TW499308B - A new method for registration of computerized brain atlas and CT image - Google Patents

Abstract

In the clinical application of surgery, mini-invasive procedures have attracted the attention of surgeons since it can extract targeted tissue while minimizing damage to adjacent structures. The difficulty of this three-dimensional (3-D) localization is increased by the fact that it often requires isolating a structure deeply buried within the body. While methods exist for imaging and displaying the 3-D structure of the body, the surgeon must still relate what he sees on the 3-D display with the patient's actual anatomy. Current methods often involve a surgeon simply utilizing traditional two dimensional (2-D) slices of MRI or CT imagery and mentally transforming them to the actual patient. Thus, there is a clear need for registered visualization techniques, where 3-D reconstruction of internal anatomy is overlaid with the surgeon's view of the patient, so that he can directly visualize important structures, and act accordingly. In this patent, we propose a new scheme for 3-D registration and fusion of the computerized stereotactic atlas and computered tomography (CT) slices. Based on 3-D geometrical analysis, the localization and correction algorithms are used to derive the accurate position of the desired target in the surgery. In summary, his scheme has the following functions. (1) 3D localization of brain CT images. (2) Registration of brain atlas and brain CT images. This scheme can provide the accurate location of targeted tissue for driving the probe to the optimal treatment position. The success of this scheme is helpful in automating the surgery system and improving the quality and efficiency of the surgery.

Description

499308 V. Description of the invention (1) Industrial application field The present invention discloses a method for aligning computer atlas of brain and CT image. In the clinical application of neurosurgery, the CT image combined with the computer atlas of the brain can quantify the surface structure of the brain tissue, which is convenient for doctors to locate invisible lesions. In Jingbei medicine, there are the following methods for the treatment of patients with Parkinson's disease: 1. Internal medicine and music therapy: the most important treatment method, physicians usually increase from a single-low-dose pretreatment f 'and then increase to appropriate maintenance. Do not forget to take the medicine on time. The type and type of the medicine and the reaction after taking the medicine must be recorded and discussed with your doctor. Do not change the medicine or change it without permission. Patients with cardiovascular, respiratory, and endocrine diseases must take special care. In addition, vitamin β6 will reduce the effectiveness of the drug, so avoid overdose. 2. Surgical treatment ················································ refers to the inhibition of the ventral nucleus of the visual mound or the ball of the Cang from the view of Wei Quan, Shansi, read and put the operation ^ to be languing ^ speed watch, despise Qiu Cai and Deya _ Ming = = reduce drug consumption and complications, stereotactic surgery was completed:, the most thorough coffee 19Γ $ 7 pages 499308 five 、 Explanation of invention (2) ^ Horsley and aarfce's surgical experiments, due to insufficient technical maturity at that time, did not attract much attention, but the surgery in the deep part of the brain was already very valuable. Following the development of this technology, in the 5G and 6G years, it even became the main method for treating Parkinson's disease. By the 1970s ~ years, due to the use of drugs, most patients switched to drug treatment and stereotactic surgery. The application has been relatively reduced. After the bile years, the production of three-dimensional granule surgery has made revolutionary progress, mainly due to computer tomography. Technology has been able to cooperate with it. Through computer tomography, it is found that the brain is small: A's ( target) '-like traditional surgery Achieved, the accuracy and depth of stereotactic surgery can make up for its shortcomings. The principle of stereotactic surgery is basically fixed on the head with special instruments, and then using x-ray development technology such as electric Linlin Ying's aging instrument calibration and assistance The target is parked in the visual mound. After interpreting the value of the red scale, the computer software 2 is used to accurately reorganize and determine the three-dimensional spatial coordinate position of the visual mound. After the seat post is known, it can be explored on the instrument by machine. The position of the needle will arrive at the ship and undergo surgery. In the past, the accuracy of the target plant was as described above. The accusation = the positioning software provided by Qiqi, which is not only costly (_ $ 10,000) The computer map of the brain provided by Hachiman is still only a two-dimensional resource and the experience of a neurosurgeon to successfully complete the operation. ^ It is still necessary to find a balanced basal nucleus by the basal wealth of 4 coffee and = Image of Chizaki Pass ~ --- w I ~ image (such as CT5 MR, pet), page 8 499308

^^^ ® # m (sUrface feature) ^^ t ^^ (def〇nnaWe m〇del)? 舄 Transfer matrix, who can't do it ^, ^) to align. However, the errors are in several coffees. As for the high-precision stereotactic surgery, it does not turn over, because the object of the surgery = the surrounding material has been discussed for several times. That's why the frame and head surgery is still adopted. Gu Gujia. In other words, the frameless system will place the positioning on the scalp, and the skin is soft tissue ... Push it immediately, pay no attention, and the wrong money comes into being. It is the so-called "poor house, lost a thousand miles." Therefore, only when the money frame is nailed to the money can we get the exact ‘object coordinates’. Secondly, on the ship ’s customs office, for the financial law, it is assumed that the computer breaks the reference axis of the cat (that is, the three-dimensional head holder is placed in the ideal position), or the offset is small enough to be ignored. _This practice is in the clinical application society, "cannot be. In other words, a complete positioning system should have this correction function. Objective of the invention

The primary object of the present invention is to reveal a new method of computer tomography scans and computer atlas alignment—the group of CT medical images obtained by the imaging device, extracting nine positioning points on the image plane, and deriving the entire group of CT medical images in The three-dimensional coordinates on the frame coordinates; then according to the CT image scales of different patients and the position of the reference axis of computer tomography in the ㈣ coordinate system, the computer map is translated by using coordinate transformation to extend (or shrink) and rotate the three-dimensional Alignment. Not only can it shorten the time it takes for a physician to search for a lesion, but it can also improve the safety of surgery.

499308

Detailed description of the invention The present invention _ a method for aligning computer brain atlas and CT image ", including: Γ: processing 'for generating three-dimensional CT image data of the brain, and computer atlas of brain; It is used to convert the CT image coordinates to the frame coordinate system;-The Si position refers to the reference axis side, and the GT provides the alignment of the GT image and the map; 4. The alignment of the map and the CT image, according to the 3D coordinate axis of the mouth image in the spatial position, the brain The map is converted to a frame coordinate system. Figure-shows the block diagram of the brain electrical spectrum and CT # image alignment method of the present invention in detail as follows: 1. Pre-processing The pre-processing of this method includes the generation of computer atlas data of the brain. * 3_P1 Qian high-level job sweeping and correction, the three axial Erlai Department graphic materials into a 512 * 512 pixel graphics building. For the generation of 3D CT image data of the brain, the step-sequential cross-section CT image read gray scale chirp method is used to reconstruct its 3D volume data. The linear gray-scale interpolation method used in the previous process is as follows:

The gray level value of the ㈣ position of the m layer 乂 ㈣ ㈣ is the relative position of the image of the / 7 ″ layer gray where M is the number of interpolated layers, / j3c, do not interpolate to the relative position of the η layer image , / Wi order value. Brain CT image localization

499308 V. Description of the invention (5) The three-dimensional fixed-value step shame description describes the structure of the thin target and the coordinates of the three-dimensional frame. And a target point can be the normal anatomical structure of the brain tissue, or it can be a little bit of surgical towel _ cells _ in the tumor. There are three requirements for three-dimensional hand shooting. — The target property is seen directly or indirectly in radiography; 2. The coordinates of the reference point on the stereotactic surface must be known; 3. There must be—the location of the image, radiography, or physiological punctuation, and the stereolocation In the operation, the position of the guiding instrument === The calculation of three-dimensional miscellaneous coordinates in Qianshu Qian includes two basic steps. First of all, in order to accurately determine the three-dimensional coordinates after the heterolocalization, the patient's head must be fixed on the head ring in a stereo positioning system, and then the stereo positioning frame is joined to the head ring. As shown in Figure 2, the CRW headband and BRW stand are used in computed tomography scans; then, a two-dimensional structure on the BRWLRiL body position frame is used to perform a computed tomography scan in CT ~ The required nine reference points are generated in the image, as shown in Figure 3. Since three points can determine a plane, so by pasting these nine reference points on the CT image, we can find the three-dimensional coordinates of any-point on the CT image. It is expected that the area where these nine reference points are located can be easily calculated by image residual calculation (MO coffee oglcal ope tender), using a 5x5 circular structure element, and then calculate the center point of each area as the reference point location, such as Figure 3 shows the positions of the reference points. In the N-type structure device, one prostitute rod and one oblique rod can be regarded as constituting two sides, and one image will cross the right three Ganyi M

Page 11 499308 5. Description of the invention (6), as shown in Figure 4. For the length of the reference point produced by the vertical end (length and hypotenuse), the length of the short side of the triangle α, so the height on the base ring of the three-dimensional frame can be obtained by formula (1), tan ^ (1 ) This. The N-type niche system can provide three points on a plane in space to define the direction and position of the plane. This -plane may be a coincident plane parallel to the stereo positioning system, or it may be the same as XY The plane produces forward and backward tilt4, side tilt, or front and back side tilt, as shown in Figure 5. Consider Q = image with nine reference points as shown in Figure 3. 2D screen coordinates * 4), and reference points (<X4'J ^), (X ...), ..., (Wi). The coordinate positioning procedure corresponding to the frame coordinates is obtained from the screen coordinates, which can be obtained by the following four steps.

Step 1 · Determine the zero point of the stereo positioning coordinate system by reference points A, C, D, F, G, and I (-^ 0 5 y〇 5 2slice) x0 + ½ + xG ten; ^ 6 (2)

6 • slice

£ i ± £ 2 + ^ 3 where is the height of each N-type device on the base ring,

499308 V. Description of the invention (7) ^ (xA-xB) 2+ (yA-yB) 2

tanS (3)

d (χE -xf) 2 + (ye -yf) 2 tmS z3 ^ j (xH -xG) 2 + (yH -yGf tan 5

Step 2. Determine the deviation angle a, 々, / a = tan i [(Zl -Z3> XA + xc XG-\-Xj, 2 2 /? = Tan_1 [(z2 1 • Less 0 ⑷ 2 γ = tan '- y〇 ~ yF X £)-X p

Step3. Determine the three-dimensional space coordinates (x3D, j; 3D, z3D) = (xs-x0) cos r + (ys-^ 〇) sin γyrei = (ys-y〇) G〇sr-(^ s- x〇) sin rx X3D = frel a (5) y ^ D = y rel C0S β ^ 3D = + ^ r, / sin ^ + xrelsun a

Step4. Determine the coordinates of the device in the Arcquadrant

lAO = X3D + βχ ⑹ y AQ = y ^ D + βγ 2 AO = Z3D + βζ where watts, eight, and A are the offset factors of the stereo positioning circular quadrant to the positioning system (B ias factors) °

Page 13 499308 V. Description of the invention (8) III. Alignment reference datum axis detection in the conversion month 丨) 'For the initial positions of dc (Anterior commissure) and PC (Postenor _miSSUre) on the CT image, It can be obtained by regiongrowing of image processing technology, and then confirmed by a professional physician. Connect the two points (ie AC-PC) to get the reference datum equation. 4. Atlas alignment with CT images In the work of computer atlases and CT images, two _ standard buttons are used. -This is the frame coordinate system on the standing mosquito bit system, which defines 4 (w) for each point on the CT image; second, for a target coordinate system, the computer map of the brain constructed by S㈣ka nd and Wahren The coordinates of each point are private, and), where the X-axis is such as the cut plane, the r-axis is the w-cut plane, and the z-axis is the secret cut plane. The alignment describes a point h on the system ‘a point 0 on the system. Xiao Xiao was selected from the computer map and converted to the frame coordinate system rUu through a coordinate conversion program. Therefore, the three-dimensional coordinates of XC (Wl, Z,) # ° PC (~ 从) can be regarded as known, then the P force and point represent two coordinate points that are known to have not shifted (that is, the situation Coordinate points, D represents especially the distance from p ”? (_ ~ L, the relationship is shown in Figure 6. Then Pj and Fan 4 can be deduced:

Page 14 499308

~ [(You plant,) 2 + (less 1 to less 2) 2 + (Z1 -Z2) 2] 1/2 ΡΑ ^ [(χλ-x2Uyx-y2 ^ zi ~ z2)] PA = [D, 0, 0] (7) PO = R4xPA = [NvN2, N3] where the dead point and its vertical point ^ is the outer product of 5 and ^ ?, which is its unit normal vector [Hm]. With the team, # 2,%] and the angle Θ between A and /, we have transformed the coordinate system Γ · (χ ', ηζ') of the computer map to the frame coordinate system Γ (Z, K, Z) on the CT image. The conversion formula is divided into four steps.

Stepl · Any point (x., Y, z) on the coordinate system r (X, r, Z) is translated to the origin TRl- 0 0-Xc X 0 1 0 -y〇y 0 0 1 a &amp; z 0 0 0 1 ⑻

Step2: Rotation amount of PO

499308 V. Description of the invention (ίο) where ~, · = 1 ·· 9 can be obtained from Θ.

Step3: Gain (Scale)

Sx 0 0 0 s = 0 Sy 0 0 0 0 Sz 0 0 0 0 1 (1〇) where (UV, A) are the gains in different slice directions for different patients.

Step4: Convert the coordinate system to Yayaina R R 2 0 0 1 0 y〇0 1 0 0 1 (11) Formally ⑻ to (11) we can get from the coordinate system ^. (You Ό) to get Z),

Page 16

499308

V. Explanation of the invention (π) In other words, through the formula (12), the computer map data can be mapped to the frame coordinate system to achieve the purpose of alignment. Figure 7 shows the graphical human-machine interface of the positioning button of the financial invention.

Table 1 shows the comparison between the method proposed in the present invention and the traditional method and clinical results (based on the 12.0mm section of the W series). Among them, the traditional method! It is assumed that the offset angle of the reference axis is zero, and then the corresponding axis scale of the map is adjusted, and then it is directly superimposed on the CT image. The traditional method 2 assumes that the offset angle of the reference axis is 0, and then applies the two-dimensional rotation formula X = xcos0 + less. Sin0 y = x * cos ^-^ * sin ^ element correction, and then adjust the corresponding axis scale of the map, Finally, it is directly superimposed with the ct image. As for the confirmation of the clinically superior target, the electrophysiological characteristics of the target can be verified. From the table, we find that the average difference of the proposed method is about 0.02 mm, which is superior to the traditional method and also confirms the effectiveness of this method. The brain mapping and CT image alignment method pioneered by the present invention is an important auxiliary tool in the surgical treatment of Parkinson's disease patients. It is a technology that manufacturers of stereotactic surgery medical software are eager to develop, so it is worth applying for patent protection.

499308

Without departing from the spirit and scope of the present invention, the scope of protection that can be regarded as a slight change and tears shall be defined by the scope of the attached patent application. To sum up, the present invention is original, novel, and progressive. Although the present invention is disclosed as above with some preferred embodiments, it is not intended to limit the present invention. Anyone skilled in this field is looking for a decoration. Therefore, the present invention is shown in Table 1. Si 12.0 mm Difference (mm) Lat AP Vert Clinical alignment Results 9.75 -0.86 -35,58 ~ traditional method 1 9.84 -0.76 -35.29 difference amount (mm) 0.09 0.1 0.29 ~~ 〇ΤΪ6 'traditional method 2 9.73 -0.93 -35.29 ----- J difference (mm) 0.02 0.07 0.29 Alignment result of this method 9.75 -0.92 -35.59 Difference (mm) 0.0 0.06 0.01 Simple diagram ^ --- ~~ --- Description of the table: Table 1: The method and traditional method and clinical Schematic illustration of the results obtained: Figure 1. Block diagram of the computer brain atlas and CT image alignment method. 2. CRW headband. (B) BRW-LR stereotactic frame. 3. The CT image will be generated by the stereo positioning frame. Nine reference points Figure 4 Cross-sectional view of CT image in N-type device ... Figure 5 Tilt angle (α, β, γ) of the stereotactic frame Figure 6: Position of AC and PC in the brain anatomy map Figure 6 (b Schematic diagram of angular deviation of the reference axis for computer tomography interface

Claims (1)

Methods for aligning electrical spectrum and CT images, including: and pre-processing of computer brain maps to generate 3D CT image data of the brain, spectrum; CT image miscellaneous, used to convert CT image coordinates to frame coordinate system; bit With reference to the base S-axis gamma, Xiao Yu provided GT images and 嶋 alignment;-maps were aligned with CT images, and the three-dimensional coordinate axes of the CT images were erected to transform the brain atlas to the frame coordinate system. 2. If the brain computer atlas and CT image alignment method of item 丨 of the patent application range, the pre-processing includes ... y '-brain electrospectrum data generation, _ with 3_pi * 3_pi or higher resolution The scanner converts the three-dimensional two-dimensional brain atlas data into a graphic file of 512,312 pixels. 3. As for the method of aligning brain computer atlas and CT image in the first item of the patent application, the linear gray-scale interpolation method used for preprocessing is as follows: rfv Μ — // 2 +1 γγι w X'y ) = μ +1 w ^^ + Μ + ΤΙη'λm = where M is the number of interpolated layers, 4 / χ, M is the grayscale value of the position 仏 ^ at the interpolated layer, and // 々M is the η The gray scale value of the relative position of the layer image. Do not set it to the gray scale value of the relative position of the 7th layer image. 4. If the brain computer atlas and CT image alignment method of item 2 of the patent application, the brain CT image positioning method is based on the 9 reference points of the 3N head frame on the CT image; set each coordinate point 2D screen coordinates are (心 凡), and the reference Patent application on page 20 &quot; One by one &quot; One-the coordinates of points (A to I) are (~, ^^, (~, heart), ((⑺)); use the following equations to achieve CT image coordinate conversion For the purpose of frame coordinates ... where 0 center is the frame coordinate of the target, (^, χ) is the screen coordinate of the target, (h'h) '(~, heart), ..., (χ /, α ) Are the screen coordinates of reference point (a to I). Jiu, eight, Α is the offset factor of the stereo positioning circular quadrant to the positioning system. XA0 = X3D ^ βχ y AO = y ^ D + βγ ZAQ = ZW + β2 X3D = Xrel y ^ o = yreiC0S ^ ^ 3D = + Vrei sin ^ + xrc / sin a Vrei ^^ s-Joicosr-C ^-^ 〇) sinr a = tan r (zx — Z3) y rXA + XC XG + XI 2 2 2 γ = tan ~ [one XF ^ / (~ one ~) 2+ (fork 4 — ^) 2 tan 5 ▲ χ £-xf) 2 + (y £-yf) 2 tmS yj (xH -xG) 2 + (yH -y〇f tan ^ 6 6 3. If the brain computer atlas and CT image alignment method of item 1 of the patent application range, the registration reference axis detection is obtained from the item 2 of the patent application range One group of CT scenes. First enlarge the original image by 4 times, and then use the image processing technology. Region growing method, automatic detection of the initial positions of AC (anterior commissures) and pc (p〇steri〇rc〇mmissures) on the CT image, and then confirmed by a professional physician, the two points are connected (ie A 〇ePC) to obtain the equation of the reference datum axis. 6. If the brain computer atlas and CT image alignment method of item 1 of the scope of patent application, the head frame detects nine reference points; first, the area where the nine reference points are located It can be easily obtained by image type calculation (using 5x5 circular structural elements), and then calculate the center point of each area as the position of the reference point, and use "+" to indicate the position of each reference point. 7 · If applying for a patent The method of aligning the brain computer atlas and radon image in the first item of the scope. The atlas is aligned with the CT image according to the position of the reference axis in the frame coordinate system using the following formula. 499308 Page 23