Patients and Physicians alike are looking for the best protocols to identify disease and to track therapy. Often these tools employ expensive nuclear diagnostics such as Positron Emission Tomography (PET) and Single Photon Emission Computed Tomography (SPECT). These two methodologies have become the gold standard diagnostic tool in combination with anatomical imaging such as computed tomography (CT), however they are still rather inaccessible to most of the world due the cost and necessary infrastructure to produce isotopes and maintain these instruments. While we push PET and SPECT technology into whole body and exceedingly multiplex capable imaging systems, newer and more cost-effective nuclear diagnostics are still very much needed worldwide.
Back in 1933, Pavel Cerenkov made a fortuitous accident, when he identified blue light coming from a solution which glowed under external radium irradiation. In 1958 he won the Nobel prize in Physics along with Ilya Frank and Igor Tamm for their discovery. Today named after Cerenkov after Pavel Cherenkov, Cerenkov radiation occurs when a charged particle travels faster than the speed of light in a medium other than a vacuum. This phenomenon was stuck mostly in the field of physics for decades until the availability of single photon counting EMCCD sensors. It was only in 2009 that Cerenkov imaging entered biomedical research and as some of the first researchers in this field saw its potential. Soon after few clinical examples were shown by Spinelli et al. and us (Thorek et al.) proving many reviewers wrong that this could never be done in patients. Fortunately, the vast majority of radioisotopes used at Memorial Sloan Kettering Cancer Center produce Cerenkov light. Having shaped the field, we were eager to provide a large clinical trial to prove the feasibility and potential of the method.
Despite a wide array of “bright” radiopharmaceuticals available for imaging, Cerenkov luminescence is very weak, on the order of a billion times less than fluorescence, requiring the patient to be in complete darkness to exclude stray ambient light. Any light from any other source, even a LED on the camera or watch, would ruin the image. In our prior first clinical work on a few patients that required the room to be dark, the door sealed and the lights to be switched off in the corridor and the surrounding rooms. Unfortunately the corridor led to the main offices, the patient restrooms and to a clinical PET scanner this approach turned out to be not very feasible in a clinical setting. This prototype imaging set-up was unsafe due to the dark environment and could still not completely isolate the patient from ambient light. As Cerenkov luminescence is an optical method with a continuous emission spectrum, it suffers from the traditional issues of absorbance and depth dependence in tissue as any other optical methods. Thanks to NIH funding (R01 CA183953), we were able to join forces with Lightpoint Medical to design a camera system with an enclosure that is easy to operate, light proof and made from just plywood, aluminum foil, two thin curtains, painted black, with a traditional procedure chair inside. The enclosure was given the nickname the spaceship amongst others by our patients. The other half of the camera system included an Andor EMCCD camera coupled to a low scintillation glass fiberscope with a filter wheel. Preclinical testing was very straight-forward with phantoms, allowing us to capture Cerenkov luminescence from many of the isotopes used in the clinic with ease. What we did not know was how this could translate to patients as adding the fiberscope to the Cerenkov camera reduced the ease at which we could image, though we gained greatly camera flexibility. More concerning was the potential loss of signal in the relatively long fiber (ca 2 m), a known phenomenon in all fiberoptics and usually not an issue due to the use of strong excitation light – a luxury, we did not have.
To our surprise, when we started imaging patients receiving Iodine-131 for adjuvant thyroid ablation, the Cerenkov signal appeared where the standard of care SPECT identified uptake of iodine. We were encouraged by these results and expanded our imaging to patients with head and neck cancers, neuroendocrine tumors, and even metastatic prostate cancer, including three new radiotracers never before imaged by Cerenkov. Each new radiotracer provided new challenges based on the isotope brightness, biodistribution, clinical dose for imaging or therapy, and also where it may be located in the patient. With each patient imaged for each isotope, we gained a better understanding of what lesions might be possible for imaging. As these images had to be acquired for 5 minutes to 15 minutes, we also needed the patients to be comfortable yet stationary, so the fiberscope was incredibly helpful for getting the right imaging angle while ensuring the patients were comfortable as well. We kept track of each patients imaging distance and position for future imaging sessions.
We also utilized optical filters to explore the Cerenkov spectrum emanating from patients with the idea to further improve Cerenkov luminescence imaging, e.g., by gaining information on the depth of the lesion. As Cerenkov light is emitted in a continuous spectrum from the UV through to the infrared, we assumed that deeper lesions would be shifted to the red part of the spectrum. Our previous clinical trials had only used a 600 nm filter to reduce scatter. We were fortunate to get several patients who were willing to sit for imaging with each filter, often taking an hour and a half for the whole series. Ultimately the 600 nm long pass filter was most useful in reducing scatter in the image overall, but we could not convincingly assign lesion depth with any of the filters without knowing already the lesion’s signal intensity.
By March 20, 2020, we had nearly 100 patients imaged but due to COVID-19, every clinical protocol not directly benefiting patient health was put on hold. We switched to processing images and refining the performance of the camera with phantoms for each radiopharmaceutical used in patients. Just before lockdown commenced, we were able to install a high-end PC in one BML’s apartment for image processing, and some gaming. Quickly it became clear that we needed an image processing method to harmonize how the images were processed, and we fortunately had Ben Mc Larney on the team who built an ImageJ program to automate the work. With a script in hand, we quickly identified the 8 x 8 binning mode we had tested on a few patients led to negligible loss in resolution while shortening the imaging time three-fold. This standardized data set furthermore allowed us to then correlate the Cerenkov intensity from each patient image relative to the radiotracer standard uptake value and depth of the lesion when measured by a radiologist. In particular, the isotopes used in therapy needed to be recorded as the average radioactivity present in the field of view from each patient, as the dose administered or brightest lesion led to poor correlations. Further refinements could be made for lesion depth, but our goal was to define how the Cerenkov camera operated compared to a PET-CT or SPECT-CT.
The willingness of the MSKCC’s Molecular Imaging and Therapy Service to house the scanner for some time and most and foremost of the patients who volunteer their additional time during their visit at MSKCC made this whole project incredibly rewarding and even fun. Many came back for subsequent imaging sessions as well, and even enjoyed being transported out of the bustling Upper East Side of Manhattan to wherever they went inside the enclosure. Each imaging session was incredibly enriching for us too as we got to hear their stories while we sat just outside of the enclosure. We were able to image the weakest light with this enclosure camera system, and we hope that many more like it will be built to help screen patients and provide imaging access where previously prohibitive.