NetForum uses cookies to ensure that we give you the best experience on our website. If you continue to use the site, we'll assume that you are happy to receive these cookies on the NetForum website. Read about our cookies.
NetForum Community
Learn. Share. Optimize.
Log in | Sign up now | Submit content | Contact
Go to similar content

TruFlight PET technology - Clinical implementation of time-of-flight

White Paper
Philips NM Marketing Philips Healthcare

Introduction

During the last 20 years, great efforts have been made to improve the diagnostic accuracy of PET through the development of new data acquisition/processing systems. PET scanners with different performance characteristics have been developed, including the hybrid gamma cameras and partial or full-ring dedicated tomographs, working in 2-dimensional (2D) or 3-dimensional (3D) mode, to meet different requirements and variable clinical workloads. The development of new reconstruction algorithms and improvements in computer speed and storage memory have led to the possibility of collecting and processing coincidence data within very short times, effectively improving image quality and patient throughput (Ref 1). Today, the state of the art systems, such as GEMINI GXL, routinely perform whole body oncology PET scans in less than 15 minutes.

 

Although overall performance of PET and PET/CT scanners has significantly improved, there are certain limitations, especially for heavy patients, where attenuation and scatter effects are increased. Increased patient dose and longer acquisition times for larger patients have shown only limited improvement in image quality as compared to smaller patients. Increasing patient thickness dramatically increases the scan duration needed to maintain detectability and may not be practical (Ref 2). 

Figure 1 Small and large patient comparison PET FDG data clearly demonstrates better image quality in a smaller patient. Data courtesy of Methodist Medical Center, Peoria, IL
Figure 1 Small and large patient comparison
PET FDG data clearly demonstrates better image quality in a smaller patient. Data courtesy of Methodist Medical Center, Peoria, IL

 

In addition, PET/CT clinical applications are expanding beyond FDG oncology examinations. Many of these new applications provide more challenging PET sensitivity requirements due to inherently low count statistics. For example, gated cardiac PET perfusion studies (with Rb-82 and N-13 Ammonia) utilize short lived isotopes and allocate those limited counts across different portions of the cardiac cycle. Dynamic studies, new specific radiopharmaceuticals, low dose isotopes, and other Molecular Imaging applications are further examples of low count studies that would also benefit from increased system sensitivity.

 

Philips Medical Systems is investigating new scintillating crystals, scanner designs, and image processing algorithms in order to overcome these limitations and further improve imaging performance. 

 

In this paper, we present preliminary results of implementation of the TruFlight time-of-flight technology that holds promise to dramatically improve PET image quality in large patients, boost system sensitivity, and significantly decrease total imaging time. Implementation of time-of-flight technology involves a full integration of the new scintillating crystals, photomultipliers, detector electronics and data reconstruction.

What is Time-of-Flight?

In PET imaging, each annihilation of the positron labeled radiopharmaceutical produces two 511 keV photons traveling in opposite directions. These photons are detected by the detectors surrounding the subject. The detector electronics are linked so that two detection events that occur unambiguously within a certain window (typically 5-15 ns, depending on the detector material) are called coincident and thus are determined to have come from the same annihilation. A coincidence event is assigned to a line of response (LOR) joining the two relevant detectors (Figure 2A). In this way, positional information is gained from the detected radiation without the need for a physical collimator.

 

With Time-of-Flight PET imaging, in addition to the identification of coincident events, the actual time difference between the detection of two coincident gamma rays is measured (typical time difference is less than 1 nsec). This time difference is then used in the data reconstruction to more accurately localize the origin of the annihilation (Figure 2B). In order to gain whole body PET image quality benefits from Time-of-Flight information, timing resolution of the system (ability to measure the time difference between two coincident gamma rays) has to be significantly shorter than 1 nsec. A shorter timing resolution results in a higher gain in image quality. Uncertainty in a position along the line of response is equal to half the timing resolution of the system (1 nsec timing resolution corresponds to 15 cm). Image quality improvement can be measured by the Time-of-Flight sensitivity gain that is defined as the ratio of the patient size (body diameter) to the uncertainty in a position (Ref 8). Considering patient body sizes between 20-40 cm in diameter, the sensitivity gain for a Time-of-Flight system with the timing resolution of 650 psec would result in 2-4 times sensitivity gain. 

Figure 2A A coincidence event is assigned to a line of response.Figure 2B Time-of-Flight information is used in the data reconstruction to more accurately localize the origin of the annihilation.
Figure 2A
Figure 2B
A coincidence event is assigned to a line of response.
Time-of-Flight information is used in the data reconstruction to more accurately localize the origin of the annihilation.

Figure 3 IEC phantom data demonstrating the benefits of Time-of-Flight technology. Top row represents data reconstruction without the Time-of-Flight information, the bottom row the same data with the Time-of-Flight reconstruction. Image quality improvement is clearly seen between the two data sets.  Also, the benefits of Time-of-Flight are higher in the larger phantom (35 cm diameter vs. 27 diameter). Data courtesy of J. Karp, Department of Radiology, University of Pennsylvania, Philadelphia, PA
Figure 3
IEC phantom data demonstrating the benefits of Time-of-Flight technology. Top row represents data reconstruction without the Time-of-Flight information, the bottom row the same data with the Time-of-Flight reconstruction. Image quality improvement is clearly seen between the two data sets. Also, the benefits of Time-of-Flight are higher in the larger phantom (35 cm diameter vs. 27 diameter). Data courtesy of J. Karp, Department of Radiology, University of Pennsylvania, Philadelphia, PA

Historical perspective (Ref 2)

Techniques and technology that attempted to utilize differential Time-of-Flight information for reconstructing positron emission tomography (PET) images were developed in the early 1980's. Although the potential image quality improvements were attractive, a Time-of-Flight clinical system was not viable at the time.

 

By 1982, two groups (Washington University and Commissariat a l'Energie Atomique-Laboratorie d'Electonique et de L'Informatique [CEA-LET1]) had designed and built the first Time-of-Flight tomographs. A third group at the University of Texas also designed and built such a system. These systems were optimized for high count-rate imaging of short-lived radiotracers for applications such as cardiac blood flow. The first system put into operation for patient scans was the Super PETT I, built at Washington University by Michel Ter-Pogossian and his colleagues. The Washington Univeristy group went on to design two additional versions of Time-of-Flight systems and the CEA-LETI group developed two basic tomograph designs.

 

As Bismuth Germanate (BGO)-based scanners were refined, it became clear that the Time-of-Flight systems could not provide the same high spatial resolution as offered by the BGO systems. The use of the fast scintillators required for Time-of-Flight also resulted in lower intrinsic sensitivity that was only partially compensated for by the effective gain in sensitivity offered by image reconstruction techniques. The characteristics of the scintillation materials available in the early 1980's had either poor timing resolution (e.g. BGO at approximately 6 ns, which corresponds to 90 cm of positioning error from the center of the field of view -- much too large to be useful for Time-of-Flight imaging of the human body) or an undesirable trade-off between spatial resolution and detector efficiency due to a longer attenuation length (e.g. BaF and CsF). For these reasons, Time-of-Flight research drastically decreased by the 1990's. Recently, there has been renewed interest in using LSO (Ref 4) or LaBr (Ref 5) to achieve timing accuracy of better than 1 ns, while maintaining other attractive properties for PET, such as short attenuation length, fast decay time, and high light output. These developments are bringing Time-of-Flight PET back into consideration as a feasible technique. As new scintillators are being integrated into commercially produced PET machines, the transition to Time-of-Flight is becoming more practical.

Integrating Time-of-Flight technology

As described in previous chapters, the timing resolution of the system is an enabling specification for the Time-of-Flight PET. However, in addition to this parameter, the system has to meet several additional requirements:

 

  • Scintillator has to provide good timing resolution, stopping power and adequate light output
  • Photomultiplier tubes (PMT) have to have fast timing resolution and provide uniform timing response
  • Detector resolution and precise encoding
  • System electronics accuracy and stability

 

 

Full integration of all of these elements will determine the future success of Time-of-Flight technology in clinical PET applications. The commercial Time-of-Flight systems need to deliver significant clinical benefits (shorter imaging time, better image quality in large patients) and perform reliably at the same time.

 

a.  Scintillator:  Stopping power & light output

The following table (Table 1) summarizes the intrinsic physical characteristics of scintillating crystals that are the most important for conventional (non Time-of-Flight) PET imaging:

 

  • Attenuation length for 511 keV energy photons
  • Light Yield
  • Decay time
  • Energy resolution

 

Attenuation length is one of the key parameters determining the sensitivity and dose utilization of the PET scanner. The shorter the length the better is the stopping power of the crystal. High light yield has a direct correlation with image quality and ability of the PET system to accurately position the detected events. Short decay time allows setting a short coincidence timing window in order to reduce scatter and randoms. The reduction of scatter and randoms can also be achieved by utilizing a narrow energy window. This capability is exploited by systems with good energy resolution.

Table 1 Intrinsic performance characteristics of different crystal materials. Data: Photonic Materials web site (November 2005) - www.photonicmaterials.com  *GSO intrinsic energy resolution - Reference 11
Table 1
Intrinsic performance characteristics of different crystal materials. Data: Photonic Materials web site (November 2005) - www.photonicmaterials.com *GSO intrinsic energy resolution - Reference 11

 

GSO, BGO, LSO and LYSO have become crystal materials of choice for conventional PET and PET/CT scanners. Time-of-Flight technology puts additonal requirements on the performance of scintillating materials. Timing resolution (directly related to decay time) becomes the most critical. Therefore, LaBr, LYSO and LSO are the materials that currently can be considered for a Time-of-Flight PET system.

 

 

b. PMT: Timing and uniformity

The choice of photomultipliers for the Time-of-Flight PET system is very significant. It directly affects the timing resolution of the system and its stability. The photomultipliers have to have the shortest possible timing resolution (in the range of 100 psec) and uniform response across the tube. That uniform response will translate into the uniform response across the crystals. The timing resolution will add to the timing resolution of the crystals.

Figure 4. represents performance measurements of the photomultipliers suitable for the Time-of-Flight PET system. It is important to realize the photomultipliers used in today's PET scanners do not meet the requirements of Time-of-Flight.

 

 

Figure 4 Graphical illustration of the performance of
Figure 4
Graphical illustration of the performance of "Time-of-Flight ready" photomultipliers. Two graphs on the left represent the uniformity of timing response. The graph on the righ is a histogram of timing resolution.

 

c. Detector: Resolution/encoding & light collection

A Time-of-Flight PET system has to integrate a conventional PET detector in the design. The better the performance of the detector without Time-of-Flight, the better the overall performance of the scanner. Sensitivity gain in the Time-of-Flight system allows for the design a PET system with better spatial resolution (smaller crystals) while maintaining the gains in the total imaging time.  Philips Pixelar detector design (continuous light guide) is uniquely positioned to provide a base for Time-of-Flight PET.

 

Figure 5 shows a representative picture of the flood obtained from the Philips Time-of-Flight system and a data acquisition block diagram. Please note that the electronics are capable of placing 25 psec tags on each event enabling precise sampling of the Time-of-Flight information.

Figure 5 PET system flood image (left) and data acquisition block diagram with 25 psec time stamp resolution.
Figure 5
PET system flood image (left) and data acquisition block diagram with 25 psec time stamp resolution.

 

d. Electronics: Calibration stability & accuracy

One of the key performance characteristics of the Time-of-Flight PET system is its stability. The research systems developed in the 1980's were able to perform within specifications only for minutes requiring hours of recalibration. Today's technology achieves stable performance over many weeks (Figure 6)  and the calibration takes only 5 minutes. This calibration can be done automatically during the daily quality control procedure for the PET system.

e.  Data reconstruction

The area of data reconstruction has possibly enjoyed the most dramatic improvements since the first generation of Time-of-Flight. New reconstruction methodologies are supported by newly developed algorithms and enhanced computing power. The state of the art Time-of-Flight reconstruction algorithm is based on the event-by-event reconstruction (list mode) utilizing a RAMLA (Row Action Maximum Likelihood Approach) iterative algorithm

 

Reconstruction geometry is defined using the line of response (LOR) approach. In a traditional (non-LOR) method, raw data is rearranged prior to reconstruction and combined to form a uniform, interpolated sinogram. This necessary interpolation step includes a "filtering" effect that degrades lesion detectability and spatial resolution. LOR reconstruction eliminates the interpolation step to preserve the maximum spatial resolution of the system.

 

Philips PET reconstruction continues to be based on "blobs". Blobs are 3 dimensional entities, based on the Kaiser-Bessel function, circularly symmetric replacing cubic voxels. By properly controlling the size, spacing and shape of those functions, one can control the resolution and statistical texture of the reconstructed image.

 

A large technical challenge for Time-of-Flight reconstruction is to convert a very large computational task (~50x compared to conventional PET reconstruction) into multiple smaller tasks and to then distribute those tasks to more computers. The existing GEMINI GXL PET reconstruction server is an important enabling tool for the Time-of-Flight reconstruction because it uses a multiple parallel processor architecture. The major elements of this reconstruction server include:

 

  • "Protocol builder" - where the data and parameters for the major processing steps are prepared.
  • Scheduler - responsible for choosing the "best" processing step and starting it on a particular computer if one is available.
  • Launcher - provides services to the scheduler for management of the processing programs, collecting its diagnostic output and reporting its final result code.
Figure 7 Technical challenge is to convert a very large task (~50x compared to conventional PET reconstruction) into multiple smaller tasks to then distribute those tasks to more computers.
Figure 7
Technical challenge is to convert a very large task (~50x compared to conventional PET reconstruction) into multiple smaller tasks to then distribute those tasks to more computers.

Performance

Integrating the Time-of-Flight architecture as described in the previous chapters enables building of a Time-of-Flight PET system with timing resolution of 650 psec. (Figure 8)

 

Figure 8 Plot of Time-of-Flight data representing system timing resolution of 650 psec and demonstrating 10 cm FWHM uncertainty in a position.
Figure 8
Plot of Time-of-Flight data representing system timing resolution of 650 psec and demonstrating 10 cm FWHM uncertainty in a position.

 

Uncertainty in a position of 10 cm would result in 2-4 times sensitivity gain in patient studies (Figure 9) This approximation is based on the measurements of typical patient body sizes that range from 20 to 40 cm in diameter. It should be noted that the benefit increases with the patient size and/or better timing resolution.

 

Figure 9 Graphical depiction of the sensitivity gain in patients of various sizes. Curves representing 10 cm FWHM uncertainty in a position are superimposed on representative patients whole body scans (non-Time-of-Flight).
Figure 9
Graphical depiction of the sensitivity gain in patients of various sizes. Curves representing 10 cm FWHM uncertainty in a position are superimposed on representative patients whole body scans (non-Time-of-Flight).

Clinical benefits

TruFlight technology demonstrates a potential to revolutionize performance of PET imaging by extracting the true benefits of Time-of-Flight. Significant sensitivity gain (2-4 times as compared to conventional PET) has been achieved in an integrated and stable system.

 

Implementation of TruFlight PET technology in a PET/CT product would provide the following benefits:

 

  • Improved image quality/greater patient throughput
  • Consistent image quality for all patients, especially large patients
  • Enabling molecular imaging applications characterized by low count rates

 

PET/CT patient throughput of 24 patients in a single, 8 hour day would become practical by shortening of the acquisition time to 10 minutes per whole body FDG scan. As an alternative, PET users could choose to scan longer (15-20 minute PET emission time, less than 30 minute PET/CT total scan time) and obtain images with better quality (less noise due to sensitivity gain). Lower doses of FDG could be used for studies where patient exposure is critical (pediatric imaging, screening tests). Faster acquisition speed would also enable routine total body scanning (head to toe). 

 

In addition to overall image quality improvements, Time-of-Flight quality gains are more significant for larger patients. TruFlight technology would close the gap in system performance for patients of different sizes. The reading physicians would be presented with consistent image quality for all patients. Diagnostic confidence and ease of interpretation would improve. Consistent imaging times for all patients would affect a department's workflow, enabling it to schedule patients at equal intervals, eliminating the need to allocate additional time for large patients.

 

Time-of-Flight technology also enables clinicians to image dynamic processes that require fast sampling, delivers longer useful imaging time (dynamic range) from short-lived isotopes and enables the use of very low doses for tracers whose synthesis suffers from low efficiency (F-choline, thymidine, etc.) By enabling imaging characterized by low count rates, TruFlight technology is opening the pathway to new molecular imaging applications.

References

  1. PET instrumentation and reconstruction algorithms in whole-body applications. Gabriele Tarantola, BE, Felicia Zito, MSc and Paolo Gerundini, MD, Department of Nuclear Medicine Ospedale Maggiore di Milano, Istituto di Ricovero e Cura a Carattere Scientifico, Milano, Italy, Journal of Nuclear Medicine Vol. 44(5):756-769.
  2. A quantitative approach to weight-based scanning protocol for PET oncology imaging. Paul Kinahan, Phillip Cheng, Adam Alessio, Tom Lewellen, University of Washington, Seattle, MIC Conference 2005.
  3. Effect of random and scatter fractions in variance reduction using time-of-flight information. J. Kimdon, J. Qi, W. Moses, Ernest Orlando Lawrence Berkeley National Laboratory, Berkeley, CA 94720.
  4. Quantitative potentials of dynamic emission computed tomography. Budinger TF, Derenzo SE, Greenberg WL, Gullberg GT, Journal of Nuclear Medicine Vol. 19(3): 309-315, 1978.
  5. Prospects for time-of-flight PET using LSO scintillator. Moses WW, Derenzo SE.  IEEE Trans Nucl Sci 46(3): 474-478, 1999.
  6. Investigation of lanthanum scintillators for 3D PET. Surti S, Karp JS, Muehllehner G, Raby PS.  IEEE Nucl Sci Symposium and Med Imag Conf 2002.
  7. Image reconstruction and noise evaluation in photon time-of-flight assisted positron emission tomography. Tomitani T.  IEEE Trans Nucl Sci 28(6): 4582-4589, 1981.
  8. Time-of-flight positron emission tomography: Status relative to conventional PET. Budinger TF.  Journal of Nuclear Medicine, 24(1):  73-78, 1983.
  9. Characterization of a TOF PET scanner based on lanthanum bromide. Karp JS, et al.  Department of Radiology, University of Pennsylvania, Philadelphia, PA; Philips Research USA, Briarcliff, NY, MIC Conference Record, 2005.
  10. Time-of-flight PET. Lewellen TK, Seminars in Nuclear Medicine, 1998; 28:268-275.
  11. Performance of a brain PET camera based on anger-logic gadolinium oxyorthosilicate detectors. Karp JS, et al.  Journal of Nuclear Medicine 2003; 44: 1340-1349.


This content has been made possible by NetForum Community.
Share this on: Share your link in twitter Share your link in facebook Share your link on LinkedIn Print Rate this article: Log in to vote

 
Rating:
Votes:
2
Views:
6496
Added:
Jul 1, 2007

Rate this:
Log in to vote
 

White Paper
Gemini TF 16
3D, abdomen, Body, Cardiac, chest, image quality, obese patient, Oncology, phantom, time-of-flight, whole body
 

Clinical News
Best Practices
Case Studies
Publications and Abstracts
White Papers
Web seminars and Presentations
ExamCards
Protocols
Application Tips and FAQ
Training
Try an Application
Business News
Case Studies
White Papers
Web Seminars and Presentations
Utilization Services
Contributing Professionals
Contributing Institutions
Become a Contributor