System Performance

FF-OCT is beneficial to yield 3-D imaging with a faster acquisition speed and less operations than conventional OCT techniques. However, some of the imaging characteristics of FF-OCT are inherited from the conventional OCT. Micrometre-scale spatial resolution and shot-noise- limited detection sensitivity are also achievable using the FF-OCT systems. This section focuses on the evaluation of the spatial resolution and the detection sensitivity of the simple FF-OCT system, based on the OCT imaging performance discussed in Section 2.4.4 (see above).

3.3.5.1 Depth-Resolution

The depth-resolution of the simple FF-OCT is determined by the spectral property of the illumination source, i.e. the infrared LED source with a central wavelength λ0= 880 nm and a spectral width∆λ = 110 nm. According to Equation 2.21 (see above), the theoretical depth-resolution of the simple FF-OCT system is 3.1 µm.

However, the resolving capability of an imaging system is fundamentally presented by its impulse response, i.e. the PSF. For all OCT variants, the depth-resolution can be alternatively verified by the FWHM of the main OCT envelope, which is the OCT PSF profile along the axial direction. For the simple FF-OCT system, the depth-resolution was thereby measured as 3.6 µm, resulting from an obtained main envelope signal, as described in Fig. 3.10. The measured resolution is slightly larger than the theoretical value, which can be explained by the presence of systematic noise.

μ

μ

Fig. 3.10 The envelope of a main interferogram with the identified depth-resolution of 3.6 µm.

3.3.5.2 Transverse Resolution

The transverse resolution is determined by the focused spot size in a point-scan OCT system. A small spot size results in a high transverse resolution with a high NA-focused beam, if the transverse scanning step size is smaller than the spot size. For a full-field illuminated OCT system, beams are not focused as a spot of light. Despite that, point sources of light from a sample appear as spots at the FF-OCT camera sensor. Considering Equation 2.22, the transverse resolution of FF-OCT is dependent upon the beam NA. The higher the NA of the total system, the better the transverse resolution.

However, for a camera-acquireden-faceFF-OCT image, the transverse sampling interval or the dimension of sample for a camera pixel must also be considered. In situations when the calculated value with Equation 2.22 (see above) is smaller than the sampling interval, the transverse resolution of the resulting digital image is limited only by the sampling interval.

The transverse PSF profile can be used to describe the transverse resolution. The image formed through the optical system is the convolution of the object’s irradiance with the transverse PSF of system [221]. In other words, the transverse profile of the FF-OCT image

wis the convolution (⋆) of the transverse profile of the objectvwith the transverse profile of the system PSFu, given by:

w=v⋆u. (3.18)

The PSF profile can then be derived through a deconvolution process [222], once bothwand

vare acquired.

A micrometre division line of a distance calibrator was taken as a sub-resolution radiating object. The division line was first imaged at the focal plane of the FF-OCT system to acquire the profilewfrom an acquireden-faceimage (Fig. 3.11 (a) and (b)), and then imaged with a

high resolution optical microscope to acquire a precise object profilev(Fig. 3.11 (c) and (d)). The transverse PSF profileuwas computed by the deconvolution ofwand v, and it is displayed in Fig. 3.11 (e). From these figures, the width of the object profile is found to be 7.6 µm, the width of the OCT image profile is found to be 11.5 µm, the width of the transverse PSF profile, i.e. the transverse resolution is measured as 10.3 µm.

Fig. 3.11 (f) (see below) further describes a PSF calculated from ideal object and image profiles with the same widths, i.e. 7.6 µm and 11.5 µm, and a slightly smaller PSF width of 10 µm is obtained, which may indicate a more accurate determination of the OCT transverse resolution. On the other hand, there are sufficient sampling points (camera pixels) within the PSF profile, suggesting that the transverse resolution is not limited by the sampling interval.

Fig. 3.11 (a) An OCT profile of the test scale line with a width of 11.5 µm; (b) an OCT image of the test scale line; (c) an object profile of the test scale line with a width of 7.6 µm; (d) a microscopy image of the test scale line; (e) a transverse PSF profile of the OCT system with a width of 10.3 µm; and (f) a PSF profile computed from the simulated dashed OCT profile and dotted object profile — the resultant width of the PSF is 10 µm.

3.3.5.3 System Sensitivity

Sensitivity is an important feature of an OCT system to denote the minimum detectable reflected optical power. It is generally calculated using the mathematical SNR formula in Equation 2.26 (see above), which can be expanded into Equation 2.29 for a time-domain OCT system. The latter SNR formula indicates that the system sensitivity is proportional to the power returning from the sample. To calculate a minimum resolvable reflectivity or the imaging sensitivity, an ND filter (with an optical density of D) followed by a mirror is normally used to attenuate the sample reflection, so that a sample with a reflectivity of

−20D in dB can be created. As mentioned before, the experimental verification of the system sensitivity is performed by measuring the OCT A-scan peak height and the noise floor, following Equation 2.37 (see above).

For the simple FF-OCT system, a window sample was used to generate a reflectivity at−28 dB (asRwin=0.04). With a mirror reference, an A-scan signal was obtained in the FF-OCT measurement of the window. In Fig. 3.12 (a) (see below), the acquired envelope signal was adjusted to present the variation of the reflectivity within the range of 0 to 0.04 as a function of the penetration depth. The smallest resolvable reflectivity is identified below

−60 dB in the logarithmic profile of the envelope for the window measurement, as shown in Fig. 3.12 (b). This indicates the imaging sensitivity for measurement of this window sample. The DR for this measurement can be also obtained within the range of−60 dB to−28 dB.

According to Equation 2.37 (see above), the expression of the system achievable sensitivity in dB can be rewritten as the reflectivity attenuation plus ten times the base- 10 logarithm of the ratio of the square of the A-scan peak intensity to the standard deviation of the reference arm noise floor, given by:

Σ=SNRdB=10 log10

I_FFOCT2

σ_noise2

+28. (3.19)

in whichIFFOCT represents the FF-OCT A-scan peak intensity,σnoise stands for the STD of the reference arm noise floor, and 28 dB is the sample arm attenuation. As the reference mirror contribute _1.1₀₄ of the total reflection power of both interferometric arms, the reference arm noise power accounts for a same proportion of the A-scan noise power. The STD of the reference noise was calculated as 1.98×10−4by using 140 intensity points in the A-scan noise floor within the depth range highlighted in Fig. 3.12 (a). By substituting the envelope peak reflectivity 0.04 and the reference noise variance 1.98×10−4into Equation 3.19 (see above), the sensitivity of the simple FF-OCT system can achieve as large as 74 dB.

Fig. 3.13 (a) and (b) demonstrates the reflectivity variation of a mirror sample in the FF-OCT measurement of the mirror / mirror envelope signal. In Fig. 3.13 (b), the weakest resolvable reflectivity is just below−40 dB, indicating that the imaging sensitivity for the measurement of the mirror sample is below −40 dB. It is also noted that the system DR for this measurement covers a range from−40 dB to 0 in Fig. 3.13 (b). The side lobes are echoes that are due to the non-Gaussian spectral shape of the light source. They are more distinct in the measurement of the sample with a higher reflectivity than the window sample. The reference noise variance was calculated as 9.37×10−4, according to A-scan noise intensity points as marked in Fig. 3.13 (a). The achievable sensitivity of the system was estimated to be 60 dB by Equation 3.19, while the sample attenuated reflectivity was zero.

Hence, the attenuation of the sample arm reflectivity can facilitate the estimation of the achievable system sensitivity of the simple FF-OCT system. The DR of the system can also be measured in terms of samples with different reflectivity.

Therefore, by using a cheap LED for the illumination of the sample at a power of 40 µW in the simple FF-OCT system, the achievable system sensitivity was obtained as 74 dB by the use of a window sample. For the imaging of the window sample, the smallest resolvable reflectivity was measured below−60 dB and a DR was acquired from−60 dB to−28 dB. For the imaging of the mirror sample, the smallest resolvable reflectivity was measured below

−40 dB and a DR was acquired from−40 dB to 0.

(a) (b)

μ μ

Fig. 3.12 Determination of the system sensitivity with a window sample; (a) an A-scan signal in a range of 0 to 0.04 to describe the window reflectivity as a function of depth; (b) the logarithmic scale of the A-scan signal in (a) with minimum identifiable reflectivity below

μ μ

(a) (b)

Fig. 3.13 Determination of the system sensitivity with a mirror sample; (a) an A-scan signal to describe the mirror reflectivity as a function of depth; (b) the logarithmic scale of the A-scan signal in (a) with minimum identifiable reflectivity below−40 dB.