Overview

Figure 1a: DTI Tractography

Mathematical models are the foundation of biomedical computing. To further our understanding of complex diseases, such as schizophrenia or Alzheimer’s disease, we need complex models that encompass many factors – models of anatomy, morphology, function, interrelation of elements, as well as changes of each as the disease progresses. Although, clearly, these models will evolve from analysis of anatomical, pathological, and clinical data, such models are limited in scope, unless they also incorporate critical information that can best be derived from medical images. This is particularly true since images now encompass techniques beyond the visible light photograph and microscopic images of biology’s early years. Imaging, today, is better viewed as a collection of geometrically arranged arrays of data samples that measure an infinite range of information. Physical attributes such as tissue type can be derived from traditional imagery, but diverse other physical and physiological properties, such as time-varying hemoglobin deoxygenation due to localized changes in neuronal metabolism, or vector-valued water diffusion through and within tissue, are now also quantifiable with modern imaging techniques. The broadening scope of imaging as a way to organize our observations of the biophysical world has led to a dramatic increase in our ability to apply processing techniques and to combine multiple channels of data to instantiate sophisticated and complex mathematical models of physiological function and dysfunction. We believe that a National Center for Biomedical Computing, dedicated to the advancement of medical image computing, will have a broad and significant impact on experimental, clinical biomedical, and behavioral research.

Figure 1b: Ventricular Shape Differences in Twin Study

It is not enough for image analysis efforts to demonstrate new scientific principles. These efforts must be converted into working systems that are easily used and accessed by scientific practitioners. The National Alliance for Medical Image Computing (NA-MIC), proposed here, will integrate the efforts of leading researchers with a shared vision for development and distribution of the tools required to advance the power of imaging as a methodology for quantifying and analyzing biomedical data. This shared vision is based on a thorough composition of computational methods, from image acquisition to analysis, that builds on the best available practices in algorithm development, software engineering, and application of medical image computing for understanding and mitigating the effects of disease and disability.

Figure 1c: Cortical Anisotropy Map

NA-MIC’s goal is to develop, integrate, and deploy computational image analysis systems that are applicable to multiple diseases, in different organs. To provide focus for these efforts, a set of key problems in schizophrenia research has been selected as the initial Driving Biological Projects (DBPs) for NA-MIC. Schizophrenia is a multi-faceted illness affecting 1% of the US population and consuming a significant portion of the healthcare budget – estimates of yearly costs are $60 billion. Yet the science of schizophrenia is only now beginning to take concrete form, primarily because neuroimaging techniques are finally providing a sufficiently detailed picture of the structure of the living brain and tracking the way the brain functions in controlled experimental settings. These sophisticated images – time-varying, multi-spectral, scalar, and vector-valued – are fruitful ground for computation, because the patient’s anatomy forms a three-dimensional coordinate system in which to accurately combine the multiple sources of information. Thus, in addition to making important contributions to the understanding of schizophrenia as an illness, we believe the richness of this problem domain will drive the creation of computational tools and techniques with broad and significant applicability to many important areas of image-based biomedical computing, particularly as we expand the scope of NA-MIC to incorporate new DBPs, both within the brain and in other organs.

Figure 1d: Hippocampal Shape Differences in Schizophrenia

Examples of the potential for computational image analysis are shown in Figure 1. Figure 1a illustrates the rich detail that can be extracted and visualized using the tools this project provides. Figure 1b demonstrates a morphology comparison of ventricles for selected comparison populations. Figure 1c demonstrates a visualization of cortical anisotropy. Figure 1d demonstrates an analysis of shape difference in hippocampus between normals and subjects with schizophrenia. These examples clearly illustrate the potential power of image analysis tools to provide insight into disease effects.