Abstract
Images are often considered as functions defined on the image domains, and as functions, their (intensity) values are usually considered to be invariant under the image domain transforms. This functional viewpoint is both influential and prevalent, and it provides the justification for comparing images using functional \(\mathbf {L}^p\)-norms. However, with the advent of more advanced sensing technologies and data processing methods, the definition and the variety of images has been broadened considerably, and the long-cherished functional paradigm for images is becoming inadequate and insufficient. In this paper, we introduce the formal notion of covariant images and study two types of covariant images that are important in medical image analysis, symmetric positive-definite tensor fields and Gaussian mixture fields, images whose sample values covary i.e., jointly vary with image domain transforms rather than being invariant to them. We propose a novel similarity measure between a pair of covariant images considered as embedded shapes (manifolds) in the ambient space, a Cartesian product of the image and its sample-value domains. The similarity measure is based on matching the two embedded low-dimensional shapes, and both the extrinsic geometry of the ambient space and the intrinsic geometry of the shapes are incorporated in computing the similarity measure. Using this similarity as an affinity measure in a supervised learning framework, we demonstrate its effectiveness on two challenging classification problems: classification of brain MR images based on patients’ age and (Alzheimer’s) disease status and seizure detection from high angular resolution diffusion magnetic resonance scans of rat brains.
Similar content being viewed by others
Notes
A word of caution here would serve in avoiding any confusion further down in the rest of this paper. In Computer Vision and Statistics, the term covariance is commonly used to indicate the second order statistic and should not be confused with the term we use here namely, “covariant”. Further, in the field of tensor algebra, the terms, covariant and contra-variant tensors are commonly used and indicate the way these tensors transform under coordinate transformations. In this paper, covariant images simply refers to images whose value set covaries or jointly varies with transformations applied to the domain.
A similar observation is also made in Sochen et al. (1998).
References
Alexander, D. C., Pierpaoli, C., Basser, P. J., & Gee, J. C. (2001). Spatial transformations of diffusion tensor magnetic resonance images. IEEE Transactions on Medical Imaging, 20(11), 1131–1139.
Ashburner, J., & Friston, K. J. (1999). Nonlinear spatial normalization using basis functions. Human Brain Mapping, 7, 254–266.
Baker, S., Scharstein, D., Lewis, J. P., Roth, S., Black, M., & Szeliski, R. (2011). A database and evaluation methodology for optical flow. International Journal of Computer Vision, 92, 1–31.
Barmpoutis, A., Vemuri, B. C., Shepherd, T. M., & Forder, J. R. (2007). Tensor splines for interpolation and approximation of DT-MRI with applications to segmentation of isolated rat hippocampi. IEEE Transaction on Medical Imaging, 26(11), 1537–1546.
Bates, D. M., Lindstrom, M. J., Wahba, G., & Yandell, B. S. (1987). GCVPACK-Routines for generalized cross validation. Communications in Statistics—Simulation and Computation, 16(1), 263–297.
Beg, M. F., Miller, M. I., Trouvé, A., & Younes, L. (2005). Computing large deformation metric mappings vis geodesic flows of diffeomorphisms. International Journal of Computer Vision, 61(2), 139–157.
Cachier, P., & Rey, D. (2000). Symmetrization of the non-rigid registration problem using inversion-invariant energies: Application to multiple sclerosis. International conference on medical image computing and computer assisted intervention, MICCAI (pp. 472–481).
Cachier, P., Bardinet, E., Mormont, D., Pennec, X., & Ayache, N. (2003). Iconic feature based nonrigid registration: The PASHA algorithm. Computer Vision and Image Understanding, 89(2–3), 272–298.
Cao, Y., Miller, M.I., Mori, S., Winslow, R.L., & Younes, L. (2006). Diffeomorphic matching of diffusion tensor images, Proceedings CVPR (pp. 17–22).
Chen, T., Rangarajan, A., & Vemuri, B. C. (2010). CAVIAR: Classification via aggregated regression and its application in classifying the OASIS brain database. IEEE international symposium on biomedical imaging (pp. 1337–1340)
Cheng, G., Vemuri, B.C., Carney, P.R., & Marcei, T.H. (2009). Non-rigid registration of high angular diffusion images represented by gaussian mixture fields. International conference on medical image computing and computer assisted intervention, MICCAI (1) (pp. 190–197).
Christensen, G. E., Rabbitt, R. D., & Miller, M. I. (1996). Deformable templates using large deformation kinematics. IEEE Transactions on Image Processing, 5(10), 1435–1447.
Christensen, G. E., & Johnson, H. J. (2001). Consistent image registration. IEEE Transactions on Medical Imaging, 20(7), 568–582.
Christensen, G. E., & Johnson, H. J. (2003). Invertibility and transitivity analysis for non-rigid image registration. Journal of Electronic Imaging, 12(1), 106–117.
Coifman, R. R., Kafon, S., Lee, A. B., Maggioni, M., Nadler, B., Warner, F., et al. (2005). Geometric diffusions as a tool for harmonic analysis and structure definition of data: Diffusion maps. Proceedings of the National Academy Science, 102(21), 7426–7432.
Coifman, R. R., & Lafon, S. (2006). Diffusion maps. Applied and Computational Harmonic Analysis, 21, 5–30.
DoCarmo, M. P. (1992). Riemannian geometry. Boston: Birkhauser.
Gur, Y., & Sochen, N. (2009). Visualization and processing of tensor fields: advances and perspectives (pp. 325–340)., Coordinate-based diffusion over the space of symmetric positive-definite matrices Berlin: Springer.
Jian, B., Vemuri, B. C., Ozarslan, E., Carney, P. R., & Mareci, T. H. (2007). A novel tensor distribution model for the diffusion-weighted MR signal. NeuroImage, 37, 164–176.
Jian, B., & Vemuri, B. C. (2007). A unified computational framework for deconvolution to reconstruct multiple fibers form diffusion weighted MRI. IEEE Transactions on Medical Imaging, 26(11), 1464–1471.
Johnson, H. J., & Christensen, G. E. (2002). Consistent landmark and intensity-based image registration. IEEE Transactions on Medical Imaging, 21(5), 450–461.
Joshi, S., Davis, B., Jomier, M., & Gerig, G. (2004). Unbiased diffeomorphic atlas construction for computational anatomy. NeuroImage, 23, 151–160.
Kimmel, R., & Sochen N.A. (1999). Geometric-variational approach for color image enhancement and segmentation, Scale-Space (pp. 294–305).
Kimmel, R., Malladi, R., & Sochen, N. A. (2000). Images as embedded maps and minimal surfaces: movies, color, texture, and volumetric medical images. International Journal of Computer Vision, 39(2), 111–129.
Koenderink, J., & van Doorn, A. J. (2002). Image processing done right. Proceedings of ECCV, 1, (pp. 158–172).
Kroon, D.J., & Slump, K. (2009). MRI modality transformation in Demon registration. IEEE international symposium on biomedical imaging (pp. 963–966).
Kurtek, S., Klassen, E., Ding, Z., Srivastava, A. (2010). A novel Riemannian framework for shape analysis of 3D objects, Proceedings of CVPR, (pp. 1625–1632).
Kurtek, S., Klassen, E., Ding, Z., Avison, M. J., & Srivastava, A. (2011). Parameterization-invariant shape statistics and probabilistic classification of anatomical surfaces. IEEE Transactions on Medical Imaging, 30(3), 49–858.
Litke, N., Droske, M., Rumpf, M., & P. Schröder. (2005). An image processing approach to surface matching, symposium on geometry processing, (pp. 207–216).
Marcus, D. S., Wang, T. H., Parker, J., Csernansky, J. G., Moris, J. C., & Buckner, R. L. (2007). Open access series of imaging studies (OASIS): Cross-sectional MRI data in young, middle aged, non-demented, and demented older adults. Journal of Cognitive Neuroscience, 19, 1498–1507.
Moakher, M. (2006). A differential geometric approach to the geometric mean of symmetric positive-definite matrices. SIAM Journal on Matrix Analysis and Applications, 26(3), 735–747.
Seo, D., & Vemuri, B. (2009). Complex diffusion on scalar and vector valued image graphs, Proceedings of EMMCVPR (pp. 98–111).
Sochen, N. A., Kimmel, R., & Malladi, R. (1998). A general framework for low level vision. IEEE Transactions on Image Processing, 7(3), 310–318.
Tagare, H. D., Groisser, D., & Skrinjar, O. (2009). Symmetric non-rigid registration: a geometric theory and some numerical techniques. Journal of Mathematical Imaging and Vision, 34, 61–88.
Tuch, D.S. (2002). Diffusion MRI of complex tissue structure, PhD thesis, MIT.
Vercauteren, T., Pennec, X., Perchant, A., & Ayache, N. (2007). Non-parametric diffeomorphic image registration with the Demons algorithm, International conference on medical image computing and computer assisted intervention, MICCAI 2, (pp. 319–326).
Xie, Y., Vemuri, B.C., & Ho, J. (2010). Statistical analysis of tensor fields. International conferences on medical image computing and computer assisted intervention, MICCAI (1) (pp. 682–689).
Yanovsky, I., Thompson, P.M., Osher, S., & Leow, A.D. (2007). Topology preserving Log-unbiased nonlinear image registration: Theory and implementation. Proceedings of CVPR (pp. 1–8).
Zhang, P., Wee, C.-Y., Niethammer, M., Shen, D., & Yap, P.-T. (2013). Large Deformation Image Classification Using Generalized Locality-Constrained Linear Coding. International conference on medical image computing and computer assisted intervention, MICCAI.
Acknowledgments
This research was supported in part by NSF Grant IIS 0916001 to JH and BCV and the NIH grant NS066340 to BCV.
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by M. Hebert.
Appendices
Appendix 1
In this section, we show how to factor out DSC basis functions in formulating the gradient descent direction w.r.t. the DSC coefficients. Here, we only show details of derivatives w.r.t. \({\mathcal {A}}_{nml}\). Derivatives w.r.t. \(\mathcal {B}_{nml}\) and \(\mathcal {C}_{nml}\) can be done in the same way described in the following. In this section we consider Eq. 14 and in the following sections Eqs. 15 and Eq. 16 will be considered. Here, we only consider the image graphs of Gaussian mixture fields.
Equation 14 is rewritten as
In the Part1 of Eq. 44, factoring out basis functions is very straight forward, therefore, we only show how to do this in the Part2 below.
Factoring out basis functions in Eq. 45 requires evaluation Eqs. 40, 41, and 42 explicitly. First, we need to evaluated derivatives of Jacobian matrix, \(\mathbf {J}\). Given
the first order derivative w.r.t. \({\mathcal {A}}_{nml}\) is given
Once we notice that
where the subscript i of \({\varDelta }\) is same with i of \(\mathbf {u}_{i{\mathcal {R}}}\), and rewrite \(\mathbf {J}^{-1}\) as
then, we can factor out basis functions in Eq. 43 as
and in the same way,
With Eq. 50, we can rewrite Eqs. 40 and 41 as
and
where
Finally we factor out basis functions and their derivatives in Eq. 45 as follows.
Appendix 2
In this section, we show how to evaluate Eqs. 15 and 16. Denote \(J_{\gamma }\) the determinant of the Jacobian, \(\mathbf {J}_{\gamma }\) defined in Eq. 7. Its derivative w.r.t. \({\mathcal {A}}_{nml}\) defined in Eq. 12 can be written as
Given the derivative of \(\mathbf {J}_{\gamma }\), evaluations of Eqs. 15 and 16 is straightforward:
and
Rights and permissions
About this article
Cite this article
Seo, D., Ho, J. & Vemuri, B.C. Covariant Image Representation with Applications to Classification Problems in Medical Imaging. Int J Comput Vis 116, 190–209 (2016). https://doi.org/10.1007/s11263-015-0841-x
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11263-015-0841-x