Imaging System Equips Ophthalmologists with Noninvasive Detection Method
By Lawrence J. Curran, Contributing Editor
An imaging system that promises to eliminate invasive procedures for detecting retinal damage and other eye pathologies is undergoing validation tests by ophthalmologists at several selected clinics before graduating to volume production. Called OASIS (ophthalmic applied spectralimaging system), the platform was developed by Applied Spectral Imaging (ASI; Migdal Ha'emek, Israel, and Carlsbad, CA).
The company has already installed more than 150 similar systems for use in the study of disease detection through multicolor chromosome classification—cytogenetics. However, the OASIS system is ASI's first initiative in ophthalmology. H. Laurence Shaw, medical doctor and chief executive officer at ASI, says retinal disease is the main cause of blindness and visual impairment in people of the developed world.
OASIS is based on ASI's SpectraCube technology, which Shaw says may provide the means to detect retinal disease early enough to save the sight of many persons at risk. "Among other things, SpectraCube displays a layered dissection of retinal tissue," he says. The technology also may eventually replace the invasive techniques used today, which usually involve dye injection to heighten image contrast, with a safer noninvasive alternative.
FIGURE 1. Applied Spectral Imaging spectral-bioimaging system—OASIS—uses the company's SpectraCube technology. The system includes a retinal (fundus) camera rather than a microscope as an optical instrument, an interferometer, a CCD camera, a Windows NT-based host computer, and ASI's custom image-acquisition and analysis software to perform noninvasive examinations of the eye's retina.
A key element in OASIS is an optical head that is connected to the fundus (retinal) camera—the optical instrument that provides the light source. It also includes an ASI-developed interferometer, which, in turn, is coupled to a modified and cooled Hamamatsu Corp. (Bridgewater, NJ) C4880-81 CCD camera with 640 x 480-pixel resolution (see Fig. 1). The spectral characteristics of this camera are tailored to image blue and green fluorochromes and the high-resolution shorter wavelengths typically used in many microscope techniques.
The acquisition and the quantitative analysis of the obtained spectral images are controlled by a Pentium-III-based 600-MHz PC provided by Gateway Inc. (San Diego, CA). This computer runs Windows NT 4.0 and ASI's custom image-acquisition and analysis software. Anthony Arciniega, an ASI technical-support specialist in Carlsbad, CA, explains that the PC has been customized by ASI as a high-end workstation. This workstation provides 256 Mbytes of memory, an Ethernet 10/100-Mbit/s network card to support the camera-acquisition board and the SpectraCube interface card, and custom boards that control the interferometer motor and scanner.
The system generates a spectral image from hundreds of black-and-white images taken by the CCD camera. All the frames acquired for each pixel are used to construct an interferogram, which is a representation of the light intensity as it changes with each changing path. Each pixel's interferogram is later transformed into that pixel's spectrum. The interferograms of all the pixels taken together allow the reconstruction of the entire image's spectrum, which can show abnormalities in retinal tissue.
Shaw says that the ability "to build a stack of spectral images, or interferograms, for each pixel is an innovative aspect of SpectraCube technology," as is ASI's spectral image-analysis software. In examining an eye's retina, the system takes the light captured by the microscope or fundus camera and directs this light through a beam splitter. The beamsplitter divides the light in the sample into two beams.
A set of mirrors guides the beams down two paths of different lengths, and then, the two beams are merged at the end of the paths.
At the merging point, the two beams are superimposed. The total intensity of the two superimposed beams is a function of the difference in the distance between the two light paths. This path difference is called the optical path difference (OPD). The intensity of the merged beam is measured by the CCD camera. As many as 100 data points might be acquired to create an interferogram by using different OPDs each time. At each OPD, this process occurs simultaneously for each pixel in the image.
Next, all the data points acquired for each pixel are used to build the interferogram. Ultimately the interferograms of all the pixels are combined to reconstruct the entire image's spectrum.
FIGURE 2. An oxygen saturation (OS) map is a digital image where each pixel is assigned a color according to its OS value. This value indicates the percentage of oxygenated hemoglobin within the total amount of hemoglobin. The color map attached to the OS map matches each color with its OS value. The OS level (70%) of green pixels represents an unhealthy retina image (left). A healthy retina shows strong red shades, which indicate the presence of adequate levels of oxygen (right).
Elaborating on the acquisition process, Shaw explains that the system's optical head rotates for a few seconds to gather the 100 or so data points. It looks at the same spot on the retina but from many different angles. The data points are captured by the CCD camera and sent to the PC host. There, the data are converted into wavefronts by means of fast Fourier transform processing, which typically takes 5 s, depending on the size and resolution of the spectral image. In this manner, a stack of wavefronts is built for each pixel.
Last, the wavefront data are handed off to the spectral-analysis software, which might vary based on what aspect of the eye is being diagnosed. For example, the diagnosis might be focused on the oxygen-saturation level of the hemoglobin in the retina, which would indicate eye damage caused by diabetes. The oxygen saturation level would show up as a color difference: a lack of sufficient oxygen in diabetic retinopathy would show up as blue, whereas a healthy retina would appear red in the same region of interest (see Fig. 2).
"OASIS can produce images equal to an angiogram, and we can provide a pseudo-angiogram without the injection of a dye (fluoroscein)," says Shaw. He hopes that eventually the technology will contribute to the detection of glaucoma, because it can look at the neural layer under the retina and reveal any damage much earlier than can conventional test methods. He also foresees a time when the system will perform an optical biopsy simply by looking at the region of concern without invasive probing.
Company Information
Applied Spectral Imaging
Carlsbad, CA 92009
Web: www.spectral-imaging.com
Gateway Inc.
San Diego, CA 92121
Web: www.gateway.com
Hamamatsu Corp.
Bridgewater, NJ 08807
Web: www.usa.hamamatsu.com