Comparison of Cherenkov excited fluorescence and phosphorescence molecular sensing from tissue with external beam irradiation

NIR fluorescent probes IRDye 680RD, IRDye 700DX, IRDye 800CW and phosphorescent probe PtG4 (Lebedev et al 2009, Esipova et al 2011) were prepared as stock solutions in the standard phosphate buffer (PBS, pH7.2) at 100 μM concentration. The absorption and emission spectra of the probes are shown in figure 2. The tissue simulating phantoms were prepared using PBS, 1% v/v Intralipid^© (Fresenius Kabi, Uppsala, Sweden) and 1% v/v whole blood (Lampire Biological Labs, Pipersville, USA). A tube containing the solution of the probe (~1 ml) at a concentration between 0.1 and 25 μM was placed inside the phantom at a depth which was varied between 1 to 25 mm from the phantom surface being imaged.

Zoom In Zoom Out Reset image size

All experiments were performed using a linear accelerator (Varian Linac 2100CD, Varian Medical System, Palo Alto, CA) based at the Norris Cotton Cancer Center, within the Dartmouth–Hitchcock Medical Center (see figure 1(a)). The initial beam cross-section was either 10  ×  10 mm² or 20  ×  20 mm², incident on the top of the inclusion tube, and adjusted for each experiment as needed. The beam energy was 6 MV x-rays, and the radiation dose rate chosen was 600 monitor units per minute (MU min⁻¹), which is standard for patient treatment. The CEL spectral measurement system has been described previously (Zhang et al 2012), but briefly, the luminescence optical signal was collected by a fiber bundle consisting of nineteen silica fibers, 200 micron diameters packed tightly together in a circle at the tissue phantom end. The bundle was 15 m long, connecting out to the entrance of the detector. The fiber end was placed at the exterior surface of the phantom container or on the animal skin for in vivo studies. This optical fiber bundle relayed the optical signal to a vertical line of fibers at the entrance slit of the spectrometer (Acton Insight 300 mm, Princeton Instruments, Acton, USA), and the spectrometer coupled the spectrally resolved light to a gated ICCD (PI-MAX3, Princeton Instruments, USA). The ICCD was cooled to  −25 °C, and the intensifier set to its maximal gain. The ICCD chip was of 1024  ×  256 pixels with acquisition using all 256 rows binned to sum the signal into a single spectrum. Radiation from a clinical 2100CD LINAC was delivered in 3.25 μs bursts at a repetition rate of 360 Hz. So the ICCD was synchronized with the LINAC for detection of the light, with luminescence detection using an acquisition delay and gate width particularly set for either continuous wave (CW) or time-gated mode. In CW CEL detection, the acquisition was coincident with the LINAC pulse so that the Cherenkov and fluorescence were captured at the same time, in the 3.25 μs bursts of signal. But when temporally delayed luminescence was being detected, the acquisition gate start was delayed to ensure no detection of the Cherenkov signal into the ICCD. In this latter time-gated CEL detection, the acquisition delay was increased to 4 μs, beyond the time of the LINAC pulse, and the acquisition gate width was expanded to 50 μs to capture as much of the delayed phosphorescence as possible. Since Cherenkov emission was instantaneous with the 3.25 μs radiation pulse, the time-gated detection effectively removing all background Cherenkov from the detected signal. This set up is as shown in figures 1(a) and (b).

In tissue simulating phantom study, CEL from 2500 radiation pulses was integrated on the CCD chip prior to read out of the integrated signal, requiring acquisition for several seconds. This process was found previously to provide an optimal signal level without saturation. The background signal was measured using a tube containing PBS with the same volume of probe.

All animal studies completed here where approved by the Institutional Animal Care and Use Committee at Dartmouth. To assess basic in vivo detection capability, the simulation of a tumor was created in vivo, using Matrigel mixed with PtG4 at 20 μM final concentration, injected subcutaneously into rat tissue. The rat tissue was injected with 250 μl of the Matrigel-PtG4 mix on the right flank and kept warm for 1 h while the Matrigel solidified. In order to guarantee the stability of the experimental process, the CEL detection experiments were completed immediately after animal euthanasia. To study the dose efficacy for CEL in vivo, the tissue was irradiated with doses which varied from 7200 pulses (about 200 cGy) down to 1 pulse (about 0.02 cGy). This last dose is the lowest value that the LINAC can reliably deliver. In each of these, the signal was acquired using the time-gated mode with phosphorescence detection.

Each spectrum was processed by an important sequence of steps, which have been determined through previously published studies. These steps included temporal median filtering over 5 successive acquisitions, followed by background subtraction from a signal without luminescence, and wavelength smoothing through the implementation of a moving window average filter spanning ten neighboring points. The probe efficacy for CEL in the tissue simulating phantoms and in vivo was evaluated by using the signal-to-noise ratio (SNR) as a standard metric, which was defined by the differencing process of signal minus background at each wavelength, integrated over the wavelength band where the emission peak was found, and then divided by the standard deviation. This is described as,

$$\begin{eqnarray}&&\text{SNR}={\int}_{{{\lambda}_{1}}}^{{{\lambda}_{2}}}\left[{{I}_{S}}\left(\lambda \right)-{{I}_{B}}\left(\lambda \right)\right]\text{d}\lambda /\sigma \end{eqnarray} \tag{ 1 }$$

where, I_S and I_B are the spectrum of the signal with and without probe respectively, and λ₁ and λ₂ are the bounds of the integrated wavelength range, and σ is the standard deviation. In the concentration sensitivity experiment, these spectra were integrated separately after subtraction of the background, and the standard deviation value for each was estimated by the variance between these samples. In the concentration sensitivity experiment, the standard deviation value for each was estimated by the variance between these samples. In the case of the experiment varying the object depth, the background changed with position of the object and the beam, so an individual background was taken for each beam position, and the standard deviation value for each position was estimated by the variance of 5 successive background acquisitions. In the animal model study, the estimation of standard deviation was the same as the method used in the concentration sensitivity experiment.

The absorption and emission spectra of IRDyes and PtG4 are shown in figures 2(a) and (b). The IRDyes and PtG4 can all efficiently absorb red and NIR light in the 600–800 nm wavelength band. As listed in table 1, compared with fluorescent probe IRDyes, phosphorescent probe PtG4 (emission maximum 772 nm) has a larger Stokes shift.

Figure 3 is the measured background Cherenkov emission spectra under CW and time-gated detection in the tissue simulating phantoms. In CW detection of the background, the Cherenkov emission during the radiation pulse was detected at the same time, so that there is a continuous emission across the visible and NIR spectrum with intensity varying as noted early (Jelley 1955). From this emission, it is obvious that the absorption of these probes were all in the range of Cherenkov emission within the tissue, so that they could all be used with this excitation light source. The absorption in 500–600 nm spectrum region was mainly from oxygenated hemoglobin in tissue simulating phantoms. When compared with the results of CW detection, Cherenkov emission was completely isolated from the radiation pulse under time-gated detection.

Zoom In Zoom Out Reset image size

The raw spectrum of CEL exhibits bloom of counts as shown in figure 4(a) (red arrow), which was removed by temporal median filtering, see figure 4(b). In addition, the smoothing method is helpful to reduce the low pass noise after temporal median filtering as shown in figure 4(c). The spectra shown below were all processed by using temporal median filtering and smoothing.

Zoom In Zoom Out Reset image size

The measured spectra of molecular probes with CW detection are shown in figures 5(a)–(d). After background subtraction, the absorption and emission spectra of the molecular probes appeared as shown in figures 5(f)–(i). Compared with CW detection, the PtG4 phosphorescence signal was effectively isolated from the Cherenkov emission under time-gated detection as shown in figure 5(e), even without background subtraction.

Zoom In Zoom Out Reset image size

According to the background subtracted spectra, the luminescence intensity of probes were integrated at the range of 713–790 nm for IRDye 680RD, 708–805 nm for IRDye 700DX, 850–890 nm for IRDye 800CW, 750–850 nm for PtG4 under CW detection and 730–880 nm for PtG4 under time gated detection. Figure 6(a) shows the results of the integrated intensity of CEL for different concentrations of IRDyes probes and PtG4 under CW and time-gated detection. The slope of concentration with integrated intensity saturated at higher concentrations, so measurement in this range would be problematic in tissue. According to equation (1), the SNR in concentration efficacy experiment was estimated using the standard deviation of the integrated intensity of the signal. The SNR versus concentration is shown in figure 6(b), with lines connecting the similar data points from the different dyes. In this experiment, the lowest concentrations used were 3.3, 1.6, 6.3 and 0.2 μM for the agents IRDye 680RD, IRDye 700DX, IRDye 800CW and PtG4, respectively. These were all with CW detection. When compared with CW results, the lowest concentration used was 0.1 μM for PtG4 under time gated detection. The SNR versus concentration plots were extrapolated to estimate the minimum sensitivity at which SNR  =  1, and these ultimate limits of detection are what was listed in table 2.

Zoom In Zoom Out Reset image size

Figures 7(a)–(e) show the background Cherenkov subtracted luminescence spectra at different depths for IRDye probes and PtG4 at the fixed concentration of 6.3 μM. The absorption and probe luminescence signals are decreased with depth increase. The luminescence intensity of probes were integrated at the range of 720–780 nm for IRDye 680RD, 710–850 nm for IRDye 700DX, 825–900 nm for IRDye 800CW, 760–810 nm for PtG4 under CW detection and 750–860 nm for PtG4 under time gated detection. In figure 8(a), the integrated fluorescence and phosphorescence intensity were plotted with depth. In depth efficacy experiment, the SNR was estimated by using different σ. Compared with using standard deviation of the integrated intensity of signal as σ, the SNR was reasonable by using the standard deviation of the detected background as σ. The SNR versus depth are shown in figure 8(b). In this study, the detectable depth was 10 mm, 20 mm and 5 mm for IRDye 680RD, IRDye 700DX and IRDye 800CW with CW detection, and PtG4 has the detectable depth of 20 mm with CW and 25 mm with time gated detection mode. The depth sensitivity was estimated for SNR  =  1, representing an ultimate lower limit, and the results are listed in table 2.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We embedded PtG4 in Matrigel solutions for in vivo measurement. The Matrigel-PtG4 phosphorescent spectra were measured 5 times and the temporal median filtered spectra of Matrigel-PtG4 (data not shown) were obtained to calculate the background, using a least-squares fitting algorithm. The original phosphorescent spectra were acquired for doses varying from 200 cGy down to 0.02 cGy. When including the number of the temporal median filtering steps used in data processing, the total radiation doses delivered for each background subtracted spectra was varied from 1000 cGy down to 0.1 cGy. The smoothed background subtracted phosphorescent spectra for different radiation doses are shown in figure 9(a). The result clearly shows the phosphorescent peak at 772 nm for Matrigel-PtG4. The integrated intensity of 80 nm spectral window centered on the emission peak is plotted for each radiation dose in figure 9(b). The integrated intensity was offset by 600 to make the last three data point at lower dose range visible in the figure. Least-squares linear fitting was applied and the results of the fitting are listed. The integrated intensity of Cherenkov-excited phosphorescence was directly proportional to the radiation dose and the monotonically proportional dynamic range was 4 orders of magnitude. The SNR of Matrigel-PtG4 phosphorescence for different radiation dose is plot in figure 10. The signal was detectable when radiation dose was 7 cGy, under the conditions stated here. The ultimate lower limit dose was 4.6 cGy when estimated for SNR  =  1 with linear fitting.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size