Automatic axis generation for virtual bronchoscopic assessment of major airway obstructions.

Virtual bronchoscopy (VB) has emerged as a paradigm for more effective 3D CT image evaluation. Systematic evaluation of a 3D CT chest image using VB techniques, however, requires precomputed guidance data. This guidance data takes the form of central axes, or centerlines, through the major airways. We propose an axes-generation algorithm for VB assessment of 3D CT chest images. For a typical high-resolution 3D CT chest image, the algorithm produces a series of airway-tree axes, corresponding airway cross-sectional area measurements, and a segmented airway tree in a few minutes on a standard PC. Results for phantom and human airway-obstruction cases demonstrate the efficacy of the algorithm. Also, the algorithm is demonstrated in the context of VB-based 3D CT assessment.

Virtual bronchoscopy for three--dimensional pulmonary image assessment: state of the art and future needs.

Virtual bronchoscopy is emerging as a useful approach for assessment of three-dimensional (3D) computed tomographic (CT) pulmonary images. A protocol for virtual bronchoscopic assessment of a 3D CT pulmonary image would have two main stages: (a) preprocessing of image data, which involves extracting objects of interest, defining paths through major airways, and preparing the extracted objects for 3D rendering; and (b) interactive image assessment, which involves use of graphics-based software tools such as surface-rendered views, projection images, virtual endoscopic views, tube views, oblique section images, measurement data, global two-dimensional section images, and cross-sectional views. Although a virtual bronchoscope offers a unique opportunity for exploration and quantitation, it cannot replace a real bronchoscope. Limitations of current virtual endoscopy systems include high cost, lack of visual aids beyond simulated endoscopic views, difficulty in performing interactive anatomic exploration, lack of quantitative information, use of surface rendering instead of volume rendering, and need for substantial off-line display computation. Future needs include development of fully integrated user-friendly virtual bronchoscopes, development of optimal CT protocols for generating artifact-free data sets, and improvements in automated preprocessing of 3D CT images.

Distributed system for processing 3D medical images.

Three-dimensional (3D) image data generated by radiological imaging modalities such as CT and MRI can provide detailed structural insight. Automating the analysis of these images can improve the consistency of the results and reduce user interaction time, but introduces a tremendous computational burden. To address this problem, we have designed a distributed processing environment for the rapid processing of 3D medical images. Our system allows a user to perform automatic 3D filtering, segmentation, and measurement on a 3D image using a heterogeneous network of processors and the PVM protocol.

Hadamard transform imager and imaging spectrometer.

An imager and a spectrometric imager, which achieve multiplexing by the use of binary optical encoding masks, have been built and tested. The masks are based on orthogonal, pseudorandom digital codes derived from Hadamard matrices. The spatial (and/or spectral) data are therefore obtained in the form of a Hadamard transform of the spatial (and/or spectral) scene; computer algorithms are used to decode the data and reconstruct images of the original scene. The hardware, algorithms processing and display facility are described. A number of spatial and spatial/spectral images, obtained in the laboratory, are presented. We present an analysis of the situations for which the multiplex advantage may be gained and of the limitations of the technique. Potential applications of the spectrometric imager are discussed. The spectrometric imager is covered by U.S. Patent 3,720,469 assigned to Spectral Imaging Inc., Concord, Mass.