/ Instrumentation Basics 1. Motivations for / imaging 2. What is a / Scanner 3. Typical Protocols 4. Attenuation Correction 5. Problems and Challenges with / 6. Examples Motivations for / Imaging Desire for aligned anatomical and functional images Adam Alessio Department of Radiology University of Washington aalessio@u.washington.edu 206.543.2419 11/27/07 1/36 image shows good anatomical detail but no functional information image contains useful functional information but poor anatomical detail 2/36 Software Based Image Fusion/Registration Registration Software kidneys What is /? A device with shared mechanical components providing aligned anatomic () and functional () images. bladder Anatomy () Function (FDG-) Logistical Challenge (schedule 2 studies, image manipulation, ) Error prone due to topology differences (and patient positioning, disease progression between scans, internal motion) Better to acquire scans at same time with same device / - Hardware Registration 3/36 4/36 1
Aside: Why are images less visually appealing than? A. Resolution is worse: limited by 1. Positron Range 2. Emitted photons are not exactly anti-parallel (slight angular spread) 3. Intrinsic spatial resolution of detectors (high energy photons travel some distance in detectors) 4. Sampling of detectors B. Signal to noise ratio worse: images are noisy/blurry because: 1. Have much less good counts than 2. Have more contributions of bad counts (scattered and random events) (Noisy images are then blurred for visual appeal leading to even worse resolution) What is a / scanner? So why bother? Answer: Sensitivity Typical whole-body system can detect ~2% of all emitted photons, thus we can image almost any biological compound. In comparison looks at an energy-weighted density and MR (mostly) looks at proton (hydrogen) density 5/36 / Scanner consists of separate and Components 6/36 Some Background First / scanner introduced in 1998 by University of Pittsburgh, NCI, and I Systems (Knoxville, TN) Combination of independent scanner components still standard Benefits of /? Primary: Mechanically Aligned Anatomical and Functional images Fusion - images not fused, simply overlaid - alpha blending Siemens/I/Hitachi Biograph/Reveal (BGO or LSO) Phillips/ADAC Gemini (GSO) Gemini TF (LYSO) GE Medical Systems Discovery LS, ST, STE (BGO) 7/36 8/36 2
Multimodality Imaging Benefits of /? Primary: Mechanically Aligned Anatomical and Functional images Fusion - images not fused, simply overlaid - alpha blending Secondary: scan provides attenuation correction information for Reconstruction Much faster than conventional transmission scan (20 sec vs. ~20 min) Less noise than conventional transmission scan Shorter overall scan time Wholebody / ~ 30 min Wholebody ~ 50 min Tertiary: scan used for scatter correction image could add in resolution recovery in image 9/36 10/36 1. Scout scan (5-20 sec) / scan protocol 2. Selection of scan region (1-2 min) Scout scan image scanner FOV The most significant error in data: Attenuation detector ring data acquisition measured counts expected counts 3 or 4. Helical (1-2 min) 3 or 4. Whole-body (6-40 min) 11/36 FDG in patient attenuation by a factor of 10 to 100 (!) Attenuation is mainly due to scatter It is by far the most important effect for both noise (due to reduced counts) and qualitative image appearance (greater than scatter, randoms, deadtime, normalization ) 12/36 3
Effects of attenuation on qualitative appearance: Patient example Attenuation Correction-Transmission Scan Methods Coincident photon Ge-68/Ga-68 (511 kev) Single photon Cs-137 (662 kev) X-ray () (~30-140kVp) Rod source Detectors Transmission: tissue density Emission: FDG (Attenuation corrected) Emission: FDG (not corrected for attenuation) high noise ~12-45 min whole body scan no bias post injection contamination lower noise ~4-20 min scan time some bias post injection contamination Virtually no noise ~20 sec scan time potential for significant bias no post injection contamination There are striking changes in appearance with and without attenuation correction 13/36 HISTORY: Past: Siemens and GE Scanners and original / scanners Now: not available for AC on any new system Past: Philips Scanners Now: not available on any new systems Past: not on Scanners Now: only option with new / systems 14/36 Why are Attenuation Correction Factors potentially biased when derived from Transmission scan? Why are Attenuation Correction Factors potentially biased when derived from Transmission scan? Source for x-rays (tube) is 1. at a lower energy level and 2. a distribution of energies For any given material a stream of 70keV photons will be attenuated differently than 511keV photons Most materials fairly straightforward to convert from 70 kev to 511keV Contrast agents complicate conversion spectra Intensity I 0 (E) 0 X-ray source positron source γ-ray source 30 120 E (kev) 511 662 Linear attenuation coefficients for bone and muscle tissue in the range of 10 to 1000 kev. Photoelectric absorption is the main contributor to the attenuation of photons at lower energies whereas Compton scattering dominates at higher energies. emission data attenuated at 511 kev 15/36 16/36 4
Why are Attenuation Correction Factors potentially biased when derived from Transmission scan? Bias in AC Factors Illustrates that mass attenuation coefficient of iodine is appreciably different from biological tissue at energies while similar at 511keV energy. 17/36 18/36 Converting Numbers to Attenuation Values Linear Scaling Method maps Hounsfield Units to Attenuation correction coefficients at 511keV Flowchart of typical / Operation Attenuation correction factors can be obtained from either (A) -based or (B) conventional transmission source (rare on recent systems) 19/36 20/36 5
Scanner Componen t GEH LS NXi (BGO) Siemens wide-bore *** PICO Accel (LSO) Siemens wide-bore HIREZ Accell (LSO) Philips Gemini Allegro (GSO) Comparing / Scanners* Component** Lightspeed Plus (4,8, or 16 slice) GEH ST wide-bore Lightspeed Plus NXi (BGO) (4,8, or 16 slice) SOMATOM (2 or 16 slice) SOMATOM (6 or 16 slice) Philips MX8000 Dual or Brilliance 16 Mode 2D/3D TX Modes X-ray 68Ge 2D/3D X-ray 3D 3D 3D X-ray 137Cs FOV diam (cm) 50 () 55 () 50 1 () 70 () X-ray 50 1 () 58.5 () 70 X-ray 50 1 () 60 () 70 50 1 () 60 () Patient co-scan port (cm) length (cm) 70 () tapered to 59 () 70 70 () 63 () 160/200 400 160 182 182 190 max bed load (lbs) 400 450 450 430 room size (feet) 14 x 24 14 x 24 15 x 25 15 x 25 14 x 24 Challenges & Problems with / 1. Functional and Anatomical Image Alignment Possible Movement between scans Respiratory Motion (breath hold protocol?) Several organs deviate in volume and position (liver, spleen, ) Calibration of gantry, gantry, and bed 2. Based Attenuation Correction (in my order of least to most important) Truncation Errors from arms down in FOV Artifacts in corresponding regions in image Biases in AC factors from conversion of X-ray energies to energies Incorrect Values in Image Contrast agents, Prostheses, implants Movement between Scans Respiratory Motion Can cause artifacts near dome of liver (in cardiac, causes defects in cardiac values) *Based on published specifications - OUT OF DATE ** sub-system options are changing rapidly ***CPS products sold through I as Reveal and Siemens as Biograph 21/36 22/36 Based AC Problems (Bias in AC Factors) General Rule: If one over attenuation corrects (uses higher ( more dense ) than true correction factors for a region), will get artifactually high values Example: Contrast agents in image, when scaled to energies, are higher than should be leading to artifactual focal hot spots. If one under attenuation corrects (uses lower than true factors for a region), will get artifactually low values Example: Respiratory motion can replace diaphragm in image with lung space in image. Causes photopenic region (banana artifact) along diaphragm in image. Artifact Examples 31/36 32/36 6