and capable of entering into photo-oxidative reactions with neighbouring molecules [15]. Conditions leading to separation of the retina from the RPE prevent proper phagocytosis of shed outer segments. A number of diverse diseases can cause an increase in autofluorescence to arise from the outer retina.

Because RPE lipofuscin is largely composed of retinoids, which because of their conjugated double bonds can absorb in the visible light spectrum. This same molecular structure allows retinoid molecules to fluoresce, which provides opportunity for imaging their presence. The commercial SLO-based systems use 488 nm light to excite lipofuscin and use a longpass filter starting at 500 nm as a barrier.

The advantages of the commercial SLO systems include the ability to have confocal imaging, where only light from the conjugate plane is used in imaging. The commercial SLO has software that automatically adjusts the brightness and contrast of each image, which simplifies making a presentable image. Disadvantages include the images have more noise, are more variable from one image to the next, and the images have a potentially variable but unknown alteration by the software in the production of the final image.

Fundus camera–based systems use filters that are placed in the light path in a manner similar to that used for fluorescein angiography, but with an excitation of 535–585 nm which is a green color and a barrier at 605– 715 nm. The wavelengths used for the fundus camera system excite the lower end of the curve for lipofuscin. Advantages of the fundus camera–based system are lower incremental cost, low noise, repeatability, and the potential to record and measure raw information from the patient. Disadvantages include the lack of any help from the software in optimizing the image. The raw image from the fundus camera often has lower contrast than the modified image automatically produced by the SLO. Although this is easy to fix, it is a repetitive task that should be automatically done in software, but unfortunately, it must be done manually by the user.

9.5Optical Coherence Tomography

9.5.1The Wave-Like Nature of Light

Light rays, due to their wave-like properties, can interfere with each other. This tendency is used in interferometers in which the light rays from a test (or sample) arm are compared with light rays from a reference arm.

Alterations in the temporal arrival of light from the sample arm become readily detectable when this comparison is made because of an ingenious use of the periodic nature of the waveform. Each photon, by virtue of its periodic character in essence, carries its own reference clock. The phase difference between the clocks in the sample and reference arms is compared. This allows detection of very small differences in time delay, far shorter than what could be determined by the use of electronic detection and gating circuits. We let the light time itself.

9.5.2Coherence Length

Light is said to be coherent when the correlation of two different waveforms exceeds a set limit. Spatial coherence is when there is sufficient correlation in two rays that vary in space, while temporal coherence is an estimate of the correlation across time. Some waveforms can show coherence across space (spatial coherence), that is one part of the section of a beam may be coherent with another part of the cross-section, but not be coherent with light in the same beam from a different period of time. The time that light rays remain coherent is called the coherence time, and the distance covered during this time is called the coherence length. Ordinary lasers, such as a helium-neon laser, have very long coherence lengths. Light from one part of the beam may show high correlation with the beam meters ahead or behind.

In an interferometer, a light with a long coherence length makes measuring small differences in path length easy. Small changes in path length will cause a demonstrable change in the interference between the two beams. The problem with long coherence lengths is that large changes in path lengths can result in the same change as do smaller changes. If the difference in path length is changed by 1000 wavelengths, the interference produced would be no different.

9.5.3Time Domain Optical Coherence Tomography

Addition of periodic waveforms results in a summation, as per Fourier, and the resultant waveform contains all of the components. Some light sources are capable of producing a range of wavelengths that sum to produce the output. This light has a bandwidth of light of various wavelengths; these wavelengths add to produce a more

complex waveform that varies with time. Due to the variability of waveforms produced, the time in which the light remains coherent is small. The coherence length of this light can be quite short. Light with a short coherence length can interfere with similar waveforms, but not others. Therefore, interference between the reference and sample arms of the interferometer will only occur if the arms are nearly the exact length. With this knowledge, we can locate specific points in the object space because we know the length of the reference arm. By moving this test point through the object space, we can assemble an A-scan of the tissue being examined. If we add up successive A-scans, we get a B-scan. This is how time domain OCT (TD-OCT) works.

The problem with TD-OCT is that the signal is generated from a very small point in the tissue even though we are shining the light through the full thickness of the tissue. OCT scans have to be completed within a reasonable time because of factors such as potential light damage to tissue and patient steadiness. This limits the time of examination, which in turn limits the time in which any particular scan can be done. The less time devoted to any given A-scan means each point in the A-scan is sampled for less time. As in any measurement technique, a smaller sample is correlated with lower signal to noise ratio.

9.5.4Frequency Domain Optical Coherence Tomography

Interference is not an all-or-nothing response. If two light beams interfere, there is an interferogram produced and there is a frequency shift present that varies with path length mismatch. In other words, the depth information is encoded in the frequency space. If a grating is used to spread the resultant spectrum onto a line CCD, this encoding can be decoded. This is somewhat equivalent to having many detectors in a time domain instrument. A Fourier transform is used in the decoding process to obtain information from the thickness of the tissue sampled. Because a Fourier transform is used to analyze the spectral information, OCTs employing this approach are called Fourier domain or spectral domain OCTs (SD-OCT). The advantage of SD-OCT is that information from more than one place in the light path is sampled simultaneously. This speeds up the operation as compared with sampling only one point as per time domain OCT. For any given time the light beam is shining into the tissue, the signal obtained from a par-

ticular point is greater for a SD-OCT implementation as compared with a TD-OCT instrument. Because there are trade-offs that can be made between scanning speed and final signal to noise ratio, each OCT device operates at a different A-scan rate. For TD-OCT, this was generally 400 A-scans per second. For SD-OCTs, usual A-scan rates vary from about 27,000 to 41,000.

9.5.5Increasing Depth of Imaging

A disadvantage of SD-OCT is that the resolution and sensitivity decline with increasing path length mismatches. Even though there is a roll-off in resolution, the increasing darkness of structures deeper in the eye is more objectionable. A technique to increase the ability to image deeper structures possible with many SD-OCT devices is to push the OCT instrument closer to the eye to obtain an inverted image [16]. This inverted image has more information from deeper images as the result of a differential in the sensitivity. This technique was called enhanced depth imaging OCT or EDI-OCT. With this technique, it is possible to image the choroid, and in case of very high myopes, image the full thickness of the sclera.

Swept source OCTs (SS-OCT) use a different means of generating an image in which the center frequency of the light source is changed rapidly. A photodiode is used to record the interferogram. One big advantage of SS-OCT is the output of the photodiode occurs essentially in real time instead of having to wait for the wells of a CCD to fill and then be sequentially measured. This enables SS-OCT instruments to operate at higher scan speed than SD-OCT. A-scan rates of 100,000 to greater than 300,000 have been shown. A secondary benefit of SS-OCT is that the roll-off in sensitivity with increasing depth is not the same as that seen with SD-OCT. SS-OCTs will likely be the next generation of commercial OCTs, so imaging and evaluating the choroid will become more commonly available in clinical medicine.

9.5.6General Optical Coherence Tomographic Imaging Characteristics of the Macular Region

The image from OCT instruments is derived from the coherence information and not the brightness of the