Two major improvements, described below, have been made in the construction and operation of a computed-tomography imaging spectrometer (CTIS). These plus future improvements can be expected to enhance the practicality and commercial viability of CTISs, which, in principle, offer unprecedented capabilities for imaging with spatial, spectral, and temporal resolution. For example, the CTIS in its present form could be used in medical and pharmaceutical applications to perform spectral imaging of transient scenes that contain fluorescent dyes. With increases in spectral accuracy and spatial resolution, it could be used for remote sensing.

This is an Experimental CTIS. The heart of this instrument is the two-dimensional diffraction grating, which spectrally disperses an image of the scene in two spatial dimensions. Superior two-dimensional gratings with tailorable properties can be in the form of PMMA computer-generated holograms fabricated by electron-beam lithography and etching.

A CTIS includes a spectral disperser in the form of a two-dimensional diffraction grating positioned between two relay lenses in a video imaging system (see figure). If the disperser were removed, the system would produce ordinary images of the scene in the field of view of the system. In the presence of the grating, the image on the focal plane of the system contains both spectral and spatial information because the multiple diffraction orders of the grating give rise to multiple, spectrally dispersed images of the scene. By use of algorithms adapted from computed tomography, the image on the focal plane can be processed into an "image cube" — a three-dimensional collection of data on the image intensity as a function of the two spatial dimensions (x and y) in the scene and of wavelength (l). Thus, both spectrally and spatially resolved information on the scene at a given instant of time can be obtained, without scanning, from a single snapshot; this is what makes the CTIS such a potentially powerful tool for spatially, spectrally, and temporally resolved imaging.

Prior to the improvements reported here, the two-dimensional gratings for CTISs were constructed by stacking and crossing one-dimensional gratings. The disadvantages of this approach are that (1) total throughput efficiency is low, (2) diffraction-order efficiencies cannot be tailored to prevent saturation of focal-plane-array (FPA) photodetectors by weakly dispersed orders, and (3) the pattern of dispered images does not fill the FPA area efficiently. This leads to the first of the two improvements, which is the use of computer-generated holograms (CGHs) as the two-dimensional dispersers. The CGHs offer high total efficiencies and can be designed to generate arbitrary patterns of diffraction-order efficiencies. The CGHs are made from poly(methly methacrylate) by analog direct-write electron-beam lithography followed by development in pure acetone.

To be able to use the computed-tomography algorithms to reconstruct a scene from an image on the focal plane, one must first determine connection weights from positions and wavelengths in the scene to detector pixels. One can determine the connection weights fairly directly by measuring pixel detector outputs while scanning a monochromator-illuminated optical fiber across the scene. Such a complete calibration procedure is hardware-intensive and is time-consuming because the entire scene must be scanned anew for each resolution element in the image cube. This leads to the second improvement, which is a modification of the calibration procedure.

In the modified procedure, one does not scan the entire scene; instead, one uses measurements taken while the single point in the center of the scene is illuminated at each wavelength of interest in the pass band of the CTIS. There are two steps in the modified calibration procedure. In the first step, the pixel outputs are measured at each wavelength. From the measurements, the corresponding system efficiencies (throughput fractions) are calculated for all diffraction orders at all the wavelengths. In the second step, the system efficiencies are used in a ray-tracing computer program that calculates the connection weights from all scene positions to all pixels on the focal plane. The calculation accounts for transmissivities of lenses and other optical elements, plus spectral responsivities of the photodetectors.

This work was done by Daniel Wilson, Paul D. Maker, and Richard Muller of Caltech and Michael Descour and Eustace Dereniak of the University of Arizona for NASA's Jet Propulsion Laboratory. For further information, access the Technical Support Package (TSP) free on-line at  under the Physical Sciences category.