Cerenkov luminescence imaging (CLI) and radioluminescence imaging (RLI) are becoming well-established methods for preclinical in vivo small animal optical imaging, as described in recent reviews.^1–3 CLI and RLI are based, respectively, on the detection of Cerenkov radiation or scintillation light in the visible range. CLI was mainly applied to image the biodistribution of radiopharmaceuticals labeled with beta plus^4–6 and beta minus^7,8 emitters. RLI was developed to obtain the biodistribution of alpha⁹ and gamma emitters¹⁰ and can also be used in combination with CLI to enhance the light output.¹¹

Recent examples of CLI applications to humans were presented in Refs. 12 and 13. In Ref. 12, Cerenkov radiation from the thyroid of a patient treated with ${}^{131}\mathrm{I}$ for hyperthyroidism was imaged, while Ref. 13 presented an endoscopic Cerenkov luminescence imaging approach to detect gastric cancers in patients injected with 2-[18F]fluoro-2-deoxy-D-glucose.

In this work, we present an application of CLI and RLI for the analysis of ex vivo fresh tumor specimens removed during neurosurgery. The main goal of such an approach is to provide a fast molecular imaging estimation of tumor uptake (almost in real time) during surgical procedures. CLI and RLI images were obtained by placing the tumor specimens in an optical imager prototype developed in our laboratory.¹⁴

The detector used in the prototype is a cooled ($-80°\mathrm{C}$) electron multiplied charge coupled device (EMCCD; Andor iXon Ultra) coupled with an $f/0.95$ 17-mm C-mount lens (Schneider Optics). The EMCCD has a back-illuminated $512\times 512$ sensor with pixel size equal to $16\mu \mathrm{m}$, resulting in a detector with total dimensions equal to $8.2\times 8.2{\mathrm{mm}}^2$.

The smallest achievable field of view (FOV) is $\sim 6.1\times 6.1{\mathrm{cm}}^2$, resulting in an image pixel size of $120\mu \mathrm{m}$. To exclude ambient light, the system is mounted on a black light-tight enclosure, in which an adjustable stage is placed for positioning the specimen.

All the CLI and RLI images were acquired using binning = 2 and 5 min of exposure time. A photographic black-and-white image (binning = 2 and exposure time = 0.01 s) was always acquired before any emission image to provide a reference for CLI and RLI. A color image of the sample was also acquired using a conventional camera to show the amount of blood present in the sample.

RLI data were acquired by placing a slab of 1-cm plastic scintillator on top of the specimen. To avoid any contamination, the scintillator was placed $\sim 1\mathrm{mm}$ from the sample. Considering that the plastic scintillator slab is almost transparent, the collected light is thus the sum of Cerenkov and scintillator light produced in the slab. This acquisition scheme ($\mathrm{CLI}+\mathrm{RLI}$) was adopted mainly to maximize the light collected by the CCD.

A patient with a grade 1 brain meningioma localized in the left parietal region was injected with 166 MBq of ${}^{90}\mathrm{Y}\text{-}\mathrm{DOTATOC}$ the day before surgery. The patient participated in an experimental protocol of a radioguided surgery technique for cerebral tumors with the use of a beta minus radiation probe,¹⁵ which uses ${}^{90}\mathrm{Y}\text{-}\mathrm{DOTATOC}$ for diagnostic purposes—contrary to its traditional use with therapeutical quantities for peptide receptor radionuclide therapy. In the new technique in Ref. 15, the ${}^{90}\mathrm{Y}\text{-}\mathrm{DOTATOC}$ injected activity is calculated based on a diagnostic positron emission tomography–computed tomography (PET-CT) imaging with the analogue ${}^{68}\mathrm{Ga}\text{-}\mathrm{DOTATOC}$ radiopharmaceutical that allows evaluation of the activity concentrations in tumors and thus can establish the activity to be administered for detection with the beta minus probe. In our case, the tumor to nontumor standard uptake value ratio was $\sim 25$. The required ${}^{90}\mathrm{Y}\text{-}\mathrm{DOTATOC}$ activities are of a few hundreds of MBq (typically 185 MBq, 166 MBq in our patient), and the ensuing tumor activity concentrations are also suitable for the study of CLI and RLI.

As shown in Ref. 16, meningiomas express somatostatin receptors; thus, the ${}^{68}\mathrm{Ga}\text{-}\mathrm{DOTATOC}$ and ${}^{90}\mathrm{Y}\text{-}\mathrm{DOTATOC}$ uptake in the tumor is specific.

Figure 1(a) shows the ${}^{68}\mathrm{Ga}\text{-}\mathrm{DOTATOC}$ image of our patient, acquired two weeks before surgery. This time is sufficient to guarantee the absence of any interaction between the two radiotracers. A similar time interval was previously considered in a study aimed at comparing the biodistribution of ${}^{90}\mathrm{Y}\text{-}\mathrm{DOTATOC}$ in the same patients under two renal protection conditions.¹⁷

Fig. 1

(a) The coronal, sagittal, and axial PET-CT images of the patient injected with ${}^{68}\mathrm{Ga}\text{-}\mathrm{DOTATOC}$. The PET images were used to estimate the activity concentration of ${}^{90}\mathrm{Y}\text{-}\mathrm{DOTATOC}$ in the tumor during surgery. (b) The T2 MR images of the patient acquired without contrast medium.

${}^{68}\mathrm{Ga}$ and ${}^{90}\mathrm{Y}$ have very different physical half-lives, but the labeling of the same peptide—leading to ${}^{68}\mathrm{Ga}\text{-}\mathrm{DOTATOC}$ and ${}^{90}\mathrm{Y}\text{-}\mathrm{DOTATOC}$—allows us to assume similar, if not identical, kinetic processes in tumors and normal organs as well, as applies for ${}^{111}\mathrm{In}\text{-}\mathrm{DOTATOC}$ used as a surrogate for ${}^{90}\mathrm{Y}\text{-}\mathrm{DOTATOC}$, as sustained by the joint IAEA, EANM, and SNMMI practical guidance on peptide receptor radionuclide therapy (PRRT).¹⁸ The use of ${}^{68}\mathrm{Ga}\text{-}\mathrm{DOTATOC}$ to foresee the outcome of therapy with ${}^{177}\mathrm{Lu}$- or ${}^{90}\mathrm{Y}\text{-}\mathrm{DOTATOC}$ has been ascertained in the community of PRRT and has been discussed in several papers.^19–22 Moreover, in the protocol of Ref. 15, the presumed activity of ${}^{90}\mathrm{Y}$ derived from the ${}^{68}\mathrm{Ga}$ estimates has also been compared with the Monte Carlo simulations of the beta probe sensitivity, showing results in good agreement (data not shown from a paper submitted by Solfaroli Camillocci et al.).

The diagnostic T2 magnetic resonance images of the patient acquired without contrast medium before surgery are shown in Fig. 1(b). The study was approved by the ethical committee of the Carlo Besta Neurological Institute, and informed consent was signed by the patient.

The tumor diameter was $\sim 3.5\mathrm{cm}$ and the volume was $\sim 18{\mathrm{cm}}^3$. From the DOTATOC uptake derived from PET images, we could estimate the activity concentration of ${}^{90}\mathrm{Y}\text{-}\mathrm{DOTATOC}$ in the tumor during surgery to be equal to $4.8\mathrm{kBq}/\mathrm{ml}$.

${}^{90}\mathrm{Y}$ ($T_{1/2}=2.66\text{days}$) decays mainly by beta minus emission with end point energy equal to 2.28 MeV and a small fraction of pair production. Because of the direct interaction with the CCD of bremsstrahlung and gamma rays emitted by ${}^{90}\mathrm{Y}$, a small sparse spike noise pattern is generated during the emission image acquisition. The spikes have been removed by applying a box median filter with a size equal to $3\times 3\text{pixels}$ on the emission data.

The specimens of the tumor removed by the surgeon were imaged using both CLI and RLI as described before. The sample activity could not be directly measured simply because surgery was carried out in a hospital that did not have a nuclear medicine division with appropriate instrumentation. The sample has a dimension of $5\times 11\times 9{\mathrm{mm}}^3$; thus, the estimated activity was equal to 2.4 kBq.

Figure 2(a) shows the photographic picture of the tumor specimen, Fig. 2(b) the Cerenkov image, and Fig. 2(c) the RLI signal. The images confirm an appropriate localization of the visible Cerenkov and radioluminescence light signal within the tumor region.

Fig. 2

(a) A color image evincing the amount of blood present in the tumor specimen, and (b) and (c) the CLI and RLI signals. As can be seen from the images in (b) and (c), there is a good agreement between the localization of the visible Cerenkov and radioluminescence light signal within the tumor region. The color scale is in arbitrary units.

To our knowledge, this is the first published experimental evidence of the possibility of detecting Cerenkov and radioluminescence light from a fresh human tumor specimen removed during surgery.

As expected, the detected light signal is higher in RLI with respect to CLI; we used the same color scale in order to provide a better comparison between CLI and RLI signals. A circular (3 mm diameter) region of interest was drawn at the center of the tumor specimen, and the mean value was calculated for both CLI and RLI. We found that the average values of the CLI and RLI signal were, respectively, equal to 582 and 652 (arbitrary units).

The higher background signal in RLI data around the specimen is due to direct interaction of the high-energy beta particles with the scintillator slab.

In summary, we showed, for the first time, examples of CLI and RLI images of a fresh human brain tumor specimen removed during neurosurgery of a patient previously administered, for investigational purposes, with 166 MBq of ${}^{90}\mathrm{Y}\text{-}\mathrm{DOTATOC}$.

The FOV of $6.1\times 6.1\mathrm{cm}$ allows the acquisition of few surgical samples simultaneously, reducing the total time needed for the imaging process.

As shown by recent Monte Carlo simulation,²³ the use of ${}^{90}\mathrm{Y}$ is ideal considering the higher number of Cerenkov photons emitted in the tissues with respect, for example, to ${}^{18}\mathrm{F}$.

The CLI signal of the specimen is well confined in the tumor region, while the RLI signal is higher but also contains higher background outside the sample. We can conclude that CLI is the best option when the Cerenkov signal is high enough to be clearly detected, while RLI can be a good alternative option when the Cerenkov light is low. More precisely, RLI can be the best imaging approach when the radiotracer uptake is low, when using isotopes with low Cerenkov light yield, or with end point energy below the Cerenkov threshold.

The approach presented in this work could provide a fast molecular imaging method for the estimation of tumor radiopharmaceutical uptake during surgical procedures using Cerenkov and radioluminescence light.