Future directions in Nuclear Medicine Instrumentation

Transcription

1 Future directions in Nuclear Medicine Instrumentation Where are we going - and why?

2 First, the disclosure list My group at the University of Washington has research support from: NIH DOE General Electric Medical Systems Philips Medical Systems Zecotek Photonics

3 Next, the disclaimer - my crystal ball is cloudy

4 Evolution or Revolution Commercial systems tend to be evolutions of designs rather then radical departures Research systems tend to be more adventurous in terms of looking at new technologies and scanner designs

5 Why bother? Measurements of metabolic function rather than anatomy Tracer concentrations => measurements do not alter physiology Used for both detection and estimation tasks Wide variety of tracers that can be used - and new and more specific tracers are under development

6 The measurement task PMT crystal collimator Single probe Gamma camera PET scanner The goal is to measure/image the amount of tracer per unit volume

7 Too much stuff for one talk Nuclear Medicine instrumentation covers a very wide range of devices and applications For example: Portable imaging systems Compact cameras for general imaging Mobile Neuro PET scanners also being looked at for portal imaging Probes and hand held cameras for use in the OR Hybrid imaging systems PET/CT SPECT/CT PET/MR SPECT/MR PET/optical SPECT/optical PET/photo-acoustic and the list goes on Dedicated portal imaging systems PET scanners taking data during proton or other heavy particle therapy sessions

8 How to look into the crystal ball and reduce the scope of this talk We will look at basic detectors and then systems concepts Single probes and portal imaging we will ignore for today We will start with a quick look at planar gamma cameras, then move into SPECT and PET.

9 More definition of scope? general imaging system application areas: 1) pre-clinical (e.g., animal) imaging; 2) dedicated limited area scanners (e.g., neuro-imaging or breast imaging); 3) whole body imaging. We can further subdivide each of these three target areas into two additional categories: 1) academic development 2) commercial development.

10 Basic detector parameters Energy resolution Scatter rejection Interaction kinematics Spatial resolution Intrinsic Stopping power Sensitivity Timing Dead time Coincidence detection (PET)

11 Basic detector types Dominated by scintillators NaI(Tl) BaF2 BGO LSO GSO LYSO LaBr3 LFS LuAP LuI3 Effective atomic no. (Z) Linear attenuation coef. (c ~0.56 Density (gm/cm 3 ) Index of refraction Light Yield (% NaI(Tl)) Peak wavelength (nm) Decay constant (ns) Hydroscopic Yes Slight No No No No No No No Yes The ideal is dense, fast, cheap, bright. We are still looking for it.

12 Solid state versus scintillator Solid state devices in some systems For example: CdZnTe (CZT) and CdTe Major advantage is energy resolution Major disadvantage is cost (and for PET - stopping power and timing)

13 Planar imaging Are single head cameras a dying breed? Certainly few are currently sold, but still useful for portable studies and procedures like lymph mapping. New development is focused on compact, portable systems - mostly by smaller vendors

14 Special purpose detectors Examples: Inter operative imaging and scinto-mamography CZT is playing a large role in the development of such systems

15 Large area planar detectors Most CZT detectors are pixelated - one element per resolution cell. Events at edges are easily detected. Most NaI(Tl) designs are monolithic crystals read out by an array of photosensors Position typically determined with weighted Anger logic and can not resolve events near the edge of the crystal Alternatives are statistical estimators that can also estimate depth and resolve events near the edge of the crystal (more on that later) detector - NaI(Tl)

16 The problem of collimators Collimators are usually the limiting factor in spatial resolution for planar gamma cameras (includes SPECT). The resolution is a function of the distance between an object and the collimator. One fix is to use a E - E detector - so far primarily an academic R&D topic Scatter detectors being considered include Silicon slabs Absorption detectors range from NaI(Tl) to CZT and Ge Another approach is to model the collimator response and include it in a statistical reconstruction algorithm - an approach in academic laboratories and implemented by virtually all of the major gamma camera vendors. Without With PSF modeling

17 Planar issues -> SPECT CZT only in dedicated cardiac and mammography scanners or pre-clinical systems (cost an issue) Collimator response modeling (much more accurate in tomographic reconstructions) Improvements in PMTs, count rate, uniformity of collimator response all carry over

18 Systems for specific tasks Combinations of collimators and detectors have lead to specialized imaging systems - particularly for cardiology

19 SPECT/CT Hybrid imaging is the area of most development SPECT/CT was developed before PET/CT First commercial systems were non-diagnostic CT Newer systems all offer diagnostic CT

20 Lots of SPECT/CT options Diagnostic or nondiagnostic CT - which is the best fit?

21 SPECT/CT - future directions? The trend is clearly moving toward quantitative accuracy Improved reconstruction algorithms with corrections for: Scatter (beyond current multi-window or ESSE) Collimator response, including septal penetration Dead time Motion Slow dynamic (limited by sensitivity and maximum SPECT rotation speeds)

22 NaI(Tl) forever? For standard, large field-of-view SPECT or SPECT/CT systems, it is hard to find a lower cost, effective scintillator than NaI(Tl). For specialized systems such as pre-clinical scanners and mammography systems, alternatives are feasible - but almost all are based on non-scintillators such as CZT. NaI(Tl) forever - basically yes!

23 MR/SPECT? Pre-clinical systems being developed at university laboratories Commercial versions in development (e.g. Gamma-Medica). Most systems being developed use solid state detectors (CZT) and multi-pinhole collimators. Options for solid state photomultiplier (SiPM)??

24 What is a SiPM? Most common are geiger-muller mode APDs Each micro cell is biased such that any charge created in the microcell causes a discharge Surrounding silicon quenches the discharge Hundreds to thousands of microcells per square mm Gains about the same as a PMT Capable of fast timing ( < 100 ps has been achieved) Work in magnetic fields Low bias voltage ( < 200 VDC) Challenges: high capacitance current sources dark current temperature sensitivity

25 Will SiPMs replace PMTs? The cost per element is potentially low since it is a CMOS process However, for a given application you still have to cover the scintillator surface area Total cost may or may not be lower due to the amount of silicon needed 2. For many hybrid applications (e.g., MRI + nuclear imaging), SiPM technology is currently generating the most interest 3. For some new technologies, such as depth-of-interaction detectors, SiPM arrays may well be the best alternative. The answer is for systems like large field of view gamma cameras, PMTs are likely the photosensor of choice. For small systems, some hybrid applications, and some PET technologies, more likely to use SiPMs instead of PMTs.

26 PET systems Essentially all systems currently being sold are PET/CT systems

27 Improving image quality Do we need to improve intrinsic spatial resolution - for whole body generally not, for dedicated neuro scanners and pre-clinical scanners generally yes. Do we need to improve sensitivity - generally yes for all applications How? Increase solid angle (smaller ring diameters, longer axial FOV) Add time-of-flight (TOF) to improve image SNR Do we need to improve system modeling in image reconstruction - yes.

28 Smaller ring diameter -> DOI Penetration of gamma ray results in miss position of the line of response (LOR) Scatter in detector complicates the problem (multiple interactions) Adding DOI measurement capability allows more accurate positioning of the events

29 Some (not all) DOI approaches Pulse shape discrimination (PSD) PMT/APD Two or more layers of scintillators that have different decay times. Use PSD to identify which crystal is the site of interaction. events time Some scintillator combinations that have been used: LSO/LSO NaI(Tl)/LSO BGO/GSO GSO/LSO/BGO/CsI(Tl)

30 Some (not all) DOI approaches Some other discrete crystal approaches optically coupled semi-transparent coupling optically decoupled position sensitive PMT readout devices readout devices Examples of several single-ended DOI detector designs: (a) Comb-slit detector design; (b) Segmented detector design; (c) University of Washington light sharing technique.

31 Depth (mm) Detector 1 Detector 2 Some (not all) DOI approaches Shameless pitch for one of the UW approaches Design interface to... A B... share large fraction of light... share very little light SiPM array Photon Count Isosceles Triangle with Black Paint Challenge is scatter in an array of crystals and the number of data channels (>40/detector module and there are 70 to 140 modules in a scanner) Preliminary data (2009) got 5 mm DOI accuracy. New goal is 2 mm accuracy.

32 Some (not all) DOI approaches Photosensors on both ends of a crystal array

33 Some (not all) DOI approaches Another shameless UW pitch

34 Why so much attention to DOI? Necessary for best resolution in pre-clinical scanners (to the extent of working on systems that can estimate the position of first interaction) Probably necessary for best TOF resolution (account for change in timing based on where in a scintillator the light is produced) Necessary for best resolution in dedicated neuro or other dedicated scanner applications May be needed if smaller diameter detector ring, longer axial FOV whole body scanners are developed.

35 TOF Increases effective SNR - shown by several groups As shown by the Penn group, TOF improves rate of convergence of statistical reconstructions with TOF resolutions on the order of 650 ps. AS TOF resolution improves, SNR improves. Current commercial scanners operate at 500 to 700 ps Goal is to get below 300 ps - will likely need DOI corrections for time skewing in the detectors

36 PET/MR Many different solutions Some design options: Use fiber optics to pipe light to sensors Use solid state sensors Put electronics in the magnet with the sensors

37 PET/MR Concept of a whole body PET/MR system from EU project web site

38 PET/MR First commercial system, a head insert by Siemens, is a bit bulky. More compact electronics and improved PET detectors are needed.

39 Where is instrument development headed? Probes - small improvements Planer imaging - focus on specialized instruments (e.g., mammography, systems for OR, etc) SPECT - focus on SPECT/CT, improved collimators, system and software improvements to provide quantitative SPECT. Development of methodologies to support dynamic SPECT including revisiting ring geometries and other schemes to increase senstivity. PET - focus on TOF improvements, possible changes in axial lengths of scanners, increased commercial development of more task specific scanners (e.g., portal imaging, mammography, etc) Hybrid imaging - refinement of XX/CT systems, propagation of XX/MR systems, and work towards other combinations such as XX/optical, XX//acoustic, and tri-mode systems such as PET/SPECT/CT For all - continued refinement of image reconstruction algorithms with ever more accurate modeling of the system.

40 That s All Folks! The best kind of small animal imaging :-)