SNMMI Podcast Series

Part 1: Quantitative SPECT Imaging with Dr. Stephen Graves & Dr. Benjamin Auer

December 02, 2022 SNMMI Season 1 Episode 4
SNMMI Podcast Series
Part 1: Quantitative SPECT Imaging with Dr. Stephen Graves & Dr. Benjamin Auer
Show Notes Transcript

In part 1 of a conversation with Dr. Stephen Graves and Dr. Benjamin Auer, our two experts discuss quantitative SPECT, its place in regards to PET, requirement, and current status.

Welcome to the latest episode of the Society of Nuclear Medicine and Molecular Imaging Podcast Series. My name is Dr. Stephen Graves. I'm a medical physicist and assistant professor at the University of Iowa in the Department of Radiology. I'll be your host today, and the focus of the podcast will be on quantitative SPECT imaging.  The role for quantitative capabilities and single photon imaging has rapidly evolved in recent years with the rise of new radiopharmaceutical therapies, as well as new radiopharmaceuticals for neurological imaging. There are now several commercial manufacturers of SPECT CT systems who are providing solutions for quantitative SPECT imaging as well as third-party software platforms for doing quantitative image reconstruction. As such, we feel that this is a particularly timely and important topic. And so I'm pleased to say that today our guest to help address this topic will be Dr. Benjamin Auer. 

Dr. Benjamin Auer is the director of nuclear medicine physics at Brigham and Women’s Hospital in Boston, and an instructor of Radiology at the Harvard Medical School. Dr. Auer completed his PhD in physics and molecular imaging at the University of Strasbourg/France in 2017, after which he completed a postdoctoral fellowship with Michael King at the University of Massachusetts Medical School. Dr. Auer is currently an intern in the Society of Nuclear Medicine and Molecular Imaging, physics, instrumentation and data sciences Council, and he's known for his significant expertise in single photon imaging, including novel instrumentation, Advanced Image reconstruction and quantitative methods. Welcome, Dr. Auer, and thank you for joining us.

Thank you, Dr. Graves for a nice introduction and for having me today. So it is my real pleasure to discuss with you about the fascinating topic of quantitative SPECT, and I hope that the auditors will have like a better understanding of what it is actually, and how valuable it has the potential to be clinically.

Great. So let's, let's get into some topics here. So first off, you know, maybe it's, it's obvious to some listeners, maybe, maybe it's not to others, but, you know, for a long time, positron emission tomography, PET imaging has been considered the, the quantitative modality in nuclear imaging. And it's really done a really remarkable job of implementing robust quantitative methods in routine clinical practice. So if that's been the case, for nearly 40 years at this point, why do you think we need quantitative SPECT imaging now in 2022?

| DR. BENJAMIN AUER 3:01 - 10:46 RUNS: 465 SECONDS
Yes, so that's a very good question. So, before answering this, I think it is important to define what quantitative imaging means right.  So nuclear medicine imaging, so PET and SPECT, provide the ability to measure in a non-invasive fashion physiology and then biochemical processes in vivo right. Here like detectionof the gamma rays emitted within the patient via like direct gamma, gamma-ray emission for SPECT and positron animation for PET. So, the fundamental concept of quantification in nuclear imaging relies on estimating the activity concentration of the radiotracer per unit volume right. And thus, to convert like images in counts per pixels into activity concentration, so [Becquerel] per ml unit.  So PET and SPECT system are not, we'd say, intrinsically quantitative.

This means that they are limited for measuring activity concentration without additional effort, as we will see later on. And as you mentioned, the concept of quantitative PET/CT has existed for many years, and has been widely used in clinical routine, for example, in the diagnosis and treatment response monitoring of multiple cancer or in nuclear cardiology.  So, the question I have is then, what about SPECT and why has it taken longer to develop, and why is it not widely used to this day, right?  So in terms of quantitative capability, the potential for PETs was emphasized, we'd say from the beginning, in SPECT it has taken longer to develop and the reason for me include like simpler correction for photon attenuation, you know, of the dual photons emitted in the annihilation process and coincidence detection that could be determined like from a transmission scan usually and this is easily implemented for PET and PET has been greatly assisted by the simplicity of correcting for photon attenuation and then the early PET systems were strictly 2D transaxially oriented with very little inclusion of scatter within the imaging plane and hence no need for further scatter correction right.

And finally, I would say that the nature of PET I mean early PET imaging was mostly research oriented. And these early studies aim to provide quantitatively accurate images right and the system had to be calibrated in the same units, namely radioactivity concentration per unit volume rate Bq/ml.  So, in contrast SPECT systems were more challenging to apply attenuation correction with acquisition consisted relatively high fraction of scatter photons in photo peak. So, typically it's 30 to 40% for Tc-99m, and this basically leads to the necessity for scatter correction algorithm right. And also the clinical orientation that most of the clinical SPECT studies in the old days were performed with the need for rapid scan report thus avoiding to spend like long periods of time applying advanced processing techniques to produce the quantitative reconstruction. So, computation power and associated costs was also a problem at that time to apply this correction routinely.  So, in short, SPECT scans have been traditionally interpreted without any correction for attenuation and scatter, and this has led the physicians to learn to read around certain effects in the reconstructed images such as, you know, attenuation artifacts in myocardial perfusion imaging. So, you have just learned to read around these without really using correction. So, in addition, SPECT is further complicated by the necessity for modeling and testing of corrections, you know, at different photon energies and photo peaks and also collimators because you have to test for basically each nuclide and collimator sets and this is different as opposed to the constant 511 Kev annihilation photon energy consistent across all PET studies, regardless of the radionuclide.  So for mainly these reasons, I would say PET has been known as quantitative but that you know that SPECT is not. However, I would say today, several advanced correction techniques have become widely available and I've enabled SPECT to join PET as a quantitative imaging modality, quantitative SPECT as a great opportunity in combined diagnostic and treatment applications. Although PET has a significant sensitivity advantage and higher spatial resolution over SPECT. SPECT has some advantages over PET so for example, physical half-lives of many SPECT radionuclides are generally longer and more aligned with the biological half-lives of physiologic processes of interest. Radiotracers are readily available and do not require relatively close proximity to a medical cyclotron and a rapid distribution network. And there is potential to simultaneously perform like multi tracer studies with different nuclides examining in, you know, examining different biologic pathway in a single imaging session.  So for example, we can assess the cerebral blood flow with Tc-99m HMPO and at the same time, we can assess uptake in the striatum via like dat scan with I-123.

The last thing is that SPECT systems are also of lower costs and have a much greater I would say clinical availability worldwide . The last thing I'm going to say is accurate and reproducible quantification has really been the key objective since the early days of nuclear medicine imaging. So both for PET and SPECT, and the ability to accurately quantify total radionuclide uptake in SPECT imaging has multiple important applications such as for disease diagnosis, monitoring and treatment responses based assessment, and also dosimetry for radionuclide therapies. So there are so many potential clinical application and it will be hard to enumerate all of them today. Right. But yeah, it's really, like, an exciting area.

I think that's a great summary for the topic. It's really pretty remarkable that we're able to take these imaging systems that fundamentally are counting things – counting events that are in the photo peak and events that are outside of the photo peak – and transforming count wise information into 3d quantitative distributions of how much radioactivity there is in a given region. It's easy to take for granted. Obviously, there are a lot of corrections that have to be applied to take a raw reconstructed data set or raw sets of projection data and generate a quantitative activity map. A lot of that was enabled with the development of iterative reconstruction techniques in the in the 90s and early 2000s. We've, we've come a long way with ability to incorporate those corrections into iterative reconstructions. So, can you outline for us what some of those corrections are and what the current state of the field is with regard to correcting for physically degrading factors?

| DR. BENJAMIN AUER 11:53 - 19:51 RUNS: 418 SECONDS
Yeah, absolutely. So, I would say that quantitative SPECT/CT has been made possible, you know, thanks to the following technological developments I would say.  So, the availability of co-registered CT data for using attenuation correction also availability of fast scatter correction mainly like window-based techniques such as you know, dual-energy windows DEW and triple energy window TEW, also like improved digitized detector performance plus mechanical and electronic stability. And as you said, improved reconstruction algorithm that can basically incorporate the underlying physics of imaging into the image formation process. Which is really not the case with the old reconstruction approach, you know, known as filtered back projection/FBP. So, also increasing computing computational power that allows sophisticated algorithms to be implemented and use clinically ina reasonable time right.

 And finally, continuing increase utilization of PET/CT, which has demonstrated, the clinical potential of quantitative radionuclide imaging, encouraging like renewed interest in producing similar measures with SPECT, so, we can say that PET/CT has really lead the way or force quantitative SPECT/CT imaging. So, I would say while correction for attenuation and scatter makes by far the greatest impact on quantitative quantification in SPECT, there are a number of other factors that need to be considered to produce like quantitative images.  So, imaging studies of patients who have just received radionuclide therapy where relatively high activities of radionuclides are usually administrated are typically affected by what we call dead time. So, these are caused mainly by the signal pileup in the gamma camera detector and electronics and daytime losses can be significant and cause an underestimation in the dose calculation in dosimetry studies and therefore, for any posttherapy imaging study where quantitative information is required, dead time correction should be performed. Also SPECT as a poor spatial resolution compared to PET imaging. So, that leads to erroneous apparent decrease in the reconstructed radioactivity concentration due to the limited spatial resolution. This is what is called partial volume effect or PVE. And it is a problem when imaging objects which are less in size that three times the spatial resolution of the imaging system.
 So, for PET for SPECT, this typically means that source objects less than 40 to 50 millimeter in size will be underestimated. So, correction in the form of recovery coefficient apply to reconstruct the data based on, you know, knowledge of the true size of radioactivity estimated from the CT may be used to provide like an estimate of the true activity quantification in small structures. So, basically, we constrain the reconstruction with anatomical prior, and this approach works great with high contrast structures on CT, but, I will say that this correction remains difficult to implement in practice.  So, modeling the distance-dependent spatial resolution of SPECT, you know, in reconstruction, and there's correcting for it may address some of the partial volume effect impact, but it is an ongoing issue for SPECT quantitation and PET as well because it has also like limited spatial resolution right, but the spatial resolution is better than that of SPECT system.  So, despite this limitation, SPECT/CT system imaging, you know, Tc-99m today report like quantitative accuracy to within plus minus 5% of the true radionuclide concentration, and I would say this is equivalent to the accuracy of current PET/CT system. And then, once you have, you know, all the correction available, which is the case today, as you said most of the vendors provide such correction routinely on their system. So, then yeah, the last step is to enable quantitative measurement.  So, all SPECT images reconstructed in counts per pixel must be converted to activity concentration, so, Bq/ml . So, there are like multiple ways to do this conversion, and the simplest of which is to use a basic conversion factor in reconstruction. You know, by acquiring a test source usually, as simple objects such as uniform cylinder for an unknown amount of time and then reconstructing this acquisition, we can then determine a conversion factor between the activity in Bq of the test source measured in reference dose calibrator and the resulting counts in a 3D image of that test source. So, this method is called like cross-calibration, because you calibrate the SPECT against dose calibrator and then once that conversion factor is obtained, it is assumed that under similar acquisition condition, the same conversion factor can be used to convert data from an actual patients scan and we will go back I think to these methods later on during that podcast.

 Unfortunately, generally in SPECT such conversion needs to be done like for each nuclide /collimator set, this is different than for SPECT for PET, because for PET you use like the image of  the 511 KeV annihilation photons, so, only a single energy and there is collimator used in PET imaging. So, also dose calibrator must be properly calibrated and must use the proper settings and geometries to accurately measure the amount of administrated activity right.  So, this is really important and SPECT system and dose calibrator must also be periodically tested here like quality control, and calibration testing, recommended by the manufacturer, and I would say in NM accreditation like the ACR and standard instrument calibration and quality control tests include like tests of uniformity, center of rotation, calibration, multiple head registration, also you have to register the SPECT and the CT if you are using like a Hybrid system. And of course, the CT system has also to be tested regularly. You know, so you're making sure that your instruments used for calibration are working as expected.

 | Music up AD BREAK - RUNS: 34 SECONDS  
SNMMI, through its Value Initiative, continues to support YOUR efforts to transform the science and practice of precision nuclear medicine for diagnosis and therapy to advance patient care.
  Whether it's working to ensure adequate and appropriate coverage and reimbursement for nuclear medicine procedures or sustaining and growing a diverse and qualified workforce prepared for current and future diagnostic and therapeutic nuclear medicine need, SNMMI is dedicated to improving your profession as you dedicate yourself to improving the lives of others. Together, let's continue to transform the profession. Visit and lend your support with a year-end, tax-deductible gift today.  Music down

You mentioned a lot of different corrections that need to be applied to convert the raw count data to activity concentration. I think we could spend a lot of time discussing in any one of the corrections that you mentioned, but just briefly, I think you talked about how SPECT scanners in general, have a lot less sensitivity and yet we still need to worry about dead time effects. So, I wonder whether you could explain this, which has to do with the collimator of course, reducing the number of photons reaching the crystal and also, as well as the difference in scintillation design where you have a PET system with a bunch of individual scintillation detectors and a conventional SPECT system that says one large crystal So, can you talk about why that induces dead time issues?

| DR. BENJAMIN AUER 20:44 - 22:57 RUNS: 133 SECONDS
Yeah, absolutely. So, I would say that, yeah, dead time correction is particularly important for radionuclides which also multiple photon emissions such as a I-131 as photon that included in the energy window also contributes to dead time. So, dead time is really affected by the detection and rejection of scattered photon and then time is really basically the camera will not be able to respond to any signal increase due to processing time, because it takes some time to emit the light right from the scintillator, and also it takes some time to processthe signal to form the image. So, dead time correction is important for theranostic application. But one thing I can add on that is this is true for what I would call conventional technology, such as you know scintillator plus PMT, but with CZT. So Cadmium, Zinc Telluride detector technology, there is virtually no dead time effect, because the conversion of the energy deposited by the gamma-ray into the detector is directly converted into a signal. So, there is like a direct conversion. And this is different than with conventional technology, where the energy deposited by the gamma ray will be converted into a number of visible photons, and then the lights will be converted into an electronic signal. And also, you will have like some processing of the electronics signals, you need to amplify that signal. So that it can be used to form the image.

One of the other corrections that you mentioned was attenuation correction. With modern hybrid imaging systems, we have both the nuclear medicine imaging system and a CT scanner attached. You know, that hasn't always been the case. I think the first SPECT CT scanner came onto the market in the early 2000s, I believe it was the GE Hawkeye, and before that we just had standalone planar and SPECT imaging systems. You know, and around that time, you also had hybrid PET CT and PET systems. And over time, the commercial manufacturers of these systems have sort of done away with the PET-only systems and really, the only thing you can buy these days is a PET CT.  So I wonder whether you see that being a trend that is going to happen in the SPECT imaging space? And if that's not the case, and maybe it has something to do with the cost of imaging and the availability and the utility of non-quantitative imaging, I wonder whether you can explain your thinking on that topic.  
| DR. BENJAMIN AUER 24:01 - 30:11 RUNS: 370 SECONDS
So, this is a really interesting topic and I would say that you know, there have been like, recent changes as I will describe. So, just to start from the beginning, so, the CT has really quite a critical role to enable like, accurate quantitation. So, as you mentioned, typically the transmission scan is obtained from an x-ray CT system attached to the SPECT system and this is why we called hybrid SPECT/CT system right. So, correction for attenuation is performed based on the attenuation map calculated from these transmission images of a patient acquired during the same image session. So, registration between the SPECT and the CT is kind of straightforward because the patient is just laying down, you know on the bed and the bed is moving from the SPECT area to the CT area also, the CT used for attenuation correction is low dose and it is used to ensure like minimal radiation exposure to the patient.  So, that as you know, it is sufficient and accurate to correct the attenuation of the SPECT data. So, attenuation correction are introduced during typically reconstruction of the SPECT data -- they are included into an iterative reconstruction algorithm. So, basically, you model attenuation during reconstruction and then you correct for that effect as well. So, the CT provides also anatomical information and this is very valuable for diagnosis and to perform like segmentation for dosimetry or other types of study.

So, in our department, when we do like you know, a SPECT study, the physician will always require the CT because the CT provides like anatomical information and this is part of setting up the diagnosis. And another thing is CT is important for segmentation because some of the organs cannot be accurately contoured from the SPECT data. And it is also easier to have like some sort of semi-automated, we'd say approach a segmentation approach based on the CT.  So, I'll ever I would think that PET only system will likely stay around for a while to be honest, some of the studies do not necessarily required accurate connotation and you know can be done via a simple planar imaging, so, without the need to rotate the head around the patient. So, for example, gastric emptying studies where the idea is to evaluate the time it takes for food to empty out of thestomach at different, you know, time points. So, in our department, we have like two SPECT only systems and we use them exclusively for planar imaging, such as you know, bone studies or gastric emptying studies. So, also there have been like multiple simple attenuation correction approach developed for SPECT-only systems that have been reported to perform with relative good accuracy compared to CT-based attenuation correction.  So, for example, the Chang method for brain imaging, these methods do not take into account the heterogeneous density of the brain structures and also patients specific attenuation that are crucial for accurate quantitation correction in whole body imaging, so, they are less important for brain SPECT because of the smaller attenuating volume and more homogeneous density and so, yeah one thing also to mention is the AI So, artificial intelligence-based attenuation correction that have been released recently such as the one name true corr for the cardiac dedicated CZT SPECT system, the DSPECT this was released by Spectrum Dynamics and this approach correct for attenuation based on solely the SPECT data and does not provide like a synthetic attenuation map. So, you won't be able to access then the anatomical information, but the approach is capable to correct for attenuation from the SPECT data on the So, without the need for any transmission transmission scam. So, compared to conventional CT-based attenuation correction approach, this AI method was found to lead tolike very good results about like, you know, a mean absolute error of 4% and the development of such artificial intelligence methods suitable for SPECT attenuation correction is currently of great, great interest.
 However, I do see like a significant increase in hybrid SPECT/CT system demand and availability in any future as we have discussed, this will be the system of choice for quantitative imaging needed for theranostic, for example, and other studies for which accurate quantitation would make it would be very beneficial.

You rightly point out that there are a lot of SPECT and planar imaging applications that don't require a high degree of quantitative accuracy. You mentioned gastric emptying bone imaging, you know, things like renal scans that are sort of relative measures. I think there's a good case to be made that the single imaging systems will stick around.


That wraps the first half of the conversation between Dr. Stephen Graves and Dr. Benjamin Auer. Stay tuned for the second half of the conversation about quantitative SPECT imaging coming soon.  This has been SNMMI Podcast Series. Keep an eye out for future episodes where we’ll continue to tackle hot-button issues in the Nuclear Medicine and Molecular Imaging profession. Thanks for listening.