1.

Introduction

All optical imaging techniques suffer from diminished spatial resolution with increasing depth into tissue because of the severe light scattering present.¹ Yet, in recent studies it has been shown that Cherenkov-based excitation of molecular probes is a feasible way to image deep into tissues, over several centimeters, with submillimeter resolution.² One important application of this is luminescence-based sensing of oxygen over macroscopic depths into tissue. The available depth of sampling is appropriate for imaging oxygenation in lymph nodes. In this study, the feasibility of imaging oxygen in lymph node-sized objects in a human body phantom was assessed to estimate applicability of this luminescence lifetime sensing technology in human imaging.

Cherenkov radiation is produced in all tissues undergoing treatment with Megaelectron Volt photons or electrons,³ generating broadband optical radiation that can be used to excite molecular probes.^4,5 Clinical linear accelerators (linacs) operate in pulsed mode with $\sim 4\text{-}\mu \mathrm{s}$ bursts,^6,7 allowing for excitation of phosphorescent probes that have 10- to $100\text{-}\mu \mathrm{s}$ range emission decay times.⁸ Previous experiments demonstrated this luminescence imaging with Cherenkov excitation in basic tissue phantoms and in rodents,^9,10 to image tissue oxygenation, with time-gated signals that were sensitive to the luminescence quenching by oxygen. The key advantage of the method was to image centimeters into tissue while retaining submillimeter spatial resolution. At the same time, imaging oxygen is relevant in radiotherapy since it is a radiosensitivity factor that correlates with outcome.¹¹

In this work, sheet-shaped radiation beams from a linac were used to produce Cherenkov photons, to excite luminescence of probes in a volume largely confined to the irradiation sheets. Thus, by shaping the x-ray beam into a thin sheet, images of Cherenkov-excited luminescence from a planar slice within the tissue can be observed. Analogous to light sheet microscopy, a series of luminescence images can be taken at different depths, allowing for three-dimensional (3-D) volumetric rendering. This imaging approach has been termed Cherenkov-excited luminescence scanned imaging (CELSI).^2,10 Knowing the position of the excitation plane allows for depth-variant attenuation correction, and postprocessed 3-D images show the luminescence distribution from within the tissue.

The aim of this study was to study CELSI in a human body geometry, to demonstrate the capability of imaging local oxygenation in regions similar in size to lymph nodes, as shown in Fig. 1. Specifically, the effects of depth, concentration, and minimum radiation dose were examined here.

Fig. 1

Measurement geometry for the body phantom is shown with (a) a schematic of the geometry, (b) a photograph of the setup, and (c) the body phantom with skin on (left) and without skin (right) as shown with five tubes fixed onto the lateral rib area.

2.

Materials and Methods

3.

Results

In the depth and concentration ranging studies, CELSI experiments were performed for an intralipid-blood phantom embedded with PtG4-containing 0.5-mm Eppendorf tubes. The results are shown in Fig. 2, where sample luminescence and roomlight images are included in (a), and SNR is plotted in (b). The SNR has an exponential shape and the values plateau to their lowest values at deeper depths, as anticipated, because at some extreme depths the presence of the luminescent region cannot be sensed. Consequently, the SNR should be expected to asymptote to 1.0 at deep depths. The relationship between imaging depth and concentration of PtG4 for $\mathrm{SNR}=1$ is described by the circled plots in Fig. 2(b), indicating that for depths $<1.7\mathrm{cm}$ there was good SNR, but below this depth, the concentration required for $\mathrm{SNR}>1$ increased dramatically. However, signal as deep at 3 cm was detectable, at concentrations just above 100 microMolar.

Fig. 2

Experimental results to investigate image quality versus depth for different concentrations of PtG4 (${\mathrm{C}}_{\mathrm{PtG}4}$) as shown in the legend, showing (a) the room light image and luminescence images of the phantom used at three different depths in the phantom, and (b) the corresponding SNR values extracted from these images are shown as a function of depth by colored dots. In the same graph, the red line shows the ${\mathrm{C}}_{\mathrm{PtG}4}$ at which $\mathrm{SNR}=1$ (values on left $y$-axis).

For the first set of geometry studies investigating phosphorescence of tubes containing PtG4 relative to control tubes containing PBS and a blank, a strong phosphorescent signal was detected. This was anticipated because the stock solution of $25\mu \mathrm{M}$ was a high concentration and was placed at a depth of only 3 mm. At 144 AoC, the SNR was computed to be 59, whereas at 60 AoC the SNR was 66. Overall, the 144 and 60 AoC cases produced similar results, with very similar SNRs, for the experiments with “skin.”

For the second set of experiments investigating different concentrations of PtG4, visible phosphorescent signals were detected from the 5- to $25\text{-}\mu \mathrm{M}$ tubes, whereas the $1\text{-}\mu \mathrm{M}$ tube did not produce a signal with sufficient SNR to be detected. The trend in signal strength, with respect to concentrations on the MIP images was as anticipated, with greater concentrations of PtG4 yielding stronger, brighter signals, with slight variation largely due to the thickness of skin overlying the tubes. Videos of the two sets of geometry studies are illustrated.

The angular sheet-beam scanning methodology that is more suitable for imaging a chest region was used, as shown in Fig. 1. Raw measurements of Cherenkov light and luminescence for four specific slice positions are shown in Figs. 3(a) and 3(b), respectively. Figure 3(c) shows a combined image of CT and luminescence maximum intensity projection (MIP). Scanned image data were reprocessed and overlaid on the 3-D volume of the CT scan, to allow visualization of the recovered positions of the tubes. The recovered images are shown in Fig. 4, with a perspective view in (a), and the three orthogonal views of the CT in (b).

Fig. 3

Raw measurement sequences are shown for (a) Cherenkov emission and (b) luminescence photons, for the sheet at four different slice locations. In (c) a composition image is shown of the phantom CT with color overlay of the CELSI MIP image. A temporal sweep of the signals in (a) and (b) can be seen in the associated video files as the sheet moved across the phantom body.

Fig. 4

3-D images of (a) CT and (b) overlaid with CELSI for three orthogonal views: coronal, sagittal, and transverse. Compositions of CT and CELSI images for varying numbers of light sheets, which correspond to less dose delivery. In these images, the number of sheets were: (c) 25, (d) 50, (e) 100, and (f) 250, respectively.

To pursue a minimum radiation dose, sparse illumination was investigated, i.e., the number of light sheets was desired to be as few as possible. Here, a total of 25, 50, 100, and 250 light sheets were used with equal spacing, i.e., with spaces 2, 1, 0.5, and 0.2 mm, respectively. Figures 4(c)–4(f) show the compositions of CT and CELSI images for different numbers of light sheet. It can be seen that the image quality was reasonable, even for 25 light sheets.

4.

Discussion

Initial studies of depth sensitivity have indicated that CELSI imaging would be viable for a depth of perhaps 2 to 3 cm, likely dependent upon the tissue type involved. Adipose tissues tend to have lower blood volume and consequently reduced absorption attenuation of the light—this may allow imaging down to 3 cm. In contrast, denser tissues would have higher absorption and therefore be more limited to $>2\mathrm{cm}$.

The set of experiments conducted here were initially focused on determining the optimum positioning of the body phantom, couch, gantry, and camera for maximal sensitivity. Three orientations were investigated such as (i) vertical with the gantry directly above the body phantom on the couch at 180-deg reference angle and the camera placed directly facing the left side of the body phantom, (ii) horizontal with the gantry directly facing the left side of the body phantom at 90-deg reference angle and a 45-deg front-surface mirror placed above the top surface of the body phantom, allowing horizontal mounting of the ICCD directly facing the mirror, and (iii) angled with the gantry at 145-deg reference angle with ICCD on the same side. The angular orientation in an epi-illumination configuration was determined to be optimum and used for all subsequent experiments. In this orientation, a Cherenkov sheet was clearly visible translating through the body phantom. This orientation was also hypothesized to result in a strong phosphorescence signal from the simulated lymph nodes, and better approximated tangential breast tumor treatment protocol. The recovery of intensity was highly modulated by the depth of tissue overlying the tubes, and it was apparent that the signal was not linear with concentration because of this effect. However with skin thickness measured by the CT scan and geometry, it is feasible to quantify these values based on depth-dependent attenuation correction.¹⁰

These studies illustrate how imaging of lymph node sized objects may have potential with CELSI imaging in radiotherapy. Sensing their oxygenation is feasible, and further exploration of agents and signal analysis could lead to the next clinical trials in molecular sensing in radiotherapy.