Non-rigid Atlas to Subject Registration with Pathologies for Conformal Brain Radiotherapy

Abstract

Abstract. Warping a digital atlas toward a patient image allows the simultaneous segmentation of several structures. This may be of great interest for cerebral images, since the brain contains a large number of small but important structures (optical nerves, grey nuclei, etc.). One important application is the conformal radiotherapy of cerebral tumor, where a precise delineation of all these structures is required. However, in this case, the variability induced by the tumor or a surgical resection, that are not present in the digital atlas, prevents an accurate registration between the atlas and the patient images. Since our registration method allows to locally control the amount of regularization, we are able to explicitly introduce those areas in the warping process. For computational efficiency, we have created a parallel implementation that can be used from the clinical environment through a grid interface.

Dense deformation field estimation for atlas-based segmentation of pathological MR brain images

“…As the deformation is only based on contours of interest, the probability of registration errors increases more we are far from these contours. As far as we know, most of the existing methods for registration of images with space-occupying tumors are either surface-based [10,12,15,35] or voxel-based [2,4,[6][7][8]36] approaches. However, in our opinion it is worth to study how to combine the advantages of both approaches.…”

“…Other methods [7][8][9] locally adapted the elasticity of the transformation, rather than modeling the deformation induced by the tumor, in a way that large deformations induced by the tumor can be captured. To avoid tumor segmentation, Duay et al in [7] did not use a special scheme to preserve the regularity of the transformation in the areas where the images differ topologically (e.g., inside the tumor).…”

“…They proposed to use an elasticity map related to the atlas to allow larger displacements over regions (the inside of the lateral ventricles) that are known to be very sensitive to the tumor deformation. On the contrary, in order to prevent bad deformations inside the lesion area, Stefanescu proposed in [8,9] to impose more rigidity on the tumor. This required the previous segmentation of the patient's tumor.…”

“…A widely used solution is to allow globally or locally more elasticity to the deformation in order to obtain more local deformation (see for instance [2,4,7,8]) with the risk of increasing the irregularity of the deformation field and thus of the contours. Moreover, this does not ensure that the sought level of precision will be obtained.…”

Incorporating Statistical Measures of Anatomical Variability in Atlas-to-Subject Registration for Conformal Brain Radiotherapy

“…The second term performs a regularization of the temporal derivative of the displacement field, similar to a fluid regularization of the velocity field in fluid mechanics. By locally weighting this "fluid" regularization with a scalar field k, [8] showed that the algorithm can be rendered robust with respect to potential pathologies in the subject image.…”