A versatile IDL toolkit for quantitative medical imaging research

Abstract

723

Learning Objectives: How to facilitate image-based research with low-cost, powerful software tools.

Abstract (summary): Many powerful medical image viewers are readily available from both commercial and open sources. These often feature refined visualization tools but typically lack in three areas important in research applications: 1) Efficient volume of interest (VOI) definition tools; 2) facilities for flexible, programmed manipulation of regions and image data; and 3) an open I/O architecture that supports interactions with databases and that can be easily extended for new image/data formats. Our toolkit results from an effort to create a software environment that overcomes these limitations and that expands the usability of already available packages. We will present a platform-independent library of software tools for medical, image-based research. It includes tools for communication with PACS and scanners (expanded query/retrieve), integration with a research-oriented relational database, and interactive image visualization and analysis (region definition, curves and ad-hoc processing). The software is developed in IDL, a high level interpreted language already rich in image display and analysis capabilities. This development environment has simplified maintenance, maximized adaptability and reduced the time for deployment of new procedures. Library features that will be demonstrated: 1) methods for querying from and inserting data into a relational database for imaging research; 2) easy-to-use, low-level routines for DICOM communications (examples: DICOM worklist access, reading/writing Radiotherapy Structure Sets or other complex DICOM p.10 files, special purpose de-identification, etc.); 3) adaptable and interactive 4D image visualization; tools for manual and semiautomatic VOI definition, storage, tracking and processing; drawing VOIs simultaneously in all 3 orthogonal views; generation of time activity curves from 4D datasets; re-slicing and reorienting image data using alignment matrices; and 4) support for arbitrarily elaborate image analyses using the full capabilities of IDL or modules written in other languages.

Research Support (if any): This research was supported in part by the Intramural Research Program of the NIH, CC.