Cerenkov Luminescence Imaging for Accurate Placement of Radioactive Plaques in Episcleral Brachytherapy of Intraocular Tumors

Abstract

Purpose: The purpose of this study was to determine the feasibility of using Cerenkov luminescence imaging (CLI) to facilitate plaque placement during episcleral brachytherapy of intraocular tumors.

Methods: Ruthenium-106 (Ru-106) decays to rhodium-106, which in turn emits high-energy beta particles. When the electrons propagate through the eyewall, the so-called Cerenkov effect leads to emission of weak light, which can be captured by high-sensitivity charge-couple device (CCD) cameras. Enucleated porcine eyes were prepared with tumor phantoms made of melanin-containing gelatin. The anterior portion of the globe was removed, and different Ru-106 plaque types (designated CCA, CCB, COB, and CIA) with activities ranging from 6.8 to 16.7 MBq were sutured to the sclera overlying the tumor phantom. The globe was placed in a transparent container with saline. CLI was performed through the anterior opening of the eye using a cooled electron-multiplying CCD camera.

Results: Exposure times between 5 and 60 seconds produced good quality images of the Cerenkov light. There was a linear relationship between plaque activity and Cerenkov radiance. The perimeters of the CCA and CCB plaques could be seen clearly as circles of light symmetrically surrounding the tumor phantoms. Notched COB and CIA plaques led to images revealing their actual positions in relation to the optic disc and ciliary body, respectively. Simulated plaque tilting resulted in diffuse demarcation of the light.

Conclusions: The study indicates that CLI is a feasible method to ensure accurate placement of Ru-106 plaques in brachytherapy of intraocular tumors. CLI may offer a new tool to improve and document plaque placement, both perioperatively and postoperatively.

Episcleral brachytherapy is currently the most widely used conservative treatment for uveal melanoma, and the method may also be used to treat other types of intraocular tumors, such as retinoblastoma, choroidal hemangioma, and solitary metastasis.¹ Spherically curved plaques containing seeds of iodine-125 or a thin foil coated with ruthenium-106 (Ru-106) deliver a high radiation dose to the tumor while reducing the dose to adjacent normal structures. Correct dosimetry and accurate placement of the plaque relative to the tumor are the two most important factors to achieve local tumor control and reduce the risk of radiation-induced side effects. 

Over the years, different techniques have been used to ensure that the radioactive plaque, sutured to the scleral surface, covers the tumor base with the required tumor-free margin. The first step is to locate the intraocular tumor, which for anteriorly located tumors can be achieved relatively easily by transpupillary or transocular transillumination.² Melanin-containing tumor tissue absorbs the transilluminated light and casts a distinct shadow onto the eyewall so that its margin can be marked on the episclera. For more posteriorly located tumors, however, the procedure is more complicated, necessitating the use of indirect ophthalmoscopy combined with scleral depression or transscleral transillumination.³ For this purpose, Damato et al.⁴ developed a dummy plaque with small holes for the insertion of a right-angled transilluminator. To ensure proper localization of the radioactive plaque both during and after primary surgery, several techniques have been described in published reports, such as 2- and 3-dimensional ultrasonography,^5–8 magnetic resonance imaging,^9,10 and plaque-mounted light-emitting diodes.^11,12 Nevertheless, for choroidal melanomas located in the posterior pole or juxtapapillary area, it is challenging to locate the tumor and to position the plaque correctly, which may explain why radiation failure is more common in this location.¹³ Furthermore, initially well-localized plaques may become displaced or tilted away from the sclera during the time the plaque is sutured to the eye (typically 3–7 days), which could reduce radiation dose to the tumor and cause local tumor recurrence.¹⁴ Current methods to determine the correct plaque position are time consuming and demanding and rather inaccurate due to generally low image resolution and different types of imaging errors.¹⁵ Thus, there is a need for improved methods to facilitate and document proper plaque placement both at the time of insertion and during the course of treatment. 

Cerenkov luminescence imaging (CLI) relies on a physical phenomenon called the Cerenkov effect, where charged particles (i.e., electrons) induce visible light when traveling through a dielectric medium. These low-light emissions are induced along the charged particle's track whenever the particle travels faster than the speed of light in the medium.^16,17 For electrons, propagating in tissue with a refractive index of approximately 1.4, the threshold for Cerenkov emission generation is approximately 0.219 MeV.¹⁷ Many radionuclides used for radiation therapy and positron emission tomography imaging satisfy this condition, rendering optical emissions that can be collected using a high-sensitivity charge-coupled device (CCD) camera to capture radionuclide uptake.¹⁸ Moreover, the effect is also present during external beam radiation therapy, allowing imaging of the beam entrance on or exit from a patient's body during breast cancer radiotherapy.^19,20 As early as 1971, Burch²¹ reported using Cerenkov emission in connection with the so-called radioactive phosphorus uptake test to diagnose uveal melanoma. Uptake of phosphorus-32 in proliferating cells could be assessed by placing an optical light guide in front of the cornea, guiding the Cerenkov emission to a sensitive photomultiplier tube. Burch postulated that the Cerenkov emission was generated in the vitreous humor by the emitted electrons from phosphorus-32, having maximum energy levels of approximately 1.7 MeV and average energy levels of approximately 0.7 MeV. 

Ru-106 and iodine-125 are currently the isotopes most frequently used for episcleral brachytherapy for uveal melanoma. Ru-106 will decay, with a half-life of 373.59 years, to the daughter nuclide rhodium-106. Rhodium-106 has a half-life of approximately 30 seconds and emits highly energetic electrons with a maximum energy of 3.5 MeV.²² Hence, the electrons from rhodium-106 have the ability to induce Cerenkov emission all along their paths through the sclera into the tumor tissue and farther into the eye. We postulate that CLI could be of clinical value in Ru-106 plaque brachytherapy of intraocular tumors. Herein, we report the feasibility of using CLI to facilitate and document proper plaque placement in ex vivo porcine eyes. 

Methods

Porcine Eyes

Twenty eyes from domestic pigs (Norwegian Landrace), each with a live weight of approximately 75 kg and 6 to 7 months of age, were obtained from a local abattoir. Enucleations were carried out within 12 hours post mortem, and the eyes were stored at 4°C in a moist chamber until preparation. Each eye was examined under a binocular dissecting microscope, and excess tissue, including conjunctiva and muscles, was removed from the sclera. Ten eyes were processed fresh, whereas the others were fixed in 70% ethanol in order to obtain a better stability of the globe, allowing for easy sectioning and firm attachment of the Ru-106 plaques. The anterior portion of the globe was removed and discarded together with the lens and vitreous. Thereby, the posterior fundus could easily be visualized and prepared for imaging. The study adhered to the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research. 

Uveal Melanoma Phantoms

Two types of uveal melanoma phantoms were made: orthotopic tumor phantoms in the suprachoroidal space and in situ tumor phantoms located on the retinal surface. Methods for the preparation of orthotopic tumor phantoms have previously been described in detail elsewhere.^23,24 Briefly, porcine skin gelatin powder (product no. G1890, type A; Sigma-Aldrich Corp., St. Louis, MO, USA) was stirred in distilled water at 37°C. Natural melanin, isolated from the ink sac of cuttlefish (Sepia officinalis; product M2649; Sigma-Aldrich Corp.), was then suspended in distilled water at 37°C and stirred for 15 minutes. By adding appropriate amounts of aqueous melanin to the gelatin solution, a dark brown, turbid suspension of melanin and 15% (wt/vol) gelatin was made. The suspension was stirred in a water bath at 37°C until used. The orthotopic tumor phantoms were made in fresh, whole eyes by slowly injecting 0.75 mL of the newly prepared melanin-gelatin suspension through a 3-mm scleral incision directly into the suprachoroidal space, between the sclera and the choroid. Immediately after the injection, the scleral incision was closed with a preplaced suture, and the eye was put into cold saline to complete the gelation of the phantom (Fig. 1A). 

Figure 1

 

(A) Cross-sectioned porcine eye with a dome-shaped, orthotopic uveal melanoma phantom located within the suprachoroidal space. (B) Bisected porcine eye with an in situ uveal melanoma phantom located on the inner retinal surface. (C) CCB-type plaque sutured to the posterior pole of a porcine eye. (D) Notched COB plaque sutured in close apposition to the optic nerve.

The in situ tumor phantoms were made in fixed eyes, after the anterior portion was removed, by simply depositing 0.05 to 0.1 mL of the melanin-gelatin suspension directly on the inner retinal surface (Fig. 1B). 

Ruthenium-106 Plaques

The Ru-106 plaques used in the study were manufactured by Eckert & Ziegler BEBIG GmbH (Berlin, Germany) and were of the following types and activity values (at the time of the experiments): type CCA, 6.8 MBq; CCA, 9.2 MBq; CCB, 6.8 MBq; CCB, 16.7 MBq; COB, 7.7 MBq; and CIA, 6.4 MBq. The plaques are 1.0 mm thick. The CCA and CCB plaques are circular with diameters of 15.3 mm and 20.2 mm, respectively. The COB plaque is 19.8 mm in diameter and has a notch for the optic nerve. The CIA plaque is 15.3 mm in diameter and has a shallow notch for treatment close to the iris. In all plaques, the diameter of the irradiation zone is somewhat smaller than the diameter of the plaque itself, leading to an annular inactive rim of 0.75 mm for the CCA, CCB, and CIA plaques and 1.0 mm for the COB plaque. Each plaque has two eyelets through which they were anchored to the sclera using 6-0 polypropylene sutures (Prolene; Ethicon, Somerville, NJ, USA) (Figs. 1C, 1D). 

Imaging System and Processing

Images were acquired using an electron-multiplying CCD (EMCCD) camera (product DU-897; iXon, Andor, Ireland) equipped with a camera objective lens (Xenon, 25 mm, f/0.95; Schneider Kreuznach, Bad Kreuznach, Germany). In the experimental set up, the camera was placed above a transparent plastic container filled with saline and containing one of the Ru-106 plaques sutured to a porcine eye (Figs. 2A, 2B). The set up was shielded from ambient light by encapsulating the equipment with a 1-mm–thick black plastic mat, providing optimal absorption of ambient lighting. 

Figure 2

 

(A) Schematic illustration of the experimental set up showing the EMCCD camera and the saline container with a Ru-106 plaque sutured to a porcine eye. (B) Illustration of the principle of Cerenkov emission. The Cerenkov light (blue arrows) is generated along the tracks of the beta particles (red arrows), which are emitted from a thin Ru-106–coated foil (red dotted line) within the plaque.

Acquisition settings were altered between three fixed settings. First, the EMCCD camera was operated as an ordinary CCD camera, using a long exposure time of 60 seconds. Second, the exposure time was 15 seconds with 250× gain on the electron-multiplying circuitry. Third, the exposure time was 5 seconds with 250× gain on the electron-multiplying circuitry. Hardware binning of 2×2 was applied for all image acquisitions of Cerenkov emission. 

Image backgrounds were subtracted and normalized with acquisition time. In order to extract the radiance from the Cerenkov luminescence images of radioactive plaques, the pixel intensities of the central part of the plaque image were averaged. The relationship between the Cerenkov radiance and activity of the plaques was assessed using the Pearson coefficient of correlation. The image processing and statistical analysis were performed in Matlab version 8.1 software (Mathworks, Inc., Natick, MA, USA). Throughout this paper, the images showing Cerenkov luminescence were false-color–coded using a rainbow color map. The false-color Cerenkov intensity was then overlaid on a grayscale photograph of the eye, acquired using the same camera and setup as the CLI image. 

Results

Cerenkov Emission From Ruthenium-106 Plaques in Relation to Plaque Activity

The Cerenkov emission from the radioactive plaques was assessed by placing each plaque in the transparent plastic container filled with saline and acquiring images using 60-second exposure times (Figs. 3A–C). The radiance was extracted by averaging a region-of-interest in the central part of the plaque and plotted as a function of activity (Fig. 3D). The Cerenkov radiance correlated significantly with radioactivity (r = 0.84; P = 0.038). Some of the deviations were most likely caused by slight tilting of the plaque when positioned in the plastic container. Furthermore, the Cerenkov emission decreased at the circumference of the plaque, because the diameter of the irradiation zone was slightly smaller than the plaque diameter. 

Figure 3

 

Cerenkov luminescence imaging of a CCA type plaque (A), a COB plaque (B), and a CIA plaque (C) placed in a saline container. The color scale to the right indicates camera counts per second. (D) Cerenkov emission for each plaque used in this study is shown plotted relative to radioactivity, normalized with the area of the plaque. Dotted line represents the linear regression of the data (r = 0.84, P = 0.038).

Cerenkov Luminescence Imaging in Eyes Without Uveal Melanoma Phantom

A series of Cerenkov luminescence images were taken of each Ru-106 plaque sutured to the posterior eye segments. Exposure times between 5 and 60 seconds produced good quality images of the Cerenkov emission generated by the different types of plaques. The perimeter of the CCA and CCB plaques could be seen clearly as circular areas of light with different diameters, corresponding to the sizes and locations of the plaques. 

Images of the notched COB and CIA plaques, sutured adjacent to the optic nerve and limbus, revealed their actual positions in relation to the optic disc and ciliary body, respectively. Simulated displacement of the COB plaque relative to the optic nerve and subsequent repositioning of the plaque were performed. The Cerenkov luminescence images (Figs. 4A–C) show the improved position of the plaque as it was moved closer to the nerve. Although the plaque finally was placed in close apposition to the optic nerve (i.e., with the nerve within the notch of the plaque), CLI demonstrated that there was a clear gap between the optic disc edge and the area of Cerenkov emission (Fig. 4C). In most of the images, the optic nerve was visible as a bright spot together with the lighted area of the Ru-106 plaque, caused by a higher transmission of light in the optic nerve than in the rest of the eyewall. It was also seen that the closer to the optic nerve the plaque was placed, the brighter the optic nerve appeared. 

Figure 4

 

Cerenkov luminescence imaging of a notched COB type plaque sutured close to the optic nerve. (A) Plaque is misaligned relative to the optic nerve. (B) Plaque is centered around the optic nerve but not close to the nerve. (C) Plaque is adjacent to the optic nerve. (D) Plaque is tilted away from the sclera, leading to an irregular and diffuse demarcation of the Cerenkov emission. Color scales indicate camera counts per second.

Plaque tilting was simulated by placing a small piece of a transparent plastic tube between the posterior edge of the COB plaque and the sclera. This resulted in the plaque tilting away from the sclera by approximately 3 to 4 mm, which in turn resulted in images displaying an irregular and diffuse demarcation of the Cerenkov emission corresponding to the posterior plaque border (Fig. 4D). 

Cerenkov Luminescence Imaging in Eyes With Uveal Melanoma Phantom

After preparation of the uveal melanoma phantoms, various types of Ru-106 plaques were sutured to the sclera overlying the tumor phantom. The orthotopic type of tumor phantoms in the suprachoroidal space were generally larger and more irregular than the in situ type of phantoms, which presented as uniform, circular tumors on the retinal surface. Given the presence of light-absorbing melanin within the gelatin structure, leading to significant attenuation of the Cerenkov light, both of the types of tumor phantoms were clearly visualized and delineated by CLI. The margins of the orthotopic tumor phantoms were more diffusely demarcated than the in situ tumor phantoms, which had a sharper and more distinct appearance of the tumor shape. 

Figures 5A through 5C illustrate how an initially displaced COB plaque caused an asymmetrical distribution of the Cerenkov light relative to a large orthotopic tumor phantom and how subsequent adjustment and centration of the plaque led to a more symmetrical spread of the light around the phantom. Similarly, Figures 6A through 6C show a small circular in situ tumor phantom, where a perfectly positioned CCA plaque created a halo of Cerenkov light surrounding the phantom. 

Figure 5

 

Cerenkov luminescence imaging of a CCB-type plaque sutured to a porcine eye with an orthotopic tumor phantom (red dashed circle). (A) Sectioned eye. (B) Plaque is misaligned relative to the tumor phantom. (C) Plaque is correctly centered relative to the tumor phantom.

Figure 6

 

Cerenkov luminescence imaging images of a CCA type plaque sutured to a porcine eye with an in situ tumor phantom. The central, blue-green, circular area represents tumor phantom, which is symmetrically surrounded by Cerenkov light emitted from the plaque. (A) Image obtained using the EMCCD camera in CCD mode at an exposure time of 60 seconds. (B) Image obtained using the EMCCD camera in electron-multiplying mode at an exposure time of 15 seconds and an EM gain of 250×. (C) Image obtained using the EMCCD camera in electron-multiplying mode at an exposure time of 5 seconds and an EM gain of 250×. Color scales indicate camera counts per second.

Cerenkov Emission From Ruthenium-106 Plaques at Different Exposure Times

The ability to perform CLI with short exposure times was tested using radioactive plaques sutured to porcine eyes with in situ tumor phantoms. Cerenkov luminescence images were generated using 5-, 15-, and 60-second exposure times. When 5- and 15-second exposures were made, the electron-multiplying capability of the EMCCD camera was employed using an EM gain of 250×. Images of the CCA plaque sutured to the sclera overlying an in situ tumor phantom, acquired using 60-, 15-, and 5-second exposures, are shown in Figures 6A, 6B, 6C, respectively. The longer exposure times had a better signal-to-noise ratio due to the fact that more photons were detected. The shorter exposure times collected fewer photons, but the electronic signals were amplified so that sufficient image contrast was achieved. 

Discussion

In this study, we have shown that it is possible to use CLI to localize radioactive plaques in enucleated porcine eyes. The results indicate that image acquisition with exposure times down to 5 seconds can be used when applying an EMCCD camera. By studying the Cerenkov emission pattern propagating through the eye wall, the plaque position relative to orthotopic or in situ tumor phantoms could be assessed. Consequently, the plaque could be guided to an accurate position given by the symmetrical emission around the tumor phantom or the proximity to the optic nerve. The high-energy beta particles emitted from the Ru-106 plaques generated the Cerenkov light. Radioactive plaques containing iodine-125 or palladium-103 will not induce such light because these plaques emit gamma photons with lower energies than those of the Cerenkov threshold in tissue.^25,26 

For the Cerenkov luminescence images acquired from Ru-106 plaques positioned in saline, the CLI radiance was found to depend linearly on the radioactivity normalized with the area of the plaque. This is in accordance with previous CLI studies of radionuclide uptake.¹⁸ Some deviations of CLI radiance were seen, and these were most likely due to slight tilting of the plaque when positioned in the saline container. In addition, the curved shape of the plaque caused the visible photons to be emitted in an angle, where the camera could not collect the light. Photons emitted in the central part of the plaque will, on the other hand, have a higher probability of being emitted within the numerical aperture of the camera system. This effect was minor when the plaque was used on highly scattering medium, such as tissues, because the photon paths were then randomized. 

The images of the different Ru-106 plaques sutured to porcine eyes showed diffuse emission boundaries. This effect is probably caused by light scattering, which is most pronounced in the sclera.²³ When the plaque was positioned close to the optic nerve, two effects occurred. First, contact between the plaque and the eye wall was reduced due to the more complex anatomy of the posterior pole, rendering a slightly more diffuse emission pattern in proximity to the optic nerve. This is also caused by the nerve sheath and connective tissue, which surround the optic nerve at its insertion into the globe, preventing a closer proximity between the plaque and the nerve itself.²⁷ Second, the optic nerve allowed for a higher optical transmission than the surrounding tissue, as shown by the fact that the emission was stronger from the optic disc. The optic nerve is composed of nerve bundles that effectively guide the light through the eye wall. This led to a lower attenuation of light within the optic nerve than in the rest of the eye wall, and therefore the optic disc appears brighter. The closer to the optic nerve the plaque is placed, the brighter the optic disc appears. This is expected because more photons will be able to propagate into the optic nerve. 

For Cerenkov luminescence images of plaques sutured to porcine eyes with an orthotopic or in situ tumor phantom, the margins of the tumor phantom could be identified by the generally reduced Cerenkov emission within the melanin-containing phantom. The margins of the orthotopic tumor phantoms were slightly more diffuse than those of in situ tumor phantoms. This was most likely caused by the location of the orthotopic tumor phantom within the eye wall, where visible light is affected by the light-scattering properties of the surrounding tissues. The in situ tumor phantom was placed on the inner retinal surface, and therefore, the photons that had propagated through the tumor phantom were subjected to minimal scattering in the vitreous humor. 

In a clinical setting, it is important to limit the exposure time as much as possible so that movement artifacts can be reduced. Unfortunately, Cerenkov emission is inherently weak; hence, care must be taken in order to optimize the collection of these photons. By using an electron-multiplying camera, the weak signal can be amplified after acquisition so that the contrast and signal-to-noise ratio are improved. The possibility of using exposure times as short as 5 seconds is encouraging for future clinical applications. 

In the experiments described herein, the anterior portion of the globe was removed and discarded together with the lens and vitreous. In a clinical setting, the cornea, lens, and vitreous will still be in place, which may affect the transmission slightly. However, this will also lead to higher image magnification and probably an improved visualization of Ru-106 plaques. 

The Cerenkov emission is very weak, and even with optimized photon collection and camera settings, heavily pigmented tissue leads to an attenuation of generated light. In the present study, we have shown that CLI is possible in porcine eyes, which are more densely and irregularly pigmented both in the retinal pigment epithelium and in the sclera than in human eyes.^28,29 Heterogeneous tumor pigmentation can also be problematic when analyzing Cerenkov luminescence images. The pigmentation will absorb Cerenkov emission differently, leading to an emission pattern that may differ from the shape of the tumor. Likewise, hemoglobin within the choroid and tumor vasculature will most likely affect the Cerenkov emission due to high absorption in the blue-green wavelength range.^24,30 On the other hand, CLI is intended to facilitate plaque placement relative to uveal melanoma. Although the Cerenkov emission may vary within the tumor area, the boundaries of the plaque surrounding the tumor, including a free margin, will probably be easy to identify due to the relatively low and homogenous pigmentation of the human eyewall. As there is a linear correlation between the Cerenkov emission radiance and the plaque radioactivity, plaques with a low level of activity will generate weak Cerenkov light. The plaques used in the present study were in radioactivity ranges lower than those normally used in episcleral brachytherapy but still readily detectable by CLI. 

The fact that the Cerenkov emission depends on the level of radioactivity has triggered studies where the absorbed radiation dose is assessed from the CLI measurements.^26,31 Such an approach could be applied for CLI during episcleral brachytherapy as well. However, in order to establish a truly quantitative CLI method, the collection efficiency (e.g., numerical aperture and camera quantum efficiency) needs to be taken into account in addition to rigorous calibration using plaques placed in saline and proper dose distribution calculations. This investigation will be accomplished in future studies. 

In this experimental study, we demonstrated the possibility of imaging radioactive plaques during episcleral brachytherapy. The study was performed using enucleated porcine eyes with the cornea and lens removed, which evidently differs from the clinical setting and may affect the image quality. Furthermore, the effects of blood in the tumor vasculature were not investigated. However, the experiments tried to mimic the clinical situation where the plaques were correctly sutured to the eyes, and we believe that the use of natural melanin in the tumor phantoms further enabled CLI to be simulated in a clinically realistic way. The results motivate two clinical applications: first, perioperative CLI to guide the placement of Ru-106 plaques during surgery. This requires the use of indirect ophthalmoscopy through a surgical microscope to which an EMCCD camera is connected. Second, postoperative CLI through a fundus camera coupled to an EMCCD camera is used to document the position of the plaques and to ensure that they are correctly located during the course of brachytherapy. The ability to guide and verify the placement of radioactive plaques during episcleral brachytherapy based on CLI is intriguing. We are currently in the process of translating this methodology into the clinic in order to determine its potential value for radiotherapy of intraocular tumors. 

Acknowledgments

This work was partly funded by Swedish Research Council Young Investigator Grant VR 621-208-5850. 

Disclosure: J. Axelsson, None; J. Krohn, None