Perelman School of Medicine at the University of Pennsylvania

Section for Biomedical Image Analysis (SBIA)

PREDICT - Atrophy Simulation

We develop an algorithm to simulate tissue atrophy. Even though we focus on simulating the atrophy primarily on brain tissue, the algorithm is general in the sense that it can be applied to simulate volumetric changes, be it expansion or shrinkage, in any other entity as well. In our approach to simulate brain tissue atrophy, we adapt the strategy delineated by the well-known Occam's razor: solution to problems must be as simple as possible, but not simpler. In the case of brain tissue atrophy, in the absence of a priori knowledge of precisely how a specific medical condition causes tissue loss and in which pattern, the only information available is that the apparent volume of the tissue is reduced. We therefore proceed to find a deformation that corresponds to a reduction of tissue volume in a specified region of the brain. Specifically, given a prescribed level of tissue volume after atrophy over an image, we solve for a warping deformation that produces the desired volumetric changes using an energy minimization approach. The advantages of the method is that it is automated, easily generalizable to other instances of spatially varying volumetric change, and can readily be modified to incorporate a priori knowledge in the form of a statistical atlas of tissue loss. Furthermore, the technique does not involve making unnecessary assumptions on how tissues deform, and produces highly realistic results.

The results generated by this approach are illustrated in the following two figures. In Figure 1, a local atrophy has been simulated over the brain tissue at degrees 10, 20, 30, 40, and 50 %. Figure 2 illustrates the effects of simulating global atrophy and expansion by a rate of 25 %. It is clear that with the atrophy, the gyry shrink down noticeably making way to the cerebro-spinal fluid, while the expansion enlarges the brain tissue which in turn pushes the cerebro-spinal fluid out and away.

Fig.1. Simulation of local atrophy at varying degrees. At higher atrophy levels, the brain tissue shrinks significantly and retreats, while the regions outside the atrophy experience no deformation.

Fig.2. Simulation of global volumetric change. Global atrophy produces an overall shrinking effect on the brain tissue without affecting the general topology of the morphological structures. Global expansion also preserves the original topology while inflating the brain tissue within the cranial space, causing the gyry to become tightly pressed against each other.


To download please visit our PREDICT NITRC page



  1. B.Karacali and C.Davatzikos, "Estimating Topology Preserving and Smooth Displacement Fields", IEEE Transactions on Medical Imaging, Vol 23, No.7, p.868-880, July 2004.
  2. B.Karacali and C.Davatzikos, "Simulation of Tissue Atrophy Using a Topology Preserving Transformation Model", submitted to IEEE Trans. on Medical Imaging, 2004.
  3. Zhong Xue, Dinggang Shen, Bilge Karacali, Joshua Stern, David Rottenberg, and Christos Davatzikos, "Simulating Deformations of MR Brain Images for Validation of Atlas-based Segmentation and Registration Algorithms", submitted, 2005.