AIM Brain atlases offer a prominent solution for the anatomical localization of the physiological processes and pathological disorders of the brain investigated by cranial SPECT and PET studies. Although the number of the available brain atlases shows a significant growth, the support of this technique by information systems is still poorly accomplished.
The purpose of our work is to develop an interactive software assisted by multiple brain atlas databases for the promotion of the multi-modal medical imaging projects proceeded in our institute.

MATERIALS & METHODS To deal with the semantic and syntactic differences of the available databases a general atlas definition was constructed, which enables the uniform handling of various atlases. As part of this system we have developed components and database models for maintaining the discrete structural maps, the maximum probability maps and the region systems contained by the atlases. By implementing the deployed uniforming model, a framework was constructed, providing tools for the integration, maintenance and utilization of atlases in various tasks. A graphical user interface application, called BrainLOC, has also been developed to perform localization and region analysis tasks. The software is built upon the MultiModal Medical Imaging software development system.

RESULTS According to our purpose, a general and extendable model and framework for atlas uniforming has been developed. Based on these, the BrainLOC application (www.minipetct.com) was deployed. The software became one of the most important tools of several institutional projects, and among others it permits of the atlas-assisted anatomical localization and region analysis of SPECT and PET data aligned into atlas-space by the registration tool of the institutional automated image processing framework. The entire system was tested and validated with the PET and SPECT data of 20 subjects. BrainLOC is published under student and academic licenses for educational and research purposes.

CONCLUSION Using BrainLOC, the database of over 20 deterministic and probabilistic brain atlases can be used simultaneously for localization and region analysis tasks emerging by cranial PET and SPECT studies. Applying the appropriate atlas or an arbitrary collection of regions from different atlases provides the opportunity for the quantitative anatomical analysis of functional images.