COLLINARUS: collection of image-derived non-linear attributes for registration using splines
Jonathan Chappelow, B Bloch, Neil M. Rofsky, Elizabeth M. Genega, Robert E. Lenkinski, William C. DeWolf, Satish E. Viswanath, Anant Madabhushi · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2009
ABSTRACT We present a new method for fully automatic non-rigid registration of multimodal imagery, including structuraland functional data, that utilizes multiple texutral feature images to drive an automated spline based non-linearimage registration procedure. Multimodal image registration is signicantly more complicated than registrationof images from the same modality or protocol on account of diculty in quantifying similarity between dierentstructural and functional information, and also due to possible physical deformations resulting from the dataacquisition process. The COFEMI technique for feature ensemble selection and combination has been previouslydemonstrated to improve rigid registration performance over intensity-based MI for images of dissimilar modali-ties with visible intensity artifacts. Hence, we present here the natural extension of feature ensembles for drivingautomated non-rigid image registration in our new technique termed Collection of Image-derived Non-linearAttributes for Registration Using Splines (COLLINARUS). Qualitative and quantitative evaluation of the COL-LINARUS scheme is performed on several sets of real multimodal prostate images and synthetic multiprotocolbrain images. Multimodal (histology and MRI) prostate image registration is performed for 6 clinical data setscomprising a total of 21 groups of in vivo structural (T2-w) MRI, functional dynamic contrast enhanced (DCE)MRI, and ex vivo WMH images with cancer present. Our method determines a non-linear transformation toalign WMH with the high resolution in vivo T2-w MRI, followed by mapping of the histopathologic cancer extentonto the T2-w MRI. The cancer extent is then mapped from T2-w MRI onto DCE-MRI using the combinednon-rigid and ane transformations determined by the registration. Evaluation of prostate registration is per-formed by comparison with the 3 time point (3TP) representation of functional DCE data, which provides anindependent estimate of cancer extent. The set of synthetic multiprotocol images, acquired from the BrainWebSimulated Brain Database, comprises 11 pairs of T1-w and proton density (PD) MRI of the brain. Following theapplication of a known warping to misalign the images, non-rigid registration was then performed to recover theoriginal, correct alignment of each image pair. Quantitative evaluation of brain registration was performed bydirect comparison of (1) the recovered deformation eld to the applied eld and (2) the original undeformed andrecovered PD MRI. For each of the data sets, COLLINARUS is compared with the MI-driven counterpart of theB-spline technique. In each of the quantitative experiments, registration accuracy was found to be signicantly(p< 0.05) for COLLINARUS compared with MI-driven B-spline registration. Over 11 slices, the mean absoluteerror in the deformation eld recovered by COLLINARUS was found to be 0.8830 mm.Keywords: quantitative imageanalysis, magneticresonance, whole-mounthistology, imageregistration,prostate,cancer, B-splines, COFEMI, hierarchical, non-rigid