醫療用電腦輔助診斷系統(CAD, computer-aided detection)目前已廣泛應用在臨床醫學診斷上，醫師藉由這些輔助可立即發現病灶，也大幅減少判讀時間。醫學影像結合影像註冊系統可以獲得病人更完整的資訊，監測腫瘤的增長及治療驗證。影像註冊應用在醫學影像相當重要，如病理切片組織重組、3D重建、不同模組影像結合等，其運用相當廣泛。影像註冊可分為剛性對位和非剛性對位，大致上來說兩者差異在於剛性對位是整張影像變形，非剛性對位是局部影像變形。剛性註冊法因為條件限制無法處理較複雜影像，在染色病理切片組織影像經常會有拍攝上旋轉、位移、顏色不均、污漬和折疊等問題，而導致影像在對位時會因這些問題無法完成對位，所以本研究主要使用非剛性註冊技術，採用現有的技術有UnwapJ、BunwarpJ、SURF和TrakEM2。實驗將測試如果尚未加前處理影像，與改良過後新註冊方法差異。
在X光影像上常會有影像因為輻射劑量而無法讓影像顯示更清晰，每個人的臉頰厚度不一致，拍攝所需輻射量也不同，須靠經驗給定參數或是重複拍攝。另外影像處現今技術處理分為，Normalize和Normalize Histogram Equalization兩種，但是效果並不理想。
本研究貢獻了兩種前處理方式，改善影像註冊時所發生的問題。第一種針對病理切片使用兩種不同前處理增加影像對比度，在註冊時取用單一色彩通道減少影像註冊時錯位問題和減少雜訊。第二種針對X光影像，前處理使目標影像和來源影像對比達成一致性，改善曝光過度及太暗影像的問題。不但改善影像亮度問題，並且讓每張影像的結構顯示更清楚
病理切片改良方法將原本的影像註冊法準確率51%提升至91%，X光影像處理錯誤率從59%下降至11%。此改良技術可提供給醫生在手術前判斷及手術後觀察。醫學影像結合影像註冊系統可以獲得病人更完整的資訊，監測腫瘤的增長及治療驗證。提升了及早發現提早治療的成效，供給癌症病患第一時間治療。

Nowadays, the computer-aided detection (CAD) has been widely used in clinical diagnosis to assist physicians to find lesions more quickly and precisely. Registered images are proving useful in a range of applications, not only providing more correlative information to aid in diagnosis, but also assisting with the planning and monitoring of therapy, both surgery and radiotherapy. Its use of a wide range of image registration applications in medical imaging is very important, such as image reconstruction, pathological reorganization, and the different modules image combination. Essentially, the image registration can be divided into rigid registration and non-rigid registration, and the rigid registration is working on the global image deformation, the other is for local image deformation. In this study, two types of medical image data are considered, including serial histopathological slides and dental x-ray images. Regarding histopathological tissue images, complex deformations such as rotation, displacement, staining artifacts, bending and stretching often exist, making image registration a very challenging task. In this study, our aim is to investigate robust image registration methods dealing with complex deformation problems in medical data. Three state-of-the-art non-rigid registration technologies (UnwapJ, BunwarpJ, SURF and TrakEM2), which were demonstrated in various biological applications, were evaluated, but the experimental results show that they do not perform well and are not stable for both serial histopathological slides and x-ray images. .
The two pre-processing methods are introduced in this work to improve image registration methods in application to histopathological slides and x-ray images. The first method uses pre-processes to increase images contrasts, and it adopts a single color channel to reduce image registration dislocation and noise. In the second method, pre-processing of target image and source image can get same contrast ratio, and it can improve brightness problems.
The registration accuracy over the serial histopathological slides has risen from 51% to 91%, and the error rate on the X-ray images has fallen from 59% to 11%. The presented methods greatly reduce the time of manual intervention, assist early diagnosis and benefit patient healthcare.

[1] Yosi K., Amir A. (2006). Multisensor Image Registration via Implicit Similarity. IEEE Transactions on Pattern Analysis and Machine Intelligence, 28(5), 794-801.
[2] Irani M., Anandan P. (1998). Robust Multi-Sensor Image Alignment. IEEE International Conference on Computer Vision.
[3] Li H., Manjunath BS., Mitra S.K. (1995). “A Contour Based Approach to multi-sensor Image Registration”, IEEE Transactions. Image Processing, 4, 320-334.
[4] Maurer CR., Fitzpatrick JM. (1993). A review of medical image registration. American Association of Neurological, 17–44.
[5] Keller Y., Averbuch A. (2003). Implicit similarity: a new approach to multi-sensor image registration. Computer Vision and Pattern Recognition, 2, II- 543-8.
[6] Lavallee S. (1996). Registration for computer-integrated surgery methodology, Computer-integrated surgery, technology and clinical applications (pp.77–97). Cambridge: MIT Press.
[7] McInerney T., Terzopoulos D. (1996). Deformable models in medical image analysis: a survey. Medicine Image Analysis, 1(2), 91–108.
[8] Davatzikos et al. (1996). Image registration based on boundary mapping. IEEE Transactions Medical Image,15(1), 112–115.
[9] Hawkes DJ. (1998). Algorithms for radiological image registration and their clinical application. Journal of Anatomy, 193, 347–361.
[10] Audette MA., Ferrie FP., Peters TM. (2000). An algorithmic overview of surface registration techniques for medical imaging. Medicine Image Analysis, 4(3), 201–217.
[11] Eberl S., Braun M. (1999). Intra- and inter-modality registration of functional and anatomical clinical images. In: Pham B., Braun M., Maeder AJ., Eckert MP (Eds.), New approaches in medical image analysis (pp.102–114). Bellingham: SPIE Press.
[12] Thurfjell L., Pagani M., Andersson JLR. (2000). Registration of neuroimaging data: implementation and clinical applications. Journal of Neuroimaging, 10, 39–46.
[13] Viergever MA et al. (2001). Registration, segmentation, and visualization of multimodal brain images. Comput Medical Imaging Graph, 25, 147–151.
[14] Hill DL., Batchelor PG., Holden M., Hawkes DJ. (2001). Medical image registration. Physics in Medicine and Biology, 46(3), 1–45.
[15] Surova-Trojanova H., Barker WC., Carrasquillo JA., Bacharach SL. (2000) Registration of planar emission images with reprojected CT data. Journal of Nuclear Medicine, 41(4), 700–705.
[16] Anuta PE. (1970). Spatial Registration of Multispectral and Multitemporal Digital Imagery Using Fast Fourier Transform Techniques. Geoscience Electronics. IEEE Transactions on, 8(4), 353-368.
[17] Brown LG. (1992). A survey of image registration techniques. ACM Computing Surveys , 24(4), 326–376.
[18] Sanchez Sorzano C.O., Thevenaz P., Unser M. (2005). Elastic Registration of Biological Images Using Vector-Spline Regularization. IEEE Transactions on Biomedical Engineering, 52(4), 652-663.
[19] Ignacio A et al. (2006). Consistent and Elastic Registration of Histological Sections using Vector-Spline Regularization. Lecture Notes in Computer Science, Springer Berlin / Heidelberg, CVAMIA: Computer Vision Approaches to Medical Image Analysis, 4241, 85-95.
[20] Herbert B., Andreas E., Tinne T., Luc VG. (2008). SURF: Speeded Up Robust Features. Computer Vision and Image Understanding (CVIU), 110( 3), 346-359.
[21] Cardona A., Saalfeld S., Schindelin J, Arganda-Carreras I., Preibisch S. (2012). TrakEM2 Software for Neural Circuit Reconstruction. PLoS ONE 7(6) [22] Saalfeld S., Fetter R., Cardona A., Tomancak P. (2012). Elastic volume reconstruction from series of ultra-thin microscopy sections. Nature Methods.
[23] Saalfeld S., Cardona A., Hartenstein V., Tomancak P. (2010). As-rigid-as- possible mosaicking and serial section registration of large ssTEM datasets. Bioinformatics, 26(12), 57-63.
[24] Bajcsy R. (2003). Digital Anatomy Atlas and Its Registration to MRI, fMRI,PET: The Past Presents a Future. Lecture Notes in Computer Science, 2717, 201–211.
[25] Slomka PJ., Hurwitz GA., Stephenson J., Cradduck T. (1995). Automated alignment and sizing of myocardial stress and rest scans to three-dimensional normal templates using an image registration algorithm. Journal of Nuclear Medicine, 36(6), 1115–1122.
[26] Unser M. (1999). Splines: A perfect fit for signal and image processing. IEEE Signal Process, 16(6), 22–38.
[27] Meijering E., Niessen W., Viergever M. (2001). Quantitative evaluation of convolution-based methods for medical image interpolation. Medicine Image Analysis, 5(2),111–126.
[28] Thevenaz P., Blu T., Unser M. (2000). Interpolation revisited. IEEE Transactions on Medical Image, 19(7), 739–758.
[39] Szeliski R., Coughlan J. (1997). Spline-based image registration. International Journal of Computation Vision, 22( 3), 199–218.
[30] West J., Fitzpatrick JM., Wang MY. (1997). Comparison and evaluation of retrospective intermodality image registration techniques. Journal of Computer Assisted Tomography, 21(4), 554–566.
[31] Barnden L., Kwiatek R., Lau Y., Hutton B., Thurfjell L et al. (2000). Validation of fully automatic brain SPET to MR coregistration. European journal of nuclear medicine, 27:147–154.
[32] Woods RP., Grafton ST., Holmes CJ., Cherry SR., Maziotta JC et al. (1998). Automated image registration. i. General methods and intrasubject, intramodality validation. Journal of Computer Assisted Tomography , 22(1), 139–152.
[31] Rohlfing T., Maurer CR., Bluemke DA., Jacobs MA. (2003). Volume-preserving nonrigid registration of MR breast images using free-form
deformation with an incompressibility constraint. Transactions on Medical Image,22(6), 730–741.
[32] Kybic J., Thevenaz P., Unser M. (1999). Compensation of unidirectional geometric distortion in EPI using spline warping. IEEE International Conference Image Processing , 2,168–172.
[33] Henn S., Witsch K. (2000). A multigrid approach for minimizing a non-linear functional for digital image matching. Computing, 64(4), 339–348.
[34] Rueckert D., Sonoda LI., Hayes C., Hill DLG., Leach MO et al. (1999). Non-rigid registration using free-form deformations: Application to breast MR images. IEEE Transactions on Medical Image, 18(8), 712–721.
[35] Bazen AM, Gerez SH. (2003). Fingerprint matching by thin-plate spline modeling of elastic deformations. Journal of the Pattern Recognition, 36(8), 1859–1867.
[36] Rohr K., Stiehl HS., Sprengel R., Buzug TM., Weese J. et al. (2001). Landmark-based elastic registration using approximating thin-plate splines. IEEE Transactions on Medical Image, 20(6), 526–534.
[37] Bookstein FL. (1989). Principal warps: Thin-plate splines and the decomposition of deformations. IEEE Transactions Pattern Analysis and Machine Intelligence, 11(6), 567–585.
[37] Amodei L., Benbourhim MN. (1991). A vector spline approximation. Journal of Approximation Theory, 67(1), 51–79.
[38] Lindeberg T. (1998). Feature detection with automatic scale selection. International Journal of Computer Vision, 30(2), 79 – 116.
[39] Lowe D. (2004). Distinctive image features from scale-invariant keypoints, cascade filtering approach. International Journal of Computer Vision, 60(2)91–110.
[40] Mikolajczyk K., Schmid C. (2002). An affine invariant interest point detector. European Conference on Computer Vision, 128 – 142.
[41] Se S., Ng H.K., Jasiobedzki P., Moyung TJ. (2004). Vision based modeling and localization for planetary exploration rovers. Proceedings of International Astronautical Congress.
[42] Tuytelaars T., Van Gool L. (2000). Wide baseline stereo based on local, affinely invariant regions. British Machine Vision Conference, 412 – 422.
[43] Viola P.A., Jones M.J. (2001). Rapid object detection using a boosted cascade of simple features. Computer Vision and Pattern Recognition, 511–518.
[44] Mikolajczyk K., Schmid C. (2001). Indexing based on scale invariant interest points. In International Conference on Computer Vision, 1, 525–531.
[45] Lindeberg T. (1998). Feature detection with automatic scale selection. International Journal of Computer Vision, 30(2), 79–116.
[46] Suloway C. et al. (2005). Automated molecular microscopy: the new Leginon system. Journal of Structural Biology, 151, 41–60.
[47] Brown M., Lowe DG. (2003). Recognising panoramas. IEEE International Conference on Computer Vision, 1218–1225.
[48] Mikolajczyk K., Schmid C. (2003). A performance evaluation of local descriptors. Conference on Computer Vision and Pattern Recognition, 2, 257–263.