亚洲免费av电影一区二区三区,日韩爱爱视频,51精品视频一区二区三区,91视频爱爱,日韩欧美在线播放视频,中文字幕少妇AV,亚洲电影中文字幕,久久久久亚洲av成人网址,久久综合视频网站,国产在线不卡免费播放

        ?

        2D-3D配準(zhǔn)中的初始位姿估計方法

        2022-11-01 08:01:34劉傳耙宋軼民馬信龍
        關(guān)鍵詞:實驗方法

        孫?濤,郭?可,劉傳耙,張?弢,宋軼民, 3,馬信龍

        2D-3D配準(zhǔn)中的初始位姿估計方法

        孫?濤1,郭?可1,劉傳耙1,張?弢2,宋軼民1, 3,馬信龍2

        (1.天津大學(xué)機構(gòu)理論與裝備設(shè)計教育部重點實驗室,天津 300354;2. 天津市天津醫(yī)院,天津 300211;3. 天津仁愛學(xué)院機械工程系,天津 301636)

        手術(shù)導(dǎo)航;2D-3D配準(zhǔn);初始位姿估計;正交雙平面投影

        在骨科手術(shù)中,通常采用成像較快的熒光透視圖像為醫(yī)生提供術(shù)中手術(shù)器械與病患部位的相對位置信息[1-2],但術(shù)中二維圖像難以準(zhǔn)確反映病患部位的空間位置和解剖信息,因此需要與術(shù)前三維CT數(shù)據(jù)結(jié)合,即通過2D-3D圖像配準(zhǔn)建立2D像素平面與3D模型空間的統(tǒng)一,實現(xiàn)術(shù)中、術(shù)前圖像在空間位置與解剖紋理上的對應(yīng)關(guān)系,從而引導(dǎo)手術(shù)的進行,這是病灶精準(zhǔn)定位和施術(shù)成功的關(guān)鍵[3].

        綜上,本文提出一種新的初始位姿估計方法,通過正交融合投影原理,構(gòu)建空間與平面參數(shù)間的映射關(guān)系,結(jié)合區(qū)域生長算法完成對多研究對象的5個三維空間參數(shù)的有效估計,以增大2D-3D配準(zhǔn)的初始變換范圍、降低配準(zhǔn)時間、提高配準(zhǔn)的精度和魯棒性,并進行實驗驗證.

        1?初始位姿估計方法

        1.1?參數(shù)映射關(guān)系

        大部分2D-3D醫(yī)學(xué)圖像配準(zhǔn)是將三維CT數(shù)據(jù)通過數(shù)字重建影像技術(shù)生成DRR(digitally reconstructed radiograph)圖像,比較DRR圖像與術(shù)中平面圖像的相似性,再不斷改變投影環(huán)境下三維物體的空間參數(shù),使得在某個空間參數(shù)下投影生成的DRR圖像與平面圖像的相似性達到最大[17],此時的空間參數(shù)就是相同相機坐標(biāo)投影環(huán)境下生成平面圖像的三維物體的位姿.因為三維物體的空間位姿的改變會引起DRR圖像的改變,所以可建立空間與平面參數(shù)間的映射關(guān)系.

        首先通過模擬術(shù)中C臂成像設(shè)備的成像過程,構(gòu)建相應(yīng)的投影環(huán)境來生成DRR圖像,其中正交融合投影過程的示意如圖1所示.所構(gòu)建的投影坐標(biāo)系原點為三維影像中心點,其坐標(biāo)軸與經(jīng)CT掃描后所成圖像自帶的坐標(biāo)軸平行.

        圖1?正交融合投影示意

        式(2)計算可得

        其中

        根據(jù)上述推導(dǎo)過程,可得空間與平面參數(shù)間的映射關(guān)系,但平面的參數(shù)信息包含在DRR圖像灰度信息之中,不能直接提取,因此需要在后續(xù)利用對應(yīng)目標(biāo)圖像與正交雙平面模板間的配準(zhǔn)來獲取平面參數(shù)信息.

        1.2?初始位姿估計原理與方法

        圖2?初始位姿估計原理

        然后,根據(jù)需要配準(zhǔn)的目標(biāo)投影生成正交雙平面模板,將目標(biāo)圖像和模板的灰度信息進行平面配準(zhǔn),計算得到正側(cè)位平面配準(zhǔn)參數(shù).本文采用歸一化互相關(guān)[20]作為相似性測度來計算初始位姿估計以及后續(xù)2D-3D配準(zhǔn)過程的平面圖像相似性.其中,歸一化互相關(guān)系數(shù)的計算過程可以表示為

        2?實?驗

        2.1?實驗數(shù)據(jù)與測試環(huán)境

        本文使用公開的顱骨數(shù)據(jù)[19]和天津醫(yī)院提供的完整股骨、包含上下兩骨塊(近端骨和遠端骨)的骨折股骨的CT數(shù)據(jù)來驗證上文提出的初始位姿估計方法.為了方便獲得任意位姿下的平面影像,本實驗采用DRR影像模擬術(shù)中熒光圖像進行配準(zhǔn).因為熒光圖像成像時,成像設(shè)備發(fā)出的X射線透過病患部位后強度會減少,不同的組織結(jié)構(gòu)造成的X射線衰減程度有差異,這種差異在熒光圖像上顯示為各組織結(jié)構(gòu)明暗的不同.而DRR影像成像原理是模擬這種透視過程,故模擬生成的熒光影像與術(shù)中熒光影像具有一致性,這樣不僅減少了患者和醫(yī)生受到的輻射,也可以將圖像生成的真實位姿作為配準(zhǔn)結(jié)果的評估標(biāo)準(zhǔn),具有可重復(fù)性.同時在實際應(yīng)用中,本文的數(shù)據(jù)處理算法對于術(shù)中熒光影像與DRR圖像的數(shù)據(jù)差異具有魯棒性.

        本實驗使用Visual Studio 2017作為軟件平臺,結(jié)合圖像開發(fā)工具包ITK(insight segmentation and registration toolkit)編程實現(xiàn)初始位姿估計與2D-3D配準(zhǔn)過程,電腦配置為Intel?Xeon?Gold 5120 CPU @2.20GHz處理器,內(nèi)存為64GB.

        2.2?評估方法

        以投影生成仿術(shù)中平片時三維物體的空間位姿作為“金標(biāo)準(zhǔn)”,用生成的匹配模板進行初始位姿估計,統(tǒng)計估計值與實際真值的誤差和不需要人工校正的比例.因為在2D-3D圖像配準(zhǔn)中使用雙平面方法的配準(zhǔn)精度要優(yōu)于單平面[22],且目前多采用Powell優(yōu)化方法[23]進行配準(zhǔn),所以本文用估計的參數(shù)值作為起始搜索位姿,采用雙平面Powell法進行2D-3D配準(zhǔn)(以下簡稱結(jié)合法),將得到的結(jié)果與不使用初始位姿估計的雙平面Powell法(以下簡稱單一法)進行配準(zhǔn)的結(jié)果進行統(tǒng)計,比較最終配準(zhǔn)結(jié)果與實際真值的誤差.

        2.3?實驗結(jié)果與分析

        表1?初始位姿估計結(jié)果

        Tab.1?Results of initial pose estimation

        顱骨實驗中,單一法與結(jié)合法都有較好的配準(zhǔn)結(jié)果,因此統(tǒng)計每次迭代后所有平移參數(shù)和所有旋轉(zhuǎn)參數(shù)配準(zhǔn)誤差絕對數(shù)的平均值(以下分別簡稱為平移配準(zhǔn)誤差和旋轉(zhuǎn)配準(zhǔn)誤差)并進行繪制,如圖3所示;在完整股骨、骨折股骨近端骨和遠端骨的實驗中,有10%的實驗配準(zhǔn)結(jié)果嚴重偏離“金標(biāo)準(zhǔn)”,且全來自于單一法,因此本文僅對剩余實驗結(jié)果進行統(tǒng)計,繪制每次迭代后的平移配準(zhǔn)誤差和旋轉(zhuǎn)配準(zhǔn)誤差圖,如圖4~圖6所示.由于大部分實驗經(jīng)過6次迭代后,配準(zhǔn)結(jié)果趨于相對穩(wěn)定,因此只統(tǒng)計各實驗前6次迭代過程中的配準(zhǔn)誤差.

        圖3?顱骨實驗中每次迭代的配準(zhǔn)誤差

        圖4?股骨實驗中每次迭代的配準(zhǔn)誤差

        圖5?近端骨實驗中每次迭代的配準(zhǔn)誤差

        圖6?遠端骨實驗中每次迭代后的配準(zhǔn)誤差

        各研究對象配準(zhǔn)統(tǒng)計結(jié)果表明,使用單一法進行2D-3D配準(zhǔn)時,平均迭代次數(shù)為7.5次,平均配準(zhǔn)時間為8089s;而基于結(jié)合法的2D-3D配準(zhǔn)平均迭代次數(shù)為3次,平均配準(zhǔn)時間(包括初始位姿估計時間)為4330s.與單一法相比,結(jié)合法在平均迭代次數(shù)上降低60%,平均配準(zhǔn)時間上縮短46.5%,并且具有精度高、魯棒性好的優(yōu)點,最終2D-3D配準(zhǔn)結(jié)果中各參數(shù)的配準(zhǔn)誤差的平均值±標(biāo)準(zhǔn)差統(tǒng)計結(jié)果如表2?所示.

        表2?2D-3D參數(shù)配準(zhǔn)誤差

        Tab.2?2D/3D parameter registration errors

        圖7?各實驗的參數(shù)配準(zhǔn)誤差精度提升率

        3?結(jié)?論

        本文提出了一種正交雙平面模板匹配初始位姿估計方法,該方法利用正交融合投影原理構(gòu)建了正交投影下三維空間參數(shù)與平面圖像坐標(biāo)參數(shù)間的映射關(guān)系,將已知位姿投影下的DRR圖像作為匹配模板與待配準(zhǔn)的圖像進行比較,通過平面坐標(biāo)參數(shù)的變化來對空間參數(shù)值進行估計,并且基于區(qū)域生長算法實現(xiàn)多個研究對象的配準(zhǔn).為了驗證本文提出的初始位姿估計方法的有效性,采用了顱骨、完整股骨、骨折股骨的近端骨和遠端骨作為研究對象,各目標(biāo)通過20組實驗進行初始位姿估計與后續(xù)2D-3D配準(zhǔn)實驗對比.得到全文結(jié)論如下.

        (2) 在2D-3D配準(zhǔn)實驗中,相較于單一法,結(jié)合法在整體配準(zhǔn)時間上平均縮短46.5%,在各實驗中單個平移參數(shù)精度最高提升46.5%,單個旋轉(zhuǎn)參數(shù)精度最高提升91.1%,同時實驗結(jié)果也兼具了良好的魯?棒性.

        (3) 通過各組實驗,驗證了在較大的初始位姿偏差下,本文提出的初始位姿估計方法對不同研究對象都具有很好的適應(yīng)性,這也將有助于為骨科手術(shù)中的多模影像導(dǎo)航系統(tǒng)提供最佳或接近最佳的搜索參數(shù)選擇.

        [1] Russakoff D B,Rohlfing T,Mori K,et al. Fast generation of digitally reconstructed radiographs using attenuation fields with application to 2D-3D image registration[J]. IEEE Transactions on Medical Imaging,2005,24(11):1441-1454.

        [2] Khamene A,Bloch P,Wein W,et al. Automatic registration of portal images and volumetric CT for patient positioning in radiation therapy[J]. Medical Image Analysis,2006,10(1):96-112.

        [3] Markelj P,Tomazevic D,Likar B,et al. A review of 3D/2D registration methods for image-guided interventions[J]. Medical Image Analysis,2012,16(3):642-661.

        [4] Frederik M,Vandermeulen D,Paul S,et al. Comparative evaluation of multiresolution optimization strategies for multimodality image registration by maximization of mutual information[J]. Medical Image Analysis,1999,3(4):373-386.

        [5] Kaiser M,John M,Heimann T,et al. 2D/3D registration of TEE probe from two non-orthogonal c-arm directions[C]//International Conference on Medical Imaging Computing and Computer-Assisted Intervention. Boston,USA,2014:283-290.

        [6] Muhit A,Pickering M R,Ward T,et al. A comparison of the 3D kinematic measurements obtained by single-plane 2D-3D image registration and RSA[C]//32nd Annual International Conference of the IEEE Engineering in Medicine and Biology Society. Buenos Aires,Argentina,2010:6288-6291.

        [7] Miao S,Wang Z J,Liao R. A convolutional neural network approach for 2D/3D medical image registration[J]. Optoelectronics and Advanced Materials-Rapid Communications,2015,4(3):636-648.

        [8] Pickering M R,Muhit A,Scarvell J M,et al. A new multi-modal similarity measure for fast gradient-based 2D-3D image registration[C]//Annual International Conference of the IEEE Engineering in Medicine and Biology-Society. Minneapolis,USA,2009:5821-5824.

        [9] Akter M,Lambert A J,Pickering M R,et al. Robust initialisation for single-plane 3D CT to 2D fluoroscopy image registration[J]. Computer Methods in Biomechanics and Biomedical Engineering:Imaging & Visualization,2015,3(3):147-171.

        [10] Livyatan H,Yaniv Z,Joskowicz L. Gradient-based 2D/3D rigid registration of fluoroscopic X-ray to CT[J]. IEEE Transactions on Medical Imaging,2003,22(11):1395-1406.

        [11] Van der Bom M J,Bartels L W,Gounis M J,et al. Robust initialization of 2D-3D image registration using the projection-slice theorem and phase correlation[J]. Medical Physics,2010,37(4):1884-1892.

        [12] Andreas V S,Tom C,Graeme P. Increasing the automation of a 2D-3D registration system[J]. IEEE Transactions on Medical Imaging,2013,32(2):387-399.

        [13] Akter M,Lambert A J,Pickering M R,et al. Improved robustness of 3D CT to 2D fluoroscopy image registration using log polar transforms[C]//International Conference on Electrical. Rajshahi,Bangladesh,2016:1-5.

        [14] Miao S,Wang Z J,Liao R. A CNN regression approach for real-time 2D/3D registration[J]. IEEE Transactions on Medical Imaging,2016,35(5):1352-1363.

        [15] Piat S,Miao S,Liao R,et al. Dilated Fully Convolutional Network for Multi-Agent 2D/3D Medical Image Registration:US,2019050999A1[P]. 2019-02-14.

        [16] Litjens G,Kooi T,Bejnordi B E,et al. A survey on deep learning in medical image analysis[J]. Medical Image Analysis,2017,7(5):60-88.

        [17] 王?雷. 影像導(dǎo)航手術(shù)中2D/3D圖像配準(zhǔn)[D]. 長春:中國科學(xué)院,2015.

        Wang Lei. 2D/3D Image Registration in Image-Guided Surgery[D]. Changchun:Chinese Academy of Sciences,2015(in Chinese).

        [18] 孫?濤,閆?巍,孫振輝,等. 外固定支架模型重建軟件開發(fā)與安裝參數(shù)識別研究[J]. 天津大學(xué)學(xué)報(自然科學(xué)與工程技術(shù)版),2020,53(6):593-600.

        Sun Tao,Yan Wei,Sun Zhenhui,et al. Software development and pose acquisition of the Taylor spatial frame[J]. Journal of Tianjin University(Science and Technology),2020,53(6):593-600(in Chinese).

        [19] Akter M,Lambert A J,Pickering M R,et al. A 2D-3D image registration algorithm using log-polar transforms for knee kinematic analysis[C]//International Conference of the IEEE Engineering in Medicine and Biology Society. Fremantle,Australia,2012:1-5.

        [20] Wu J,Kim M,Peters J,et al. Evaluation of similarity measures for use in the intensity-based rigid 2D-3D registration for patient positioning in radiotherapy[J]. Medical Physics,2009,36(12):5391-5403.

        [21] Zhu Z,Li G. An automatic 2D-3D image matching method for reproducing spatial knee joint positions using single or dual fluoroscopic images[J]. Computer Methods in Biomechanics and Biomedical Engineering,2012,15(11):1245-1256.

        [22] Zhu Z,Massimini D F,Wang G,et al. The accuracy and repeatability of an automatic 2D-3D fluoroscopic image-model registration technique for determining shoulder joint kinematics[J]. Medical Engineering & Physics,2012,34(9):1303-1309.

        [23] Ding H Y,Bian Z F. A sub-pixel registration approach based on powell algorithm[J]. Acta Photonica Sinica,2009,38(12):3322-3327.

        Initial Pose Estimation Method in 2D/3D Registration

        Sun Tao1,Guo Ke1,Liu Chuanba1,Zhang Tao2,Song Yimin1, 3,Ma Xinlong2

        (1. Key Laboratory of Mechanism Theory and Equipment Design of Ministry of Education,Tianjin University,Tianjin 300354,China;2. Tianjin Hospital,Tianjin 300211,China;3. Department of Mechanical Engineering,Tianjin Renai College,Tianjin 301636,China)

        In this paper,we propose a new initial pose estimation approach based on parameter mapping and image segmentation,focusing on the problems of small initial parameter transformation range and single registration object in the 2D/3D medical image registration process. This approach expands the range of initial motion deviation of 3D object spatial parameters and solves the issues regarding few estimable parameters and multibone image registration difficulty of the surgical navigation system during the registration process. First,the region growth algorithm is used to segment the multibone image and extract the target of interest to complete the target-data acquisition to be registered. Second,a projection imaging model is constructed based on the principle of orthogonal fusion projection to decompose the spatial rigid body transformation parameters into the positive lateral plane and establish the mapping relationship between spatial and plane parameters. Then,the orthogonal biplane template generated using the projection is matched with the corresponding target image,and the registration parameters of the obtained positive and lateral plane are converted into spatial parameters,to achieve the effective estimation of the initial pose of the 3D object. The method was verified based on computed tomography data of the skull,intact femur,and fractured femur. Then,the method was compared with the traditional 2D/3D image registration method. The experimental results showed that the proposed method could make the initial transformation range of the five spatial parameters reach ?±30mm or ±20°,and different research objects within this range show a good initial pose estimation effect. Compared with the traditional image registration method,the overall registration time of the 2D/3D iterative optimization based on the above method is 46.5% shorter on average,the registration accuracy of a single translation parameter is up to 69.2%,and the registration accuracy of a single rotation parameter is up to 91.1%. Simultaneously,the final registration results have adequate robustness.

        surgical navigation;2D/3D registration;initial pose estimation;orthogonal biplane projection

        the National Key Research and Development Program of China(No.2018YFB1307800),the National Natural Science Foundation of China(No.51775367),Tianjin Science and Technology Plan Project(No.18PTLCSY00080,No.20201193,No.18YFSDZC00010).

        10.11784/tdxbz202010007

        TP391

        A

        0493-2137(2022)02-0143-08

        2020-10-07;

        2021-02-25.

        孫?濤(1983—??),男,博士,教授,stao@tju.edu.cn.Email:m_bigm@tju.edu.cn

        宋軼民,ymsong@tju.edu.cn.

        國家重點研發(fā)計劃資助項目(2018YFB1307800);國家自然科學(xué)基金資助項目(51775367);天津市科技計劃資助項目(18PTLCSY00080,20201193,18YFSDZC00010).

        (責(zé)任編輯:王曉燕)

        猜你喜歡
        實驗方法
        記一次有趣的實驗
        微型實驗里看“燃燒”
        做個怪怪長實驗
        學(xué)習(xí)方法
        可能是方法不對
        NO與NO2相互轉(zhuǎn)化實驗的改進
        實踐十號上的19項實驗
        太空探索(2016年5期)2016-07-12 15:17:55
        用對方法才能瘦
        Coco薇(2016年2期)2016-03-22 02:42:52
        四大方法 教你不再“坐以待病”!
        Coco薇(2015年1期)2015-08-13 02:47:34
        賺錢方法
        开心久久婷婷综合中文字幕| 久久视频在线| 亚洲va中文字幕无码久久不卡| 91精品全国免费观看青青| 亚洲午夜精品国产一区二区三区| 日韩人妻中文字幕专区| 久久精品国产成人| 国产精品久久久久久久免费看 | 隔壁人妻欲求不满中文字幕| 97丨九色丨国产人妻熟女| 无码人妻一区二区三区在线视频| a国产一区二区免费入口| 亚洲高清视频在线播放| 日韩美女av二区三区四区| 亚洲一区二区三区av无| av网站在线观看入口| 成人久久久久久久久久久| 国产精品一区二区久久乐下载 | 中文有码无码人妻在线| 精品人妻人人做人人爽夜夜爽| 日本口爆吞精在线视频| 国产成人综合亚洲国产| 国产偷国产偷亚洲高清视频| 一本一道av中文字幕无码| 亚洲自拍愉拍| 亚洲成在人线天堂网站| 亚洲精品1区2区在线观看| 综合色就爱涩涩涩综合婷婷| 成人无码视频| 亚洲天堂av社区久久| 97人妻中文字幕总站| 国产熟妇与子伦hd| 国产微拍精品一区二区| 国产成人自拍视频在线免费| 国产精品无套一区二区久久| 少妇无码av无码专区| 2021国产精品久久| 视频一区视频二区自拍偷拍| 国产精品免费一区二区三区四区 | 国产麻豆久久av入口| 男人j进女人j啪啪无遮挡|