparatus for performing a spinal pusher and a pusher disc split using a magnetic resonance image. Background technique
  • MRI Magnetic NMR Resonance Imaging
  • a typical MRI spine disc scan process is described as: A physician or technician places each set of scan lines on a sagittal disc on a sagittal location image to ensure that the line passes through the center of the pushpad. The position and angle of the line set need to be adjusted repeatedly, and the line group placement and adjustment process is complicated and time consuming. Therefore, if the automatic extraction of the MRI spine disc can be realized, the intelligent scanning positioning of the push disc can be realized, and the automatic extraction of the push disc can adopt an image processing method to automatically segment or recognize the push or push in the sagittal image. plate.
  • the conventional image segmentation method is used to segment the pushpad, it is necessary to use the difference between the pushpad and the pusher grayscale.
  • the T1 weight image is white on the pusher and the pushpad is black, but this feature is Not absolutely, due to the influence of different weights, the gray scale difference between the pusher and the pusher disc is often not obvious.
  • the T2 weight or STIR weight image even the reversed push tray appears, so that the traditional method is difficult to fully apply.
  • the extraction of the push body has a method of matching with the deformable model. The problem with this method is that the number of visible pushes in the spine image is not necessarily the same. If the visible pusher is too much or too little, the match will be invalid.
  • the main technical problem to be solved by the present invention is to provide a spine pusher extraction method and a pusher disc splitting method and apparatus therefor.
  • a spine pusher extraction method including:
  • the pusher region is extracted by the seed region growth method based on the seed point inside the pusher.
  • a spinal pusher extracting apparatus comprising: a spinal cord positioning unit for positioning a spinal cord using magnetic resonance image data; an axial section determining unit for determining a pusher axis section according to a spinal cord
  • the seed point positioning unit is configured to position the inner seed point of the push body according to the cross section of the push body; and the push body area extracting unit is configured to extract the push body area by using the seed region growth method based on the internal seed point of the pusher.
  • a spinal interposer segmentation method comprising: locating a spinal cord using magnetic resonance image data;
  • the center points of the opposite sides of the pusher disc are respectively calculated by using the vertices of the adjacent two pusher regions; the center line of the pusher disc is determined according to the center points of the opposite sides.
  • a spinal interposer dividing device comprising: a spinal cord positioning unit for positioning a spinal cord using magnetic resonance image data; an axial section determining unit for determining a pusher axis according to the spinal cord a seed point positioning unit, configured to position a seed point inside the push body according to a cross section of the push body; a push body area extracting unit, configured to extract a push body area by using a seed region growth method based on a seed point inside the push body; a unit, configured to locate a vertex of the pusher region on the extracted pusher region; a pushpad center point positioning unit, configured to calculate a center point of the opposite sides of the pushpad by using the vertices of the adjacent two pusher regions
  • the push tray center line determining unit is configured to determine the center line of the push tray according to the center point of the two opposite sides.
  • the present invention also provides a magnetic resonance imaging system comprising the above-described spinal interverter disc dividing device or spinal pusher extracting device.
  • the invention utilizes the characteristics of the gray scale approximation hook in the push body, the obvious boundary between the push body and the push pad, and the shape of the push body is basically fixed, and the push and the push disk are adaptively extracted, and the extraction result is not affected by the push object or the push pad gray scale.
  • the influence of inconsistency is not affected by the weight of the image. It can adaptively extract all the pushes in the image, and then calculate the position and angle of the midline of each pusher. It can be effectively applied to the MRI (magnetic resonance image) spine disc. Fully automated positioning of the scan. DRAWINGS
  • FIG. 1 is a schematic structural view of a magnetic resonance imaging system in an embodiment
  • FIG. 2 is a schematic structural view of a spinal pusher extraction device according to an embodiment of the present invention
  • FIG. 3 is a schematic structural view of a pusher region extraction unit according to an embodiment of the present invention
  • Figure 5 is a T1 weight sagittal spine image
  • FIG. 6 is a flow chart of positioning a spinal cord in an embodiment of the present invention.
  • FIG. 7 is a schematic diagram of a set of maxima/minimum points detected after obtaining a first derivative
  • FIG. 8 is a schematic diagram of a boundary of a body surface detected in an embodiment of the present invention
  • FIG. 9 is a schematic diagram of a result of extracting a maximum value point after obtaining a second derivative in an embodiment of the present invention.
  • FIG. 10 is a schematic view of a spinal cord obtained after screening according to an embodiment of the present invention.
  • Figure 11 is a gray cross-sectional view of a pusher axis obtained in an embodiment of the present invention.
  • FIG. 12 is a flow chart of positioning a seed point inside a pusher in an embodiment of the present invention.
  • FIG. 13 is a schematic diagram showing the distribution of seed point points of a pusher body obtained in an embodiment of the present invention
  • FIG. 14 is a schematic diagram showing distribution of seed points after screening and moving according to an embodiment of the present invention
  • 16 is a schematic diagram of extracting a pusher region in an embodiment of the present invention.
  • Figure 17 is a flow chart showing the division of the spine disc in an embodiment of the present invention.
  • FIG. 18 is a schematic diagram of a center line of a push tray extracted in an embodiment of the present invention. detailed description
  • magnetic resonance imaging system 100 includes a magnet system 110 , a gradient magnetic field system 120 , a radio frequency system 130 , and a control and processing system 140 .
  • the magnet system 110 includes a magnet 111, a gradient magnetic field line 112, a transmission line 113, and a receiving line 114.
  • the magnet 111 may employ a permanent magnet or a normal conducting magnet for providing a constant main magnetic field to an object to be measured, such as a patient.
  • the gradient magnetic field line 112 is used to generate a gradient magnetic field in a three-dimensional space, and the emission line 113 is used to provide a radio frequency F) pulse to excite the spin of the atomic core in the object to be tested, and the receiving line 114 It is used to detect the echo signal emitted by the object to be tested.
  • the gradient magnetic field system 120 is coupled to the control and processing system 140 for driving the gradient magnetic field lines 112 under the control of the control and processing system.
  • the RF system 130 is coupled to the control and processing system 140 for generating RF pulses under control of the control and processing system and applying them to the transmit line 113 after amplification.
  • the control and processing system 140 is used to control both the various components and the echo signals.
  • the echo signal detected by the receiving coil 114 is transmitted to the control and processing system 140, which includes a spinal pusher extraction device for imaging the magnetic resonance image based on the obtained magnetic resonance image. Extract the spine pusher of the test object.
  • the spinal pusher extraction device 200 includes a spinal cord positioning unit 210, an axial section determining unit 220, a seed point positioning unit 230, and a pusher region extracting unit 240.
  • the spinal cord positioning unit 210 is configured to locate the spinal cord by using magnetic resonance image data
  • the axial section determining unit 220 is configured to determine a pusher axis section according to the spinal cord
  • the seed point positioning unit 230 is configured to position the internal seed point of the pusher according to the pusher axis section.
  • the pusher region extracting unit 240 is configured to extract the pusher region by using the seed region growing method based on the internal seed point of the pusher.
  • the spinal cord positioning unit 210 uses the feature that the gray scale of the pusher body is approximately uniform, and extracts the boundary between the two sides of the measured object (for example, the human torso), and detects the spinal cord according to the distance defined by the boundary between the two sides of the body surface. line.
  • the spinal cord positioning unit 21G includes a boundary detecting subunit 211 and a spinal cord detecting subunit 212.
  • the boundary detection sub-unit 211 is configured to detect the boundary surface of the two sides of the magnetic resonance image of the test subject by using the transition characteristics of the tissue and the background of the test subject, and the spinal cord detection sub-unit 212 is configured to measure the width of the measured subject according to the boundary of the body surface. The spinal cord is detected.
  • the axis section determining unit 220 obtains the pusher axis by using the feature that the spinal cord and the pusher axis are aligned and the distance between the two is very close, and the image of the pusher axis cross section is obtained according to the pusher axis, and then the seed
  • the point positioning unit 230 uses the feature that the pusher and the pusher disc have obvious boundaries, and positions the seed point inside the pusher according to the cross section of the pusher axis.
  • the axis section determining unit 220 includes a translation subunit 221 and a weighting subunit 222.
  • the translation subunit 221 is configured to translate the spinal cord according to a plurality of set translation distances to obtain a plurality of pusher axes, and the weighting subunit 222 is used for multiple pairs.
  • the gray scales of the pusher axes are weighted and summed to obtain the cross section of the pusher axis.
  • the seed point locating unit 230 includes a boundary point detecting sub-unit 231, a screening sub-unit 232, and a moving sub-unit 233.
  • the boundary point detecting sub-unit 231 is configured to calculate a gray-scale variation gradient of the cross-section of the pusher axis, and use a point whose gradient value is greater than a set threshold as a candidate boundary point of the pusher and the adjacent pusher disk.
  • Filter subunit 232 is used to detect The candidate boundary points are filtered according to a predetermined rule, for example, a third predetermined rule mentioned later, to obtain a boundary point between the push body and the adjacent push pad.
  • the movement subunit 233 is configured to move the coordinates of the boundary point in the direction of the axis of the push body by a set size, and the point corresponding to the obtained new coordinate is recorded as a seed point inside the pusher.
  • the pusher region extracting unit 240 obtains the pusher region by the region growing method by using the feature in which the pusher shape is substantially fixed, in combination with the area and shape of the pusher.
  • the push region extracting unit 240 includes an initial subunit 241, a region growing subunit 242, a calculating subunit 243, a judging subunit 244, a lookup and extraction subunit 245, and a growth threshold transform subunit 246.
  • the initial sub-unit 241 is used to set an initial threshold, a growth target minimum area, and a growth target maximum area, and assign an initial threshold to the growth threshold.
  • the region growing subunit 242 unit 243 is for calculating the area perimeter ratio of the filled area, estimating the desired area perimeter ratio based on the area of the filled area, and calculating the difference between the area perimeter ratio and the desired area perimeter ratio.
  • the decision subunit 244 is for determining whether the growth satisfies the predetermined condition.
  • the find and extract subunit 245 is configured to control the stop growth when the growth satisfies the predetermined condition, and find the smallest difference in the difference of the area of the filled area between the minimum area of the growth target and the maximum area of the growth target, which is the minimum
  • the growth threshold corresponding to the difference is recorded as an optimal growth threshold, and the filled region corresponding to the minimum difference is extracted as a push region.
  • the growth threshold transform sub-unit 246 is configured to reduce the growth threshold according to a fourth predetermined rule when the growth is not satisfied, to generate a new growth threshold, and then the control region growth sub-unit 242 performs region growth based on the new growth threshold.
  • the push region extraction unit 240 further includes a region analysis sub-unit 247 for analyzing the extracted push region, and removing the redundant region according to the fifth predetermined rule. The rest.
  • a spinal pusher extraction method is as shown in FIG. 4, and includes the following steps: Step S1, using magnetic resonance image data to locate the spinal cord;
  • Step S2 determining a cross section of the push body axis according to the spinal cord
  • Step S3 positioning the seed point inside the push body according to the cross section of the push body axis
  • Step S4 extracting the pusher region by using the seed region growth method based on the seed point inside the pusher.
  • the spinal cord is positioned using the features of the grayscale approximation in the pusher.
  • the spinal cord is an important feature of the spine image because its gray scale is relatively uniform and for T1 weight
  • T2 weight or STIR weight image the difference in brightness between the spinal cord and the surrounding tissue is basically stable.
  • the T1 weight image is darker than the surrounding tissue.
  • the T2 image is brighter than the surrounding tissue.
  • the left side of the figure is the front of the subject, and the right side of the figure is the back of the subject.
  • step S1 is as shown in FIG. 6, and includes the following steps:
  • Step S11 detecting the boundary between the two sides of the magnetic resonance image of the test subject by using the transition characteristics of the tissue and the background of the test subject.
  • the extraction of body surface boundaries utilizes the grayscale transition characteristics of human tissue and background tissue. For example, as shown in Figure 5, the left side of the body is a transition from black to white, and the right side of the body is a transition from white to black.
  • This transition can be detected using the first derivative, which can be detected by the first-order horizontal derivative extrema because the body surface boundary is roughly vertical.
  • the first derivative is obtained in a direction perpendicular to the boundary direction of the body surface (for example, the horizontal direction), and the first derivative can be calculated as follows:
  • Dx dx where / is the grayscale data of the input image, indicating the J direction (horizontal direction) first order dx
  • G is a Gaussian template
  • * is a convolution.
  • the maximum and minimum values of the first derivative in the horizontal direction are detected respectively.
  • the position of the first derivative is transformed along the boundary direction of the body surface, and the above steps are repeated to detect a plurality of maximum value points and minimum value points, and some of the maximum value points are connected into a line to form a set of maximum value points, some minimum values.
  • the points are connected into a line to form a set of minimum value points.
  • the maximum/minimum point image is shown in Fig. 7, where the white point is the maximum value and the gray point is the minimum value.
  • the maximum value point set is used for detecting the first side body table boundary
  • the minimum value point set is used for detecting the second side body table. boundary.
  • the plurality of maxi point sets and minima point sets are filtered according to a first predetermined rule, and the first predetermined rule may be according to the length of the line, the line. The value of each point and the distance between the line and the boundary of the image are considered together.
  • the surface boundary of the first side for example, the left side
  • the surface of the second side for example, the right side
  • step S12 After the first predetermined rule is filtered, only one left boundary and one right boundary are left, and after obtaining the boundary between the two sides, step S12 is performed.
  • the first predetermined rule is in addition to the rule disclosed in this embodiment. In addition, it can be other rules, as long as the boundary between the two sides is selected in many boundaries.
  • the spinal cord is detected according to the width of the measured subject defined by the boundary of the two sides of the body surface (ie, the distance between the boundary surfaces of the two sides).
  • the spinal cord has a narrow narrow-band shape in the image, and its gray scale is in sharp contrast with its left and right.
  • the T1 weight image has a prominent black line
  • the T2 weight or STIR weight image has a prominent white line.
  • the second derivative can be used to detect the spinal cord.
  • the second derivative is sensitive to the linear structure, but it is necessary to specify an appropriate filtering scale, which can be effectively detected only when the filtering scale matches the width of the spinal cord.
  • the filtering scale can be estimated according to the distance between the first side body table boundary and the second side body table boundary, and the second derivative of the magnetic resonance image is obtained by using the estimated filtering scale;
  • the second derivative calculation expression is as follows : Where ⁇ is the standard deviation of the Gaussian template, and the filter scale is estimated according to the distance between the boundary of the first side body surface and the boundary of the second side body table; 4 is the average distance between the boundary surfaces of the two sides, and ⁇ is a proportional constant, which can be Experiment confirmed.
  • the extreme value of the image is still taken, that is, the extreme value of the second derivative image is detected. If it is the T1 weight image, the maximum value is taken, and if it is the ⁇ 2 weight or the STI R weight image, the minimum value is taken. The value, thus obtaining a set of extreme points connected by extreme points into a line, the result of the extreme point extraction is shown in Figure 9. After extracting the extreme points, boundary filtering is still required. Therefore, the plurality of extreme point sets are screened according to the second predetermined rule, and the second predetermined rule (ie, the screening rule) can be integrated according to factors such as the horizontal position of the spinal cord in the human tissue, the vertical position, the length of the spinal cord, and the like.
  • the horizontal position of the spinal cord is raised to the right and the vertical position, and then the length of the spinal cord is comprehensively considered, thereby screening
  • the spinal cord is obtained, as shown in Figure 10.
  • the second predetermined rule may be other rules in addition to the rules disclosed in the embodiment, as long as the spinal cord is screened out among a plurality of boundaries.
  • the method of locating the internal seed point of the pusher utilizes two features of a clear boundary between the pusher and the pusher disc and a certain distance between the pusher and the pusher.
  • the process of seed point location is as follows: extract the pusher axis section, calculate the pusher boundary point, select the boundary point, and move the boundary point.
  • the cross-section of the pusher axis is first determined according to the spinal cord, and the basic idea is: The pusher axis is determined according to the spinal cord, and then the gray scale of the pusher axis is calculated, and a vector diagram representing the gray scale of the pusher axis is formed, that is, the pusher axis cross section.
  • the axis of the pusher can be obtained by translating the obtained spinal cord through the single tube.
  • the translation distance can be set according to experience or according to the boundary of the two sides.
  • the defined width range is determined.
  • the pusher axis referred to herein is not limited to a line passing through the center of the pusher, as long as it can pass through all of the pushers in the longitudinal direction, thus allowing a certain positioning error.
  • a pusher axis can be obtained by translation, or multiple pusher axes can be obtained by multiple translations. The purpose of introducing multiple translations is to suppress the effects of noise, since it is easy to miss the push body by taking only one translation.
  • the extracted pusher axis section is a grayscale vector that expresses an approximate grayscale change along the axis of the pusher, as shown in Fig. 11, where the position where the grayscale changes drastically corresponds to the edge of the pusher and the pusher disc.
  • the method of locating the seed point inside the pusher is as shown in FIG. 12, and includes the following steps:
  • Step S31 calculating a gray level change gradient of the cross section of the push body axis, and using a point whose gradient value is greater than the set threshold value as a candidate boundary point of the push body and the adjacent push pad. Because the cross section exhibits a dramatic change at the pusher boundary point, the boundary point can be detected based on the gradient feature.
  • the boundary point extraction result is shown in Fig. 13 , where the black point is the boundary point.
  • Step S32 The detected candidate boundary points are filtered according to a third predetermined rule to obtain a boundary point between the push object and the adjacent push pad.
  • These detected boundary points have a lot of redundancy.
  • a boundary often detects multiple boundary points, which requires boundary point screening.
  • the distance characteristics of the boundary points can be used for screening so that the distance d between the last two adjacent boundaries satisfies: Among them, N 2 is an empirical value and can be selected according to experience.
  • the third predetermined rule may be other rules in addition to the above-mentioned selection rule in the embodiment.
  • the boundary points are extracted. The purpose is to obtain as few seed points as possible in the future. If the seed points in the push body are directly extracted by using the uniformity, a lot of candidate points are obtained, which makes the algorithm less efficient.
  • step S33 the coordinates of the boundary point are moved according to the set size in the direction of the axis of the pusher, and the point corresponding to the new coordinate obtained is recorded as the seed point inside the pusher.
  • the filtered boundary point is moved vertically upwards or downwards in order to move the position of the boundary point to the inside of the pusher.
  • the basis for this is:
  • the area of the pusher is larger than the area of the pusher disc, and the seed point of the pusher can be obtained by slightly moving the boundary point. Therefore, the distance traveled can be set empirically, for example, a value greater than the width of the upper and lower pads.
  • the gradient of the gray scale on the axis of the pusher is used, and the gradient represents the change of the gray scale, which is independent of the level of the gray scale itself, and can be reflected on the gradient as long as the gray scale changes.
  • the interpolated disc of the T1-weighted image is generally black, and the interpolated disc of the T2-weighted image is generally white, but this feature is sometimes not obvious, that is, the gray scale difference between the push bone and the push disc is sometimes Very small, if the seed point is directly extracted by other methods, it will always be interfered by this problem, and whether it is a T1-weighted image or a T2-weighted image, there is a clear boundary between the pushbone and the push-pull disk, which can be based on the gradient.
  • the region growth method using the adaptive threshold is used for the pusher region growth.
  • the growth threshold can be iteratively calculated using the area of the pusher and the shape of the pusher.
  • the area coordinates after growth satisfy: ⁇ ,. y ⁇ - I seedxj, seedy j Among them, " ⁇ , seedy ) is the coordinate of the jth seed, (x ; , y ⁇ ) is the isolated neighborhood set of e ; ⁇ , seedy j ), which is the number of seed points, and Hj is based on the jth The growth threshold of the seed.
  • the key here is to determine the optimal growth threshold Hj.
  • the region growing method extracting the pusher region includes the following steps:
  • Step S41 setting an initial threshold ⁇ ⁇ , a growth target minimum area ⁇ and a growth target maximum area and assigning an initial threshold ⁇ ⁇ to the growth threshold ⁇ ; wherein the initial threshold ⁇ ⁇ may be a relatively large fixed value, or may be a push axis
  • the maximum gray scale and the most d, the gray value difference is multiplied by a coefficient to obtain a value, and ⁇ can be used to estimate, such as SM 2 and ST 2, ⁇ and L 2 are empirically selected.
  • step S42 region growth is performed based on the growth threshold, and the grown region is subjected to region filling. The region growth in this step can be regionally grown and filled based on seed spots using existing techniques.
  • the area of the back area, P is the perimeter of the area after filling.
  • Step S44 estimating a desired area perimeter ratio based on the area of the filled area.
  • the shape of the pusher body it is basically fixed and has an approximately square shape. Assuming that the pusher shape is a square, the circumference is calculated according to the area S of the filled area, and then the area perimeter ratio is calculated again, that is, the desired area perimeter ratio is calculated.
  • SPR, SPR ⁇ .
  • Step S45 calculating the difference between the area perimeter ratio and the desired area perimeter ratio.
  • the difference between the area circumference ratio and the desired area circumference ratio may be the difference between the two, or may be the ratio of the two.
  • Step S46 determining whether the growth satisfies the predetermined condition, and the predetermined condition may be that the growth threshold H is already small and less than the set threshold.
  • the predetermined condition may also be that the area S of the filled area is already small, less than the set threshold. If the growth satisfies the predetermined condition, step S48 is performed to stop the growth based on the seed, and find the smallest difference in the difference of the area of the filled area between the minimum area of the growth target and the maximum area of the growth target, the minimum difference
  • the corresponding growth threshold is recorded as The optimal growth threshold, the post-filled region corresponding to the minimum difference is recorded as the push region based on the seed growth; otherwise, step S47 is performed;
  • Step S47 growing a threshold value.
  • the growth threshold is decreased in accordance with the fourth predetermined rule, and then the flow proceeds to step S42, where the region growth is performed based on the new growth threshold, and the cycle is performed until the growth satisfies the predetermined condition.
  • the fourth predetermined rule may be that the growth threshold is successively decreased according to the set step size, or the growth threshold may be decreased according to a set curve, and the growth threshold may be sequentially decreased irregularly or randomly.
  • the embodiment combines the shape information of the push body, and at the same time considers the pusher area, so that the grown area is most similar to the real pusher area.
  • the pusher region corresponding to the seed point is also obtained.
  • the region needs to be analyzed and judged, because the number of seed points obtained in the past still has redundancy, mainly in two cases:
  • the extracted pusher region is analyzed, and the redundant portion of the pusher region is removed in accordance with the fifth predetermined rule. Redundancy can be removed by all or part of the following information: The difference in SP and SPR after growth of each seed point is examined;
  • the position information of the area after the growth of the two adjacent seed points on the axis of the pusher is investigated.
  • the difference in the area of the two adjacent seed points on the axis of the pusher is investigated. This information can effectively solve the above two problems.
  • the pusher region is obtained as shown in Fig. 16.
  • control and processing system includes a spinal push disc dividing device
  • spinal push disc dividing device includes the spinal pusher extracting device, the pusher region vertex positioning unit, and the push tray center in any of the above embodiments.
  • Point positioning unit and push pad center line determining unit are configured to locate the vertices of the pusher region on the extracted pusher region, and the pusher disc center point locating unit is configured to calculate the opposite sides of the pusher disc by using the vertices of the adjacent two pusher regions respectively.
  • the center point of the push tray center line determining unit is configured to determine the center line of the push tray according to the center point of the two opposite sides.
  • the spine disc dividing method is as shown in Fig. 17, and includes the following steps:
  • Step S1 using the magnetic resonance image data to locate the spinal cord.
  • Step S2 determining a cross section of the pusher axis according to the spinal cord.
  • Step S3 positioning the seed point inside the pusher according to the cross section of the pusher axis.
  • Step S4 extracting the pusher region by using a seed region growth method based on the seed point inside the pusher.
  • Step S5 Positioning the vertices of the pusher region on the extracted pusher region. To locate the scan line of the push pad, you need to determine the expression of the center line of the push pad. In the result of the push area obtained in the previous step, you need to detect the coordinates of the four vertices of each pusher, and calculate the coordinates of the vertices of the adjacent two pushers. Push the center point of the disc.
  • the spiral scan of the cartridge can be used to extract the vertices of the pusher, and the four vertices of the pusher can be obtained by using different scan directions respectively.
  • step S6 the center points of the opposite sides of the pusher disc are respectively calculated by using the vertices of the adjacent two pusher regions.
  • the midpoint coordinates of the apex of the lower left corner of the previous pusher and the apex of the upper left corner of the latter pusher are the left center point of the pusher disc, and the midpoint coordinates of the vertices of the lower right corner of the previous pusher and the top right corner of the latter pusher are the pushroom The right center point of the disc.
  • Step S7 determining the center line of the push pad according to the center point of the two opposite sides.
  • the center line of the last extracted push tray is shown in Figure 18.
  • Steps S1 to S4 in this embodiment may adopt all or part of the steps provided by the present invention, or may be implemented in whole or in part using the prior art.
  • the invention utilizes the characteristics of the gray scale approximation hook in the push body, the obvious boundary between the push body and the push pad, and the shape of the push body are basically fixed, and the push and the push pad are adaptively extracted, and the extraction result is not subject to the push or push.
  • the influence of the disc gradation inconsistency is not affected by the image weight. It can adaptively extract all the push bodies in the image, and then calculate the midline position and angle of each push disc, which can be effectively applied to the MRI (magnetic resonance image) spine. Fully automatic positioning of the push tray scan.
  • the invention is not affected by the field strength and can be applied to any field strength MRI imaging system.
  • the above describes how to extract the spinal pusher or perform the spine disc split based on the magnetic resonance image, but when the image of the measured object is obtained by other means, the spine push or the method can also be extracted according to the method and/or device provided by the present invention. Perform a spine disc split.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Heart & Thoracic Surgery (AREA)
  • General Health & Medical Sciences (AREA)
  • Pathology (AREA)
  • Medical Informatics (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Biomedical Technology (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Biophysics (AREA)
  • Radiology & Medical Imaging (AREA)
  • High Energy & Nuclear Physics (AREA)
  • Magnetic Resonance Imaging Apparatus (AREA)

Description

脊柱推体和推间盘分割方法、 装置、 磁共振成像系统 技术领域
本发明涉及一种磁共振成像设备， 尤其涉及采用磁共振图像进行脊柱 推体和推间盘分割的方法、 装置。 背景技术
MRI ( Magnet ic resonance imaging, 磁共振成像）检查由于其无损伤、 任意断面和多参数成像等特点而日益普及， 尤其在中枢神经和脊柱临床应 用中优势更为突出。 典型的 MRI 脊柱推间盘扫描过程描述为： 由医师或技 师在矢状面定位像上将每一组扫描线放置在有病变的推间盘上， 为保证线 组穿过推间盘中心， 需要反复调整线组的位置和角度， 线组放置和调整过 程繁复而费时。 因此如果能实现 MRI 脊柱推间盘自动提取， 就可以实现推 间盘的智能扫描定位， 推间盘自动提取可采用图像处理方法， 在矢状面图 像中自动分割或者识别出推体或者推间盘。 如果采用传统的图像分割方法 对推间盘进行分割， 就必须利用推间盘与推体灰度的差异性， 如 T1权重图 像上推体呈现白色， 推间盘呈现黑色， 但这种特征并不绝对， 由于不同权 重的影响， 推体和推间盘的灰度差异往往不明显， 在 T2权重或 STIR权重 图像上甚至出现反色的推间盘， 以至传统方法难以充分应用。 当然， 推体 的提取有利用可变形模型匹配的方法， 这种方法的问题在于脊柱图像中可 见推体数量并不一定相同， 如果可见的推体过多或者过少， 都会导致匹配 失效。 还有的分割方法需要医生或者技师进行一定的交互操作， 如选择特 征点等， 也影响了效率。 发明内容
本发明要解决的主要技术问题是， 提供一种脊柱推体提取方法和推间 盘分割方法及其装置。
根据本发明的一方面， 提供一种脊柱推体提取方法， 包括：
利用磁共振图像数据定位脊髓线；
根据脊髓线确定推体轴线截面； 根据推体轴线截面定位推体内部种子点；
基于推体内部种子点采用种子区域生长法提取推体区域。
根据本发明的另一方面， 提供一种脊柱推体提取装置， 包括： 脊髓线 定位单元， 用于利用磁共振图像数据定位脊髓线； 轴线截面确定单元， 用 于根据脊髓线确定推体轴线截面； 种子点定位单元， 用于根据推体轴线截 面定位推体内部种子点； 推体区域提取单元， 用于基于推体内部种子点采 用种子区域生长法提取推体区域。
根据本发明的另一方面， 提供一种脊柱推间盘分割方法，包括： 利用磁共振图像数据定位脊髓线；
根据脊髓线确定推体轴线截面；
根据推体轴线截面定位推体内部种子点；
基于推体内部种子点采用种子区域生长法提取推体区域；
在提取的推体区域上定位该推体区域的顶点；
利用相邻两个推体区域的顶点分别计算推间盘两相对边的中心点； 根据两相对边的中心点确定推间盘中心线。
根据本发明的又一方面， 提供一种脊柱推间盘分割装置，包括： 脊髓线 定位单元， 用于利用磁共振图像数据定位脊髓线； 轴线截面确定单元， 用 于根据脊髓线确定推体轴线截面； 种子点定位单元， 用于根据推体轴线截 面定位推体内部种子点； 推体区域提取单元， 用于基于推体内部种子点采 用种子区域生长法提取推体区域； 推体区域顶点定位单元， 用于在提取的 推体区域上定位该推体区域的顶点； 推间盘中心点定位单元， 用于利用相 邻两个推体区域的顶点分别计算推间盘两相对边的中心点； 推间盘中心线 确定单元， 用于根据两相对边的中心点确定推间盘中心线。
本发明还提供一种包括上述脊柱推间盘分割装置或脊柱推体提取装置 的磁共振成像系统。
本发明利用推体内灰度近似均勾、 推体与推间盘有明显边界、 推体形 状基本固定等特征自适应提取推体和推间盘， 提取结果不受推体或者推间 盘灰度不一致的影响， 也不受图像权重的影响， 可自适应提取图像中的全 部推体， 进而计算出各个推间盘的中线位置和角度， 能有效应用于 MRI (磁 共振图像）脊柱推间盘扫描的全自动定位。 附图说明
图 1为一种实施例中磁共振成像系统的结构示意图；
图 2为本发明一种实施例中脊柱推体提取装置的结构示意图； 图 3为本发明一种实施例中推体区域提取单元的结构示意图； 图 4为本发明一种实施例中脊柱推体提取的流程图；
图 5为 T1权重矢状面脊柱图像；
图 6为本发明一种实施例中定位脊髓线的流程图；
图 7为经求一阶导数后检测出的极大值 /极小值点集合示意图； 图 8为本发明一种实施例中检测出的体表边界示意图；
图 9 为本发明一种实施例中经过求二阶导数后极大值点提取结果示意 图；
图 10为本发明一种实施例中筛选后得到的脊髓线示意图；
图 11为本发明一种实施例中得到的推体轴线灰度截面图；
图 12为本发明一种实施例中定位推体内部种子点的流程图；
图 13为本发明一种实施例中得到的推体轴线种子点分布示意图； 图 14为本发明一种实施例中筛选移动后的种子点分布示意图； 图 15为本发明一种实施例中提取推体区域的流程图；
图 16为本发明一种实施例中提取推体区域示意图；
图 17为本发明一种实施例中脊柱推间盘分割的流程图；
图 18为本发明一种实施例中提取的推间盘中心线示意图。 具体实施方式
下面通过具体实施方式结合附图对本发明作进一步详细说明。
请参考图 1 ,在一种实施例中，磁共振成像系统 100包括磁体系统 110、 梯度磁场系统 120、 射频系统 130和控制及处理系统 140。 磁体系统 110包 括磁体 111、 梯度磁场线圏 112、 发射线圏 113和接收线圏 114 , 磁体 111 可以采用永磁体或常导磁体， 用于给待测物体（例如病人）提供一恒定的 主磁场，梯度磁场线圏 112用于在三维空间产生一梯度磁场，发射线圏 113 用于提供射频 F )脉沖以激发待测物体内原子核的自旋， 接收线圏 114 用于检测由待测物发出的回波信号。 梯度磁场系统 120和控制及处理系统 140连接， 用于在控制及处理系统的控制下驱动梯度磁场线圏 112。 射频系 统 130和控制及处理系统 140连接， 用于在控制及处理系统的控制下产生 RF脉沖并经放大处理后施加给发射线圏 113。
控制及处理系统 140 既用于对各部分进行控制， 也用于对回波信号进 行处理。将接收线圏 114检测到的回波信号传输到控制及处理系统 140 ,控 制及处理系统 140 包括脊柱推体提取装置， 脊柱推体提取装置用于基于得 到的磁共振图像， 在磁共振图像上提取出被测物的脊柱推体。
在如图 2所示的一种实施例中， 脊柱推体提取装置 200 包括脊髓线定 位单元 210、 轴线截面确定单元 220、 种子点定位单元 230和推体区域提取 单元 240。脊髓线定位单元 210用于利用磁共振图像数据定位脊髓线， 轴线 截面确定单元 220用于根据脊髓线确定推体轴线截面，种子点定位单元 230 用于根据推体轴线截面定位推体内部种子点， 推体区域提取单元 240用于 基于推体内部种子点采用种子区域生长法提取推体区域。
在一种具体实例中， 脊髓线定位单元 210 利用推体内灰度近似均匀的 特征， 采用提取被测物 （例如人体躯干） 两侧体表边界， 根据两侧体表边 界界定的距离检测出脊髓线。脊髓线定位单元 21 G包括边界检测子单元 211 和脊髓线检测子单元 212。边界检测子单元 211用于利用被测者组织和背景 的过渡特性检测被测者磁共振图像的两侧体表边界，脊髓线检测子单元 212 用于根据体表边界界定的被测者组织宽度检测脊髓线。
在另一种具体实例中， 轴线截面确定单元 220 利用脊髓线和推体轴线 走向一致且两者距离很近的特征先得到推体轴线， 根据推体轴线得到推体 轴线截面的图像， 然后种子点定位单元 230 利用推体与推间盘有明显边界 的特征，根据推体轴线截面定位出推体内部种子点。 轴线截面确定单元 220 包括平移子单元 221和加权子单元 222 ,平移子单元 221用于将脊髓线按照 多个设定的平移距离平移后得到多个推体轴线， 加权子单元 222 用于对多 个推体轴线的灰度求加权和， 得到推体轴线截面。 种子点定位单元 230 包 括边界点检测子单元 231、 筛选子单元 232和移动子单元 233。 边界点检测 子单元 231 用于计算推体轴线截面的灰度变化梯度， 将梯度值大于设定阈 值的点作为推体与相邻推间盘的候选边界点。 筛选子单元 232 用于将检测 出的候选边界点按照一预定规则， 例如后文提到的第三预定规则进行筛选， 得到推体与相邻推间盘的边界点。 移动子单元 233用于将边界点的坐标沿 推体轴线方向按照设定大小移动， 得到的新坐标所对应的点记为推体内部 的种子点。
在又一种具体实例中， 推体区域提取单元 240 利用推体形状基本固定 的特征， 结合推体的区域面积和形状， 采用区域生长法得到推体区域。 推 体区域提取单元 240如图 3所示， 包括初始子单元 241、 区域生长子单元 242、 计算子单元 243、 判断子单元 244、 查找及提取子单元 245和生长阈 值变换子单元 246。初始子单元 241用于设置初始阈值、生长目标最小面积 和生长目标最大面积， 并将初始阈值赋予生长阈值。 区域生长子单元 242 单元 243用于计算填充后区域的面积周长比， 根据填充后区域的面积推算 出期望的面积周长比， 并计算面积周长比和期望的面积周长比的差异。 判 断子单元 244用于判断生长是否满足预定条件。 查找及提取子单元 245用 于在生长满足预定条件时， 控制停止生长， 并在位于生长目标最小面积和 生长目标最大面积之间的填充后区域面积的差异中查找出最小差异， 将所 述最小差异所对应的生长阈值记为最佳生长阈值， 将所述最小差异所对应 的填充后区域提取为推体区域。 生长阈值变换子单元 246 用于在生长不满 足预定条件时按照第四预定规则减小生长阈值， 生成新的生长阈值， 然后 控制区域生长子单元 242基于新的生长阈值进行区域生长。
在推体区域提取单元 240 的另一具体实例中， 推体区域提取单元 240 还包括区域分析子单元 247 , 用于对提取的推体区域进行分析，按照第五预 定规则去除推体区域的冗余部分。
基于以上装置， 一种脊柱推体提取方法如图 4所示， 包括以下步骤： 步骤 S1 , 利用磁共振图像数据定位脊髓线；
步骤 S2 , 根据脊髓线确定推体轴线截面；
步骤 S3 , 根据推体轴线截面定位推体内部种子点；
步骤 S4 , 基于推体内部种子点采用种子区域生长法提取推体区域。 在一种具体实例中， 利用推体内灰度近似均勾的特征对脊髓线进行定 位。 脊髓线是脊柱图像的重要特征， 因为其灰度比较均匀， 而且对于 T1权 重、 T2权重或 STIR权重图像来说， 脊髓与周围组织的亮度差异基本上是稳 定的， T1权重图像脊髓比周围组织暗， 如图 5所示， T2图像脊髓比周围组 织亮。 在图 5中， 图的左侧为被测者的前面， 图的右侧为被测者的后面。
对于脊髓线为大致呈现竖直方向的情况， 提取脊髓线的过程为： 提取 被测物第一侧（例如左侧）体表边界、 提取第二侧（例如右侧）体表边界、 提取脊髓线。 因此本实施例中， 步骤 S1如图 6所示， 包括以下步骤：
步骤 S11 ,利用被测者组织和背景的过渡特性检测被测者磁共振图像的 两侧体表边界。 体表边界的提取利用人体组织与背景组织的灰度过渡特性。 例如图 5所示， 左侧体表为由黑到白的过渡， 右侧体表为由白到黑的过渡。 利用一阶导数可检测这种过渡， 由于体表边界大致为垂直走向， 可用一阶 水平导数极值来检测。 在与体表边界方向垂直的方向 （例如水平方向）上 求一阶导数， 一阶导数可按如下表达式计算：
dl τ dG
—— * G = I *
dx dx 其中， /为输入图像的灰度数据， 表示 J方向 （水平方向）一阶导 dx
数， G为高斯模板， *为卷积。 选择合适的高斯模版计算一阶导数后， 分别检测出水平方向的一阶导 数的极大值点和极小值点。沿体表边界方向变换求一阶导数的位置，循环上 述步骤， 检测出很多极大值点和极小值点， 一些极大值点连接成线， 形成 极大值点集合， 一些极小值点连接成线， 形成极小值点集合。 极大值 /极小 值点图像如图 7 , 其中白色点为极大值， 灰色点为极小值。 可以看出， 图像 中形成了多个极大值点集合和极小值点集合， 极大值点集合用于检测第一 侧体表边界， 极小值点集合用于检测第二侧体表边界。 因存在过多的极大 值 /极小值点集合， 因此对多个极大值点集合和极小值点集合按照第一预定 规则进行筛选， 第一预定规则可以是按照线的长度、 线上各点导数值以及 线与图像边界的距离来综合考虑， 筛选后得到第一侧 （例如左侧）体表边 界和第二侧 （例如右侧）体表边界， 如图 8 所示， 经第一预定规则筛选后 只留下一条左侧边界和一条右侧边界， 得到两侧体表边界后执行步骤 S12。 当然， 本领域技术人员应该理解， 第一预定规则除了本实施例中公开的规 则外， 还可以是其他规则， 只要实现在众多边界中筛选出两侧边界即可。 步骤 S12 ,根据两侧体表边界界定的被测者组织宽度（即两侧体表边界 的距离）检测脊髓线。 脊髓线在图像中呈狭长的窄带形状， 其灰度与其左 右形成鲜明对比， 如 T1权重图像脊髓线为突出的黑色， T2权重或 STIR权 重图像脊髓线为突出的白色。 利用这个特征， 可用二阶导数来检测脊髓线。 二阶导数对线状结构敏感， 但需要指定合适的滤波尺度， 只有当滤波尺度 与脊髓线宽度相匹配时才能有效检测。 因为已经提取出体表边界， 可根据 第一侧体表边界和第二侧体表边界的距离估算滤波尺度， 采用估算的滤波 尺度对磁共振图像求二阶导数； 二阶导数计算表达式如下：
Figure imgf000009_0001
其中， σ为高斯模板标准差， 根据第一侧体表边界和第二侧体表边界 的距离估算， 体现滤波尺度； 4为两侧体表边界的平均距离， Μ为一比例 常数， 可依实验确定。
通过求二级导数滤波后仍然对图像取极值， 即检测二阶导数图像的极 值， 如果为 T1权重图像， 则取极大值， 如果为 Τ2权重或 STI R权重图像， 则取极小值， 从而得到多个由极值点连接成线的极值点集合， 极值点提取 结果如图 9 所示。 提取极值点后， 仍然需要进行边界筛选。 因此对多个极 值点集合按照第二预定规则进行筛选， 此时的第二预定规则 （即筛选规则） 可按照脊髓线在人体组织中的水平位置、 垂直位置、 脊髓线的长度等因素 综合考虑， 例如， 本实施例中， 根据从被测者侧面获取图像， 如图 5所示， 脊髓线的水平位置靠右、 垂直位置靠上， 然后再综合考虑脊髓线的长度等 因素， 从而筛选后得到脊髓线， 如图 10所示。 第二预定规则除了本实施例 中公开的规则外， 还可以是其他规则， 只要实现在众多边界中筛选出脊髓 线即可。
在一种具体实例中， 定位推体内部种子点方法利用了推体与推间盘之 间有明确边界、 并且推体与推体之间有一定的距离间隔两个特征。 种子点 定位的过程为： 提取推体轴线截面、 计算推体边界点、 边界点 选、 边界 点移动。 本实施例中， 首先根据脊髓线确定推体轴线截面， 基本思路是： 根据脊髓线确定推体轴线， 然后计算推体轴线的灰度， 形成表征推体轴线 灰度的向量图， 即推体轴线截面。 因为推体轴线与脊髓线距离很近， 且两 者走向一致， 因此把前面得到的脊髓线通过筒单的平移即可得到推体轴线， 平移距离可根据经验设定或根据两侧体表边界限定的宽度范围确定。 这里 所说的推体轴线并不限定是从推体正中心穿过的线， 只要能够纵向穿过所 有的推体即可， 因此允许有一定的定位误差。 可以通过平移得到一条推体 轴线， 也可以通过多次平移得到多条推体轴线。 引入多个平移的目的是为 了抑制噪声的影响， 因为只取一个平移容易错过推体。 本实施例中， 以通 过多个设定的平移距离平移后得到多条推体轴线进而得到推体轴线截面为 例进行说明。 推体轴线截面可按照下面表达式进行计算， 它对多个推体轴 线的灰度求 推体轴线截面： spine (1) = ∑an = l
Figure imgf000010_0001
其中， "e(t)为推体轴线截面向量， ί为索引， (·ψ), _ν( ))为脊髓线坐 标， α„为加权系数， 可根据经验设定， Μ„为平移距离相对左右体表边界距 离的倍数， 可根据经验设定， S为推体轴线数量。
提取出的推体轴线截面是一个灰度向量， 它表达了沿着推体轴线方向 的近似灰度变化， 如图 11 , 其中灰度剧烈变化的位置对应推体与推间盘的 边缘。
当确定出推体轴线截面后， 定位推体内部种子点的方法如图 12所示， 包括以下步骤：
步骤 S 31 , 计算推体轴线截面的灰度变化梯度， 将梯度值大于设定阈值 的点作为推体与相邻推间盘的候选边界点。 因为在推体边界点处截面呈现 剧烈变化， 可依据梯度特征来检测边界点。 通过计算截面梯度， 将满足 |g( )| > r条件的候选点作为推体边界点， 其中 g(t)为截面梯度， ： Γ为阈值， 可取 r = meim(|g(t)|)。 边界点提取结果如图 1 3 , 其中， 黑色点为边界点。
步骤 S 32 , 将检测出的候选边界点按照第三预定规则进行筛选，得到推 体与相邻推间盘的边界点。 这些检测出的边界点有很多冗余， 如一条边界 往往会检测出多个边界点， 这需要进行边界点筛选。 可利用边界点的距离 特性进行筛选， 使得最后任意两个相邻边界之间点的距离 d满足： 其中， N2为经验值， 可依据经验进行选择。
第三预定规则除了本实施例中上述 选规则外， 还可以是其他规则， 要实现在众多边界点中筛选出符合条件的边界点即可， 以减少边界点的 本实施例中， 提取边界点的目的是为了后续得到尽量少的种子点， 如 果直接利用均匀性提取推体内部种子点则会得到非常多的候选点， 使算法 效率变低。
步骤 S 33 ,将边界点的坐标沿推体轴线方向按照设定大小移动，得到的 新坐标所对应的点记为推体内部的种子点。 例如将筛选后的边界点向上或 下进行垂直移动， 目的是将边界点的位置移动到推体内部。 这样做的依据 是： 推体面积大于推间盘的面积， 将边界点进行微小移动即可得到推体内 部的种子点。 因此移动的距离可根据经验设定， 例如大于推间盘上下宽度 的一个值。 筛选移动后的图如图 14所示， 很多种子点（图中为黑点）移到 了推体内部。
因本实施例中采用的是推体轴线上灰度的变化梯度， 梯度表示灰度的 变化， 它与灰度本身的高低无关， 只要灰度有变化， 都可在梯度上反映出 来。 在磁共振推体图像中， T1加权图像的推间盘一般黑色， T2加权图像的 推间盘一般白色， 但这种特征有时并不明显， 也就是推骨与推间盘的灰度 差异有时很小， 如果用其他方法直接提取种子点， 总会受到这个问题的干 扰， 而无论是 T1加权图像还是 T2加权图像， 推骨与推间盘之间都是有明 确边界的， 都可以根据梯度检测到该边界。 所以本实施例在提取推体内部 的种子点时不受推体或者推间盘灰度不一致的影响， 也不受权重的影响， 可更准确地提取到推体内部的种子点， 以进行后续的推体区域生长。
在又一种具体实例中， 推体区域生长采用自适应阈值的区域生长法， 本实施例中可利用推体的面积、 推体的形状迭代计算生长阈值。 生长后的 区域坐标满足： υί ，. y^ - I seedxj， seedy j 其中,其中, 《^· , seedy )为第 j个种子的坐标, (x; , y{ )为 e ;^ , seedy j ) 的孤立邻域集合， 为种子点数量， Hj为基于第 j个种子的生长阈值。 这里 关键需要确定最优的生长阈值 Hj。
如图 1 5所示， 以单个种子为例， 区域生长法提取推体区域包括以下步 骤：
步骤 S41 ,设置初始阈值 Ηίηί、生长目标最小面积^和生长目标最大面 积 并将初始阈值 Ηίηί赋予生长阈值 Η ; 其中初始阈值 Ηίηί可以是一个比 较大的固定值， 也可以是推体轴线的最大灰度和最 d、灰度差值乘以一个系 数得到的一个值， 和^可利用 进行估算， 如 SM二 和 ST二 ， ^和 L2依经验选择。 步骤 S 42 ,基于生长阈值进行区域生长，将生长后的区域进行区域填充。 本步骤中区域生长可采用已有的技术基于种子点进行区域生长并填充。
步骤 S4 3 , 计算填充后区域的面积周长比 SP , SP = - , 其中， S为填充
P
后区域的面积， P为填充后区域的周长。
步骤 S44 ,根据填充后区域的面积推算出期望的面积周长比。根据推体 形状基本固定且呈近似正方形的特征， 假设推体形状为正方形， 根据填充 后区域的面积 S , 推算出周长， 然后再次计算面积周长比， 即推算出期望的 面积周长比 SPR , SPR = ~。
4V5 - 4 步骤 S45 , 计算面积周长比和期望的面积周长比的差异。 本步骤中， 面 积周长比和期望的面积周长比的差异可以是两者的差值， 也可以是两者的 比值。
步骤 S46 , 判断生长是否满足预定条件，预定条件可以是生长阈值 H已 经很小， 小于设定阈值。预定条件也可以是填充后区域的面积 S已经很小， 小于设定阈值。 如果生长满足预定条件， 则执行步骤 S48 , 停止基于该种 子的生长， 并在位于生长目标最小面积和生长目标最大面积之间的填充后 区域面积的差异中查找出最小差异， 将该最小差异所对应的生长阈值记为 最佳生长阈值， 将最小差异所对应的填充后区域记为基于该种子生长的推 体区域； 否则执行步骤 S47;
步骤 S47 , 生长阈值变换。 按照第四预定规则减小生长阈值， 然后转向 步骤 S42 ,基于新的生长阈值进行区域生长，循环执行直到生长满足预定条 件。 第四预定规则可以是使生长阈值按照设定的步长逐次减小， 也可以使 生长阈值按照一个设定的曲线递减， 还可以无规律或随机地逐次减小生长 阈值。
因为推体面积不能精确估算， 如果只利用推体面积来决定阈值则很可 能得到形状不规则的区域， 分割效果也不佳。 因为噪声等因素的影响使得 推体灰度均勾性并不理想， 灰度分布无规律， 不同的灰度阈值生长出的区 域可能面积相差不大， 但形状却差异明显， 只有所提取的区域与推骨正好 吻合时， 形状最规则， 因此本实施例结合了推体的形状信息， 同时又考虑 了推体面积， 使得生长后的区域与真实推体区域最相似。
对每个种子点， 得到最优的 H后， 也得到了该种子点对应的推体区域。 这时需要对该区域进行分析判断， 因为前面得到的种子点数量仍然存在冗 余， 主要有两种情况：
1 )某些种子点并不在推体以内， 而在推间盘上， 这时即使进行上述的 区域生长， 得到的区域也不是推体区域；
2 )某些种子点同时处于同一推体内部， 它们生长出的区域会有重叠部 分。
为了解决上面两个问题， 对提取的推体区域进行分析， 按照第五预定 规则去除推体区域的冗余部分。 可通过以下全部或部分信息来去除冗余： 考察每个种子点生长后的 SP和 SPR的差异；
考察推体轴线上上下相邻两个种子点生长后区域的位置信息； 考察推体轴线上上下相邻两个种子点生长后区域的面积差异等等。 这些信息可以有效解决上面两个问题。
经过区域生长、 区域分析判断后， 得到推体区域如图 16所示。
根据本发明公开的内容， 本领域技术人员应该理解， 在进行脊柱推体 提取时， 可采用上述实施例中的全部步骤， 也可采用部分步骤进行组合了， 例如， 在定位脊髓线时采用现有技术， 而在定位推体内部种子点和进行区 域生长时采用本发明实施例中的方案， 或者在定位脊髓线和进行区域生长 时采用本发明实施例中的方案， 而其他步骤采用现有技术。
在另一实施例中， 控制及处理系统包括脊柱推间盘分割装置， 脊柱推 间盘分割装置包括上述任一实施例中的脊柱推体提取装置、 推体区域顶点 定位单元、 推间盘中心点定位单元和推间盘中心线确定单元。 推体区域顶 点定位单元用于在提取的推体区域上定位该推体区域的顶点， 推间盘中心 点定位单元用于利用相邻两个推体区域的顶点分别计算推间盘两相对边的 中心点， 推间盘中心线确定单元， 用于根据两相对边的中心点确定推间盘 中心线。
基于上述脊柱推间盘分割装置， 脊柱推间盘分割方法如图 17所示， 包 括以下步骤：
步骤 S1 , 利用磁共振图像数据定位脊髓线。
步骤 S2 , 根据脊髓线确定推体轴线截面。
步骤 S3 , 根据推体轴线截面定位推体内部种子点。
步骤 S4 , 基于推体内部种子点采用种子区域生长法提取推体区域。 步骤 S5 , 在提取的推体区域上定位该推体区域的顶点。 定位推间盘扫 描线需要确定推间盘中心线的表达式， 在前面得到的推体区域结果上， 需 要检测每个推体的四个顶点坐标， 用相邻两个推体的顶点坐标计算推间盘 中心点位置。 可利用筒单的螺旋扫描来提取推体的顶点， 分别采用不同的 扫描方向即可分别得到推体的四个顶点。
步骤 S6 , 利用相邻两个推体区域的顶点分别计算推间盘两相对边的中 心点。 前一推体左下角顶点与后一推体左上角顶点的中点坐标为推间盘的 左中心点， 前一推体右下角顶点与后一推体右上角顶点的中点坐标为推间 盘的右中心点。
步骤 S7 , 根据两相对边的中心点确定推间盘中心线。 最后提取的推间 盘中心线如图 18。
本实施例中的步骤 S1至 S4可以采用本发明提供的全部或部分步骤， 也可以是全部或部分采用现有技术实现。
本发明利用推体内灰度近似均勾、 推体与推间盘有明显边界、 推体形 状基本固定等特征自适应提取推体和推间盘， 提取结果不受推体或者推间 盘灰度不一致的影响， 也不受图像权重的影响， 可自适应提取图像中的全 部推体， 进而计算出各个推间盘的中线位置和角度， 能有效应用于 MRI (磁 共振图像）脊柱推间盘扫描的全自动定位。
本发明不受场强高低的影响， 可应用于任何场强的 MRI成像系统。 以上说明了如何基于磁共振图像提取脊柱推体或进行脊柱推间盘分 割， 但当采用其他方式获得被测物组织图像时， 同样可按照本发明提供的 方法和 /或装置提取脊柱推体或进行脊柱推间盘分割。
以上内容是结合具体的实施方式对本发明所作的进一步详细说明， 不能认 定本发明的具体实施只局限于这些说明。 对于本发明所属技术领域的普通 技术人员来说， 在不脱离本发明构思的前提下， 还可以做出若干筒单推演 或替换， 都应当视为属于本发明的保护范围。

Claims

权利要求书
1. 一种脊柱推体提取方法，其特征在于， 包括：
利用磁共振图像数据定位脊髓线；
根据脊髓线确定推体轴线截面；
根据推体轴线截面定位推体内部种子点；
基于推体内部种子点采用种子区域生长法提取推体区域。
2. 如权利要求 1所述的方法， 其特征在于， 所述定位脊髓线包括： 利用被测者组织和背景的过渡特性检测被测者磁共振图像的两侧体表 边界；
根据体表边界界定的被测者组织宽度检测脊髓线。
3. 如权利要求 2 所述的方法， 其特征在于， 所述检测被测者磁共 振图像的两侧体表边界包括：
在与体表边界方向垂直的方向上求一阶导数；
检测计算出的一阶导数的极大值点和极小值点；
沿体表边界方向变换求一阶导数的位置， 循环上述步骤， 检测出多个 极大值点集合和极小值点集合， 所述极大值点集合用于检测第一侧体表边 界， 极小值点集合用于检测第二侧体表边界；
对多个极大值点集合和极小值点集合按照第一预定规则进行 选， 得 到第一侧体表边界和第二侧体表边界；
所述根据体表边界界定的被测者组织宽度检测脊髓线包括：
根据第一侧体表边界和第二侧体表边界的距离估算滤波尺度； 采用所述滤波尺度对磁共振图像求二阶导数；
检测二阶导数图像的极值， 得到多个极值点集合；
对多个极值点集合按照第二预定规则进行 选， 得到脊髓线。
4. 如权利要求 1至 3 中任一项所述的方法， 其特征在于， 根据脊 髓线确定推体轴线截面包括：
将脊髓线按照多个设定的平移距离平移后得到多个推体轴线； 对多个推体轴线的灰度求加权和， 得到推体轴线截面。
5. 如权利要求 1至 4 中任一项所述的方法， 其特征在于， 根据推 体轴线截面定位推体内部种子点包括： 计算推体轴线截面的灰度变化梯度， 将梯度值大于设定阈值的点作为 推体与相邻推间盘的候选边界点；
将检测出的候选边界点按照第三预定规则进行筛选， 得到推体与相邻 推间盘的边界点；
将边界点的坐标沿推体轴线方向按照设定大小移动， 得到的新坐标所 对应的点记为推体内部的种子点。
6. 如权利要求 1至 5 中任一项所述的方法， 其特征在于， 基于推 体内部种子点采用种子区域生长法提取推体区域包括：
基于推体内部种子点进行区域生长， 使生长后区域的坐标满足： )为第 j个种子
Figure imgf000017_0001
的坐标， (x;, _y; ;>为 ( , ^^y )的孤立邻域集合， 为种子点数量， Hj为基 于第 j个种子的生长阈值。
7. 如权利要求 6所述的方法， 其特征在于， 第 j个种子的的最优 生长阈值的确定方法包括以下步骤：
初始步骤， 用于设置初始阈值、 生长目标最小面积和生长目标最大面 积， 并将初始阈值赋予生长阈值；
区域生长步骤， 用于基于生长阈值进行区域生长；
区域填充步骤， 用于将生长后的区域进行区域填充；
第一计算步骤， 用于计算填充后区域的面积周长比；
第二计算步骤， 用于根据填充后区域的面积推算出期望的面积周长比； 第三计算步骤， 用于计算面积周长比和期望的面积周长比的差异； 判断步骤， 用于判断生长是否满足预定条件， 若是则停止基于该种子 的区域生长， 并在位于生长目标最小面积和生长目标最大面积之间的填充 后区域面积的差异中查找出最小差异， 所述最小差异所对应的生长阈值为 最佳生长阈值， 所述最小差异所对应的填充后区域为推体区域； 否则执行 以下步骤；
生长阈值变换步骤， 用于按照第四预定规则减小生长阈值， 然后循环 执行区域生长步骤至判断步骤。
8. 如权利要求 7 所述的方法， 其特征在于， 在提取推体区域之后 还包括：
对提取的推体区域进行分析， 按照第五预定规则去除推体区域的冗余 部分。
9. 一种脊柱推体提取装置，其特征在于， 包括：
脊髓线定位单元， 用于利用磁共振图像数据定位脊髓线；
轴线截面确定单元， 用于根据脊髓线确定推体轴线截面；
种子点定位单元， 用于根据推体轴线截面定位推体内部种子点； 推体区域提取单元， 用于基于推体内部种子点采用种子区域生长法提 取推体区域。
10. 如权利要求 9 所述的装置， 其特征在于， 所述脊髓线定位单元 包括：
边界检测子单元， 用于利用被测者组织和背景的过渡特性检测被测者 磁共振图像的两侧体表边界；
脊髓线检测单元， 用于根据体表边界界定的被测者组织宽度检测脊髓 线。
11. 如权利要求 1 0所述的装置， 其特征在于， 所述边界检测子单元 用于在与体表边界方向垂直的方向上求一阶导数， 检测计算出的一阶导数 的极大值点和极小值点， 沿体表边界方向变换求一阶导数的位置， 循环上 述步骤， 检测出多个极大值点集合和极小值点集合， 所述极大值点集合用 于检测第一侧体表边界， 极小值点集合用于检测第二侧体表边界， 对多个 极大值点集合和极小值点集合按照第一预定规则进行 选， 得到第一侧体 表边界和第二侧体表边界； 所述脊髓线检测单元用于根据第一侧体表边界 和第二侧体表边界的距离估算滤波尺度， 采用所述滤波尺度对磁共振图像 求二阶导数， 检测二阶导数图像的极值， 得到多个极值点集合， 对多个极 值点集合按照第二预定规则进行 选， 得到脊髓线。
12. 如权利要求 9至 11中任一项所述的装置， 其特征在于， 所述轴 线截面确定单元包括：
平移子单元， 用于将脊髓线按照多个设定的平移距离平移后得到多个 推体轴线； 加权子单元， 用于对多个推体轴线的灰度求加权和， 得到推体轴线截 面。
1 3. 如权利要求 9至 12中任一项所述的装置， 其特征在于， 所述种 子点定位单元包括：
边界点检测子单元， 用于计算推体轴线截面的灰度变化梯度， 将梯度 值大于设定阈值的点作为推体与相邻推间盘的候选边界点；
筛选子单元， 用于将检测出的候选边界点按照第三预定规则进行筛选， 得到推体与相邻推间盘的边界点；
移动子单元， 用于将边界点的坐标沿推体轴线方向按照设定大小移动， 得到的新坐标所对应的点记为推体内部的种子点。
14. 如权利要求 9至 1 3中任一项所述的装置， 其特征在于， 所述推 体区域提取单元用于基于推体内部种子点进行区域生长， 使生长后区域的 坐标满足： [J , y; ) I l (x;, y; ) - l seedxj , seedy j ) < H ， 其中 , ( seedxj , seedy j )为 第 j个种子^坐标， (x; , _y; )为 ( Zx , e y )的孤立邻域集合， 为种子点数 量， Hj为基于第 j个种子的生长阈值。
15. 如权利要求 14所述的装置， 其特征在于， 所述推体区域提取单 元包括：
初始子单元， 用于设置初始阈值、 生长目标最小面积和生长目标最大 面积， 并将初始阈值赋予生长阈值；
区域生长子单元， 用于基于生长阈值进行区域生长并将生长后的区域 进行区域填充；
计算子单元， 用于计算填充后区域的面积周长比， 根据填充后区域的 面积推算出期望的面积周长比， 并计算面积周长比和期望的面积周长比的 差异；
判断子单元， 用于判断生长阈值是否满足预定条件；
查找及提取子单元， 用于在生长满足预定条件时， 在位于生长目标最 小面积和生长目标最大面积之间的填充后区域面积的差异中查找出最小差 异， 将所述最小差异所对应的生长阈值记为最佳生长阈值， 将所述最小差 异所对应的填充后区域提取为推体区域；
生长阈值变换子单元， 用于在生长阈值不满足预定条件时按照第四预 定规则减小生长阈值， 生成新的生长阈值， 然后控制区域生长子单元基于 新的生长阈值进行区域生长。
16. 如权利要求 16所述的装置， 其特征在于， 所述推体区域提取单 元还包括：
区域分析子单元， 用于对提取的推体区域进行分析， 按照第五预定规 则去除推体区域的冗余部分。
17. 一种脊柱推间盘分割方法，其特征在于， 包括：
如权 1-8中任一项所述的脊柱推体提取方法；
在提取的推体区域上定位该推体区域的顶点；
利用相邻两个推体区域的顶点分别计算推间盘两相对边的中心点； 根据两相对边的中心点确定推间盘中心线。
18. —种脊柱推间盘分割装置，其特征在于， 包括：
如权 9-16中任一项所述的脊柱推体提取装置；
推体区域顶点定位单元， 用于在提取的推体区域上定位该推体区域的 顶点；
推间盘中心点定位单元， 用于利用相邻两个推体区域的顶点分别计算 推间盘两相对边的中心点；
推间盘中心线确定单元， 用于根据两相对边的中心点确定推间盘中心 线。
一种磁共振系统， 其特征在于， 包括如权利要求 9至 16中任一项所述的脊 柱推体提取装置或权利要求 18所述脊柱推间盘分割装置。
PCT/CN2012/073131 2011-03-31 2012-03-27 脊柱椎体和椎间盘分割方法、装置、磁共振成像系统 Ceased WO2012130132A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US14/041,919 US9445744B2 (en) 2011-03-31 2013-09-30 Methods, systems, and devices for spine centrum extraction and intervertebral disk dividing

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201110080865.XA CN102727200B (zh) 2011-03-31 2011-03-31 脊柱椎体和椎间盘分割方法、装置、磁共振成像系统
CN201110080865.X 2011-03-31

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US14/041,919 Continuation US9445744B2 (en) 2011-03-31 2013-09-30 Methods, systems, and devices for spine centrum extraction and intervertebral disk dividing

Publications (1)

Publication Number Publication Date
WO2012130132A1 true WO2012130132A1 (zh) 2012-10-04

Family

ID=46929466

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2012/073131 Ceased WO2012130132A1 (zh) 2011-03-31 2012-03-27 脊柱椎体和椎间盘分割方法、装置、磁共振成像系统

Country Status (3)

Country Link
US (1) US9445744B2 (zh)
CN (1) CN102727200B (zh)
WO (1) WO2012130132A1 (zh)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104780838A (zh) * 2012-11-22 2015-07-15 株式会社东芝 磁共振成像装置
WO2016086744A1 (en) * 2014-12-02 2016-06-09 Shanghai United Imaging Healthcare Co., Ltd. A method and system for image processing
CN118447017A (zh) * 2024-07-05 2024-08-06 大连杰伍科技有限公司 一种基于区域生长的腰椎间盘突出检测方法

Families Citing this family (26)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP6238568B2 (ja) * 2013-05-20 2017-11-29 東芝メディカルシステムズ株式会社 磁気共鳴イメージング装置
CN103479380B (zh) * 2013-09-17 2015-05-13 飞利浦(中国)投资有限公司 椎骨和椎间盘形状位置参数的识别方法
EP3047455B1 (en) 2013-09-17 2020-07-08 Koninklijke Philips N.V. Method and system for spine position detection
CN103606148B (zh) * 2013-11-14 2017-10-10 深圳先进技术研究院 一种磁共振脊柱影像混合分割方法和装置
CN103854287B (zh) * 2014-03-11 2017-06-06 深圳市旭东数字医学影像技术有限公司 基于磁共振图像的半月板分割的方法及其装置
CN105096291A (zh) * 2014-04-25 2015-11-25 上海联影医疗科技有限公司 一种中心线的提取方法
EP3244799B1 (en) * 2015-01-16 2018-05-23 Koninklijke Philips N.V. Vertebral feature identification
JP6401083B2 (ja) * 2015-03-12 2018-10-03 富士フイルム株式会社 医用画像処理装置、方法およびプログラム
WO2016201120A1 (en) 2015-06-09 2016-12-15 University Of Virginia Patent Foundation Insulin monitoring and delivery system and method for cgm based fault detection and mitigation via metabolic state tracking
EP3369072B1 (en) * 2015-10-29 2025-04-02 Broncus Medical Inc. Learning-based spine vertebra localization and segmentation in 3d ct
CN106780520B (zh) * 2015-11-18 2021-04-13 周兴祥 一种mri腰椎图像中椎骨的自动提取方法
CN105433988B (zh) * 2015-12-28 2018-10-16 深圳开立生物医疗科技股份有限公司 一种目标图像识别方法、装置及其超声设备
WO2018161257A1 (zh) 2017-03-07 2018-09-13 上海联影医疗科技有限公司 生成彩色医学影像的方法及系统
WO2019023891A1 (en) * 2017-07-31 2019-02-07 Shenzhen United Imaging Healthcare Co., Ltd. SYSTEMS AND METHODS FOR AUTOMATIC SEGMENTATION AND IDENTIFICATION OF VERTEBRA IN MEDICAL IMAGES
JP6522077B2 (ja) * 2017-10-30 2019-05-29 キヤノンメディカルシステムズ株式会社 磁気共鳴イメージング装置
CN108520525B (zh) * 2018-04-12 2021-11-02 重庆理工大学 一种基于凸约束种子区域生长的脊髓分割方法
CN109493317B (zh) * 2018-09-25 2020-07-07 哈尔滨理工大学 基于级联卷积神经网络的3d多椎骨分割方法
CN109978861B (zh) * 2019-03-27 2021-03-26 北京青燕祥云科技有限公司 骨髓灰质检测方法、装置、设备和计算机可读存储介质
CN109934824B (zh) * 2019-03-27 2021-02-12 清华大学 颈椎脊髓高信号的检测方法及系统
CN113516614A (zh) * 2020-07-06 2021-10-19 阿里巴巴集团控股有限公司 脊柱影像的处理方法、模型训练方法、装置及存储介质
CN112528991A (zh) * 2020-12-04 2021-03-19 华南理工大学广州学院 一种对腰椎间盘进行标注的方法、装置、设备和存储介质
CN112690778B (zh) * 2021-01-14 2023-02-14 东莞安科医疗科技有限公司 一种生成脊柱椎间盘定位线的方法及系统
CN112734757B (zh) * 2021-03-29 2021-06-25 成都成电金盘健康数据技术有限公司 一种脊柱X光图像cobb角测量方法
CN114549505B (zh) * 2022-03-01 2025-03-14 卫宁健康科技集团股份有限公司 肋骨分割方法、装置、计算机设备及存储介质
CN115147686B (zh) * 2022-09-06 2022-11-25 杭州健培科技有限公司 一种脊柱椎体校正识别方法、装置及应用
CN119006479B (zh) * 2024-10-25 2024-12-31 陕西省中医医院 一种用于腰椎病变检测的mri图像分析方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000126150A (ja) * 1998-10-23 2000-05-09 Ge Yokogawa Medical Systems Ltd 関心領域設定方法、画像処理装置および医用画像処理装置
US6490472B1 (en) * 1999-09-03 2002-12-03 The Mcw Research Foundation, Inc. MRI system and method for producing an index indicative of alzheimer's disease
CN1449721A (zh) * 2002-04-03 2003-10-22 株式会社东芝 具有医学图像测量功能的医学图像处理装置
CN101301224A (zh) * 2007-05-07 2008-11-12 西门子公司 左心房和肺静脉的三维重建
CN101515367A (zh) * 2009-04-01 2009-08-26 西北工业大学 三维大脑磁共振图像的大脑皮层表面上脑沟区域分割方法

Family Cites Families (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2003520658A (ja) * 2000-01-27 2003-07-08 コーニンクレッカ フィリップス エレクトロニクス エヌ ヴィ 脊柱の幾何学的データを抽出する方法及びシステム
WO2005055008A2 (en) * 2003-11-26 2005-06-16 Viatronix Incorporated Automated segmentation, visualization and analysis of medical images
US8014575B2 (en) * 2004-03-11 2011-09-06 Weiss Kenneth L Automated neuroaxis (brain and spine) imaging with iterative scan prescriptions, analysis, reconstructions, labeling, surface localization and guided intervention
US7561728B2 (en) * 2005-03-23 2009-07-14 Siemens Medical Solutions Usa, Inc. Detection of intervertebral disk orientation in spine images using curve evolution
US7804986B2 (en) * 2005-09-13 2010-09-28 Siemens Medical Solutions Usa, Inc. System and method for detecting intervertebral disc alignment using vertebrae segmentation
US7876938B2 (en) * 2005-10-06 2011-01-25 Siemens Medical Solutions Usa, Inc. System and method for whole body landmark detection, segmentation and change quantification in digital images
US7903851B2 (en) * 2005-10-17 2011-03-08 Siemens Medical Solutions Usa, Inc. Method and system for vertebrae and intervertebral disc localization in magnetic resonance images
US8175349B2 (en) * 2006-08-16 2012-05-08 Siemens Medical Solutions Usa, Inc. System and method for segmenting vertebrae in digitized images
US20120143090A1 (en) * 2009-08-16 2012-06-07 Ori Hay Assessment of Spinal Anatomy
CA2797302C (en) * 2010-04-28 2019-01-15 Ryerson University System and methods for intraoperative guidance feedback
CN103300856B (zh) * 2012-03-13 2015-11-25 深圳迈瑞生物医疗电子股份有限公司 Mri图像的颈椎椎体轴线及相关组织的定位方法与装置

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000126150A (ja) * 1998-10-23 2000-05-09 Ge Yokogawa Medical Systems Ltd 関心領域設定方法、画像処理装置および医用画像処理装置
US6490472B1 (en) * 1999-09-03 2002-12-03 The Mcw Research Foundation, Inc. MRI system and method for producing an index indicative of alzheimer's disease
CN1449721A (zh) * 2002-04-03 2003-10-22 株式会社东芝 具有医学图像测量功能的医学图像处理装置
CN101301224A (zh) * 2007-05-07 2008-11-12 西门子公司 左心房和肺静脉的三维重建
CN101515367A (zh) * 2009-04-01 2009-08-26 西北工业大学 三维大脑磁共振图像的大脑皮层表面上脑沟区域分割方法

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104780838A (zh) * 2012-11-22 2015-07-15 株式会社东芝 磁共振成像装置
WO2016086744A1 (en) * 2014-12-02 2016-06-09 Shanghai United Imaging Healthcare Co., Ltd. A method and system for image processing
US10181191B2 (en) 2014-12-02 2019-01-15 Shanghai United Imaging Healthcare Co., Ltd. Methods and systems for identifying spine or bone regions in computed tomography image sequence
CN118447017A (zh) * 2024-07-05 2024-08-06 大连杰伍科技有限公司 一种基于区域生长的腰椎间盘突出检测方法

Also Published As

Publication number Publication date
US9445744B2 (en) 2016-09-20
CN102727200A (zh) 2012-10-17
US20140046169A1 (en) 2014-02-13
CN102727200B (zh) 2016-03-30

Similar Documents

Publication Publication Date Title
WO2012130132A1 (zh) 脊柱椎体和椎间盘分割方法、装置、磁共振成像系统
CN103300856B (zh) Mri图像的颈椎椎体轴线及相关组织的定位方法与装置
JP4544745B2 (ja) 統計的識別特性を利用した適応性断面積計算方法
CN109949349B (zh) 一种多模态三维图像的配准及融合显示方法
US11229419B2 (en) Method for processing 3D image data and 3D ultrasonic imaging method and system
CN101421745B (zh) 空间-时间肿瘤检测，分割和诊断信息提取系统及方法
US7218763B2 (en) Method for automated window-level settings for magnetic resonance images
JP5731373B2 (ja) 患者の安全性、及び走査性能の改良のためのリアルタイム性の局所的及び大域的なｓａｒ推定
US8777859B2 (en) Method and apparatus for processing ultrasound image
CN112946547B (zh) 用于从校准图像生成诊断扫描参数的系统和方法
JP6218569B2 (ja) 磁気共鳴イメージング装置
JP2004535874A (ja) 磁気共鳴血管造影法及びその装置
CN106133789B (zh) 用于分割感兴趣区域的图像处理装置和方法
Hwang et al. Estimating voxel volume fractions of trabecular bone on the basis of magnetic resonance images acquired in vivo
CN112166332B (zh) 使用磁共振指纹识别的异常检测
US10718838B2 (en) Systems and methods for calibrated multi-spectral magnetic resonance imaging
CN102968633B (zh) 基于ct侧位定位片的腰椎识别方法和装置
US20250160784A1 (en) Systems and methods for detecting angles of hip joints
CN117745704A (zh) 一种用于骨质疏松识别的脊椎区域分割系统
CN113850729B (zh) 图像处理装置、医用摄像装置以及存储介质
JP2017516590A (ja) 微細構造解析データのデータ品質を評価し向上させる方法
KR100942699B1 (ko) 요골원위부 골간단 경계추출 방법 및 장치
Steines et al. Segmentation of osteoarthritic femoral cartilage using live wire
JP6513493B2 (ja) 磁気共鳴撮像装置
Liu et al. Automatic segmentation of prostate boundaries in transrectal ultrasound (TRUS) imaging

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 12763628

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205N DATED 06/12/2013)

122 Ep: pct application non-entry in european phase

Ref document number: 12763628

Country of ref document: EP

Kind code of ref document: A1