2.1.

Camera System

A schematic drawing of a reflective-mode time-resolved optical imaging system is shown in Fig. 1 . Two sets of mode-locked Ti-sapphire lasers (Tsunami, Spectra-Physics Lasers, USA), which are tunable to conventionally used wavelengths to measure brain activity (e.g., 780, 805, or $830\mathrm{nm}$ ; $80\mathrm{MHz}$ , 10 to $16\mathrm{ps}$ , 800 to $1100\mathrm{mW}$ ), were used as light sources. Pulsing of the $730\text{-}\mathrm{nm}$ laser was synchronized with that of the $830\text{-}\mathrm{nm}$ laser by a lock-to-clock system (Lock-to-clock, Spectra-Physics Lasers). Stabilities of laser pulsing were monitored by a photodiode (PD). Both lasers are guided into separate optical fiber switchers (F-SM19, Piezosystem Jena GmbH, Germany) through objective lenses. One laser pulse within 1 sec was incident on one optical fiber, and it was switched to the other optical fibers one by one by an optical fiber switcher controlled by a personal computer (PC). Each optical fiber output was introduced into a set of optical elements that consisted of a ball lens, iris, and convex lens, which collimate output beams [Fig. 2b ]. Laser powers were reduced to 1/100 by neutral density (ND) filters placed immediately after the laser output. Then, the laser beams were led to the optical switching system. Each laser power at the sample surface was also adjusted to $100\mathrm{\mu }\mathrm{W}$ by the output iris. Pulse width at the sample was around $100\mathrm{ps}$ (initial laser pulse widths were about 10 to $15\mathrm{ps}$ ).

Fig. 1

Block diagram of noncontact B-TRIS system.

Fig. 2

Noncontact illumination system. (a) Schematic drawing of the optical system around the sample. (b) Laser illumination assembly. (c) Photographic image of entire optical assembly.

Images were captured by a charge-coupled device (CCD) camera (Imager3, LaVision GmbH, Germany) equipped with a time-resolved intensifier (Picostar, La Vision). Laser output triggered a high-voltage gate pulse for the time-resolved intensifier. Timing of image acquisition and laser illumination was controlled by the PC. Acquired images were transferred to the PC independently from laser pulse timing $(80\mathrm{MHz}\gg 10$ to $100\mathrm{Hz})$ and analyzed by a separate software (Image Pro Plus, Media Cybernetics, USA).

2.2.

Noncontact Illumination System

To achieve noncontact illumination, we have prepared a light illumination assembly [Figs. 2a, 2b, 2c] encircling an objective lens [ $f=150\mathrm{mm}$ , diameter $\left(\varphi \right)=150\mathrm{mm}$ , borosilicate crown glass (BK7), antireflection (AR) coated]. A schematic drawing of the optical system around the sample is shown in Fig. 2a. The sample in the drawing is a large phantom block and the signal from the phantom is imaged by the objective lens through a large aperture on the upright optical bench. The sample image is relayed by a combination of the objective lens, and two sets of camera lenses (Nikkor $50\mathrm{mm}$ F1.2S, Nikon, Japan). The upright optical bench board holds multiple laser illuminating assemblies shown in Fig. 2b. The laser beam emanating from the optical fiber is collimated by the ball lens and focused on the sample surface by a convex lens and iris to a spot with a diameter less than $1\mathrm{mm}$ . Time lags of the laser output between each optical fiber caused by the optical switcher were between 25 and $200\mathrm{ps}$ . However, fibers with the closest output times were selected so that the time difference was maintained at less than $50\mathrm{ps}$ for every experiment. A photograph of the optical assembly is shown in Fig. 2c. Orange fibers are the incident optical fibers from a fiber switcher. In the present study, the distance between the phantom surface and the detector (and the convex lens for illumination) depends on the focal length of the objective lens $(f=150\mathrm{mm})$ , and the distance between the phantom surface and CCD was about $700\mathrm{mm}$ .

3.1.

Measurement of Optical Properties of Phantom Sample

For the phantom, we have used white polyacetal as a light-scattering medium and black polyacetal particles for absorbers. Temporal profiles of light intensity reflected by the large bulk phantom (Fig. 3 ) were measured by noncontact B-TRIS. Plots of the light intensity (solid lines) and the fitting curve using Patterson’s half infinite model²⁹ are shown in Fig. 3. Their optical properties were obtained by the present system (noncontact B-TRIS) and the conventional fiber-based time-resolved system (TRS-10, Hamamatsu Photonics, Hamamatsu, Japan)³⁰ and are shown in Table 1 . Although there is difference in the wavelength used here, $780\mathrm{nm}$ for noncontact B-TRIS and 761 and $795\mathrm{nm}$ for TRS-10, the scattering coefficient $\left(\mathrm{\mu }{\mathrm{s}}^{\prime }\right)$ obtained by both instruments gave similar results. On the other hand, absorption coefficient $\left(\mathrm{\mu }\mathrm{a}\right)$ showed about twofold difference between these instruments. This probably indicates that the order $\sim {10}^{-3}{\mathrm{mm}}^{-1}$ is the sensitivity limit of $\mathrm{\mu }\mathrm{a}$ for B-TRIS and/or TRS-10. However, optical properties of the human brain tissue (cortex) are $\mathrm{\mu }{\mathrm{s}}^{\prime }=\sim 1.00{\mathrm{mm}}^{-1}$ and $\mathrm{\mu }\mathrm{a}\cong 0.02{\mathrm{mm}}^{-1}$ .³¹ The value of $\mathrm{\mu }\mathrm{a}$ of the human cortex is about $20\mathrm{times}$ larger than that of the present phantom sample. Thus, the B-TRIS and TRS-10 may be practical to measure the optical properties of the human brain tissue.

Fig. 3

Temporal profiles of light intensity at $780\mathrm{nm}$ reflected by large bulk phantom (solid lines) and fitting curves (dashed lines) using Patterson’s half infinite model (Ref. 29) assuming refractive index $\left(n\right)$ of the phantom was 1.48. Measurements were performed every $25\mathrm{ps}$ . Light intensities, detected at distance $\left(\rho \right)$ of 12, 15, 21, and $26\mathrm{mm}$ from light illumination spot, were integrated over $3\times 3{\mathrm{mm}}^2$ of the detection area. Estimated optical properties of the phantom are shown in Table 1.

Table 1

Optical properties obtained with near-infrared time-resolved spectroscopy.

3.2.

Time-Resolved Imaging in Scattering Media

The captured images and calculation procedures for detecting small absorber particles buried inside the polyacetal phantom are illustrated in Figs. 4a, 4b, 4c, 4d . The illumination spots marked with filled red circles and the position of the absorber marked with an open black circle, where illuminating laser spots were placed hexagonally to cover the imaging area of about $90\mathrm{mm}$ in diameter, are illustrated in Fig. 4e [see also Fig. 2a].

Fig. 4

Detection of black absorber. (a) Images without an absorber; bright spots denote illuminating positions. (b) Images with an absorber. (c) Images subtracted (b) from (a); bright spot predicts the position of absorber. (d) Integrated image of seven images in (c). (e) Illumination spots (red circles) illustrated on the image (d); position of the absorber is shown by a black open circle. Wavelength of $805\mathrm{nm}$ , which corresponds to an isosbestic point of oxygenated and deoxygenated hemoglobin, was used to capture images.

Backscattered images from the phantom without an absorber are shown in Fig. 4a, and the same images with an absorber buried at a depth of $17.5\mathrm{mm}$ are shown in Fig. 4b. The absorber is $10\mathrm{mm}$ in diameter (φ) and $5\mathrm{mm}$ thick ( $t$ ). A pulse laser was illuminated sequentially to seven different spots [see light spot in images in Figs. 4a and 4b], and images were obtained with a 30-ms CCD integration time and a 600-ps time window (1800 to $2400\mathrm{ps}$ after the laser illumination, see Fig. 8 ) of the gated intensifier. Images in Fig. 4c were obtained by subtracting images of Fig. 4b from those of Fig. 4a, where specularly reflected light, which appeared as bright spots in Figs. 4a and 4b, were removed by subtraction and only diffusely reflected lights were obtained in Fig. 4c. An integrated image of the seven images in Fig. 4c is shown in Fig. 4d. As referred to in Fig. 4e, Fig. 4c indicates decreases in the reflected signal (“shadow,” i.e., light blue and yellow area) just above the absorber (except for numbers 2 and 3, whose illuminating spots are too far from the absorber). The center of the shadow is less than $2\mathrm{mm}$ from the center of the absorber. Those correspondences occurred not only for the integrated image [Fig. 4d] but also for each subtracted image [Fig. 4c].

Fig. 8

Changes in normalized light intensity obtained at different depths of the absorber. Configuration of incident optical fibers and an absorber was the same as that shown in Fig. 4e, but the depth of absorber was changed. Wavelength of $805\mathrm{nm}$ was used, and each plot was obtained with a time window of $600\mathrm{ps}$ . The data points indicate the light intensity integrated over the region of $3\times 3{\mathrm{mm}}^2$ on the phantom surface at a distance of $40\mathrm{mm}$ from the incident light spot. The absorber was buried in the middle of incident and detection points.

In the present study, the time window was selected to be $600\mathrm{ps}$ (1800 to $2400\mathrm{ps}$ after the illumination of the pulse laser), which gave a broadening of the diameter of the shadow [full width at half maximum $\left(\mathrm{FWHM}\right)=25\mathrm{mm}$ ] 2.5 times larger than the actual diameter of the absorber $\left(10\mathrm{mm}\right)$ . If earlier arriving photons are captured, the backscattered light from the superficial area may become dominant and its intensity may increase. As a result, the signal from deeply buried absorbers (cf. Fig. 8) is lost.

3.3.

Spatial Resolution of B-TRIS

For noncontact B-TRIS, the sensitivity to the target object depends on the geometry of the illuminator, but the detected position of the target object is probably unaffected by the position or geometry of the detector (Fig. 4), because the detection area of B-TRIS is the whole area (i.e., $\sim 90\text{-}\mathrm{mm}$ diameter of field of view using $\varphi =150\text{-}\mathrm{mm}$ objective lens). The spatial configuration of light illumination and detection of B-TRIS are illustrated in Fig. 5. Figure 5a shows the position of three absorbers buried in the phantom. The three absorbers buried at a depth of $20\mathrm{mm}$ in the phantom. Figure 5b shows the region of light illumination (closed circles) and the region of detection (slashed mark region). Images obtained using noncontact B-TRIS, in which individual shadows of three absorbers can be visualized clearly, are shown in Figs. 5c and 5d [Fig. 5c, pseudo colored image; Fig. 5d, black and white image]. Furthermore, the position and shape of shadows obtained by noncontact B-TRIS did not change even when the relative positions of sample and light illumination were changed (data not shown).

Fig. 5

(a) Position of absorbers buried at the depth of $20\mathrm{mm}$ in the phantom. (b) Illustration of light illumination position and receiving areas of noncontact B-TRIS. Closed circles indicate the region of illumination and the slash mark indicates the region of detection. (c) Pseudo color image obtained with B-TRIS. (d) Grayscale image of (c). Wavelength of $805\mathrm{nm}$ was used. Time window of noncontact B-TRIS was $600\mathrm{ps}$ (between 1800 and $2400\mathrm{ps}$ ).

Further, to evaluate the spatial resolution of noncontact B-TRIS quantitatively, we placed two absorbers in the phantom at the depth of $20\mathrm{mm}$ (inset of Fig. 6 ). The profiles of the reflected light intensity along the line connecting the centers of absorbers are shown in Fig. 6. The time window was set between 1800 and $2400\mathrm{ps}$ , where the reflected light intensity is at its maximum at a depth of $20\mathrm{mm}$ (see Fig. 8). One can clearly notice the two separate peaks when the distance of two absorbers is longer than $20\mathrm{mm}$ . However, two absorbers could be discriminated from a two-dimensional image even when the gap was $15\mathrm{mm}$ .

Fig. 6

Profiles of the reflected light intensity along the line connecting centers of the absorbers. Inset is an illustration of two absorbers placed in the phantom at the depth of $20\mathrm{mm}$ . Wavelength of $805\mathrm{nm}$ was used, and the time window of noncontact B-TRIS was $600\mathrm{ps}$ (between 1800 and $2400\mathrm{ps}$ ).

3.4.

Sensitivity Depending on the Sample Position

The uniformity of the sensitivity throughout the target area obtained by noncontact B-TRIS is shown in Fig. 7 . We measured peak intensity changes in backscattered light from a single absorber that moved across the field of view, as shown with arrows (trace A and B) in the inset of Fig. 7, where the illumination positions are shown with closed circles. The sensitivity of B-TRIS is relatively flat near the center and it gradually decreases as the target position approaches the edge of the target area. Furthermore, the position of the absorber measured by B-TRIS coincided the actual position with accuracy $<2\mathrm{mm}$ as described before.

Fig. 7

Top, illustration of illumination (black circles) and detection (white plane) positions of B-TRIS, and traces of absorber (A and B). Bottom, changes in reflected light intensity accompanying the movement of absorber. Experimental conditions were the same as those of Fig. 5.

3.5.

Depth Measurement

Normalized temporal changes in light intensity depending on the depth of the absorbing samples are shown in Fig. 8 . As the depth of the absorber increased, the backscattered light path and the time of flight also increased. The peak value of the backscattered light intensity decreases exponentially and the time-of-flight profile is broadened with increasing absorber depth. This result indicates that the present noncontact B-TRIS provides not only the two-dimensional distribution of the absorbers but also their depth information.