The proposed computational anatomy project explores how well a set of formalisms derived from combinatorial network optimization theory fit as models for brain anatomy. The basic hypothesis is that long-range connections in the brain are a critically constrained resource, hence there is great pressure to finely optimize their deployment. The project would evaluate how accurately quantitative neuroanatomy data conforms to network optimization concepts. One concept may govern component placement in the nervous system: Given the specification of all interconnections among, for instance, the dozen ganglia of the roundworm C. elegans, are these components so positioned as to minimize total length of those connections? Our preliminary studies indicate that the actual layout of the ganglia in fact requires less connection length than any of the nearly 40,000,000 other possible placements. The planned experiments first involve compiling and/or refining appropriate neuroanatomical databases, principally from published anatomy, with initial focus including C. elegans and macaque and cat cerebral cortex. The proposed work thus meshes with current efforts toward a Neural Circuitry Database project--indeed, network optimization formalisms may prove a valuable top-down guide for the latter initiative. The experiments next involve evaluation of connection-optimality of the neuroanatomy. Network optimization problems are among the most computationally costly known; in general, only exhaustive search of all possible layouts can guarantee exact solutions. However, some probabilistic/approximation procedures; developed for microcircuit chip design are promising candidates for """"""""quick but dirty"""""""" biological mechanisms in optimization of neuroanatomy: The proposed experiments would include study of-such mechanisms via parallel computer simulations. A grasp of such very general principles governing generation of the human nervous system should be a useful step in understanding how the embryology can be disrupted, and in turn can be protected from birth defects. Such basic structural principles would also be valuable in the growing field of research on therapeutic regeneration of the nervous system, particularly use of tissue transplants.
| Cherniak, C (1994) Component placement optimization in the brain. J Neurosci 14:2418-27 |
