Engineering
Goals
The NAC Engineering Core serves as a critical link between the specialized domains of NAC scientific work and the larger community of computational tool development. This link work in two directions: technology inflow refers to the process of drawing from the work of other researchers and adapting it for use within NAC; technology outflow supports the dissemination efforts of the center by making the unique contributions of NAC researchers available to address biomedical imaging research at other sites.
Open Source Research Software Platform
After careful considerations of various software licensing arrangements and their impact on delivery of biomedical research software, the NAC has adopted a BSD-style license for software developed under NAC funding. The Slicer License has the following key elements:
- software can be distributed in binary form
- source code for modifications does not need to be released
- commercialization of the software is at the sole risk of the party or parties engaged in such commercialization
We have selected this approach based on the following considerations:
- NIH funding should support research that can be applied directly to improve human health
- Only FDA-approved treatments can become part of routine clinical care
- Commercialization is the proven method by which new technologies are able to have a major impact on health
- NAC is not a commercial enterprise
3D Slicer Version 3 is the NAC research software platform.
Close Collaborations
The complexity of large-scale software systems, together with the rigorous demands for accuracy and reliability inherent in biomedical computation requires a coordinated effort by many developers. Although NAC plays a critical role in supporting overall tool development, significant leverage is obtained by teaming with like-minded colleagues.
Other Centers
The NAC Engineering Core works closely with the following groups:
- The National Alliance for Medical Image Computing
- The National Center for Image Guided Therapy
- The Biomedical Informatics Research Network
NAC Cores
The primary motivation for new developments is the needs of our fellow NAC Core activities.
Diffusion Infrastructure
We are working to adapt the general purpose diffusion analysis framework being developed as part of the NA-MIC effort to support the advanced white matter analysis efforts into the underlying connectivity correlates of schizophrenia.
fMRI and Population Analysis
Group wise statistical analysis of population data from fMRI is an important challenge that has not previously been addressed by a comprehensive set of tools. We are working to incorporate new techniques pioneered within NAC into the 3D Slicer framework.
Ontology Atlases
The advent of remarkably detailed and reliable automated segmentation of brain images has led to a significant need for the application of neuroanatomical knowledge to the basic architecture and implementation of neuroimage analysis systems. Working closely with the CCA Core, the Engineering Core is creating a framework in which meaningful information tags can be added to image-based computational models. These tags are being associated with detailed ontologies that allow users of these systems to access related information resources which place the data in an appropriate overall anatomical and functional context.
Time-Series Engineering
With the expert consultation of Dominik Meier, the Time-Series Analysis (TSA) engineering effort is dedicated to developing tools to assist in the creation of dynamic temporal models of the morphological correlates of disease. Based on pioneering work developed in earlier generations of the software, customized 3D Slicer-based tools are being created to support new methods to build and analyze four-dimensional data sets obtained from longitudinal/serial MRI. Such fused datasets permit multiscale analysis of temporal change. Novel TSA paradigms aim to improve the robustness of the identification and quantification of MRI-visible change. Sensitivity boosts arise from avoiding the data-reduction associated with first segmenting a structure of interest before analyzing its change. TSA also enables some new questions about the processes underlying morphological change. In longitudinal studies, measures of the rates or dynamics of change are often of similar or greater importance than measures of the actual magnitude. This becomes particularly relevant when processes are non-linear, and estimates of rates defined as magnitude divided by the time interval become increasingly inaccurate. The TSA concept is being applied by our collaborators to a variety of clinical questions: e.g. white matter changes in normal aging, lesion development in multiple sclerosis, effects of HIV neuroencephalitis on (temporal) white matter change, subtraction imaging as advanced tool for radiological examination of follow-up data.