TTK 4165 Signalbehandlingsteknikker i medisinsk bildediagnostikk Signal Processing in Medical Imaging Faglærer: Hans Torp Institutt for sirkulasjon og bildediagnostikk Story: Myocard. Deseases important. Noninvasive methods desirable (Viability, perfusion, mechanical stress/strain) Ultrasound principles (one slide). Increasing aperture Cardiac imaging from bathtube to realtime) Stress echo Poor image quality need for contrast Perfusion imaging - second harmonic Octave improves image quality Doppler blood flow important for valve and diastolic function Color flow give detailed filling info, difficult to relate to regional function Tissue Doppler more directly related to muscular movement Integrated effect from apex TVI simultaneous vel. Measurement over 2D plane Displacement imaging Strain rate and strain imaging High frame-rate methods Q-stress Hans Torp NTNU, Norway

Introduksjon Litt ultralydfysikk og historisk tilbakeblikk Ultralyd avbildning Ultralyd Doppler for måling/avbildning av hastighet Oversikt over faget TTK4165 Story; M-mode to real-time 2D Next step 3D ??? No, No, ……… Stress echo Poor image quality need for contrast Perfusion imaging - second harmonic Octave improves image quality Hans Torp NTNU, Norway

lecture overview Physical principles of ultrasound Ultrasound imaging Ultrasound Doppler and flow imaging Overview TTK4165 Story; M-mode to real-time 2D Next step 3D ??? No, No, ……… Stress echo Poor image quality need for contrast Perfusion imaging - second harmonic Octave improves image quality Hans Torp NTNU, Norway

Sound field depends on source size and wavelength

Ultrasonic M-Mode (Motion Mode) Echoes from tissue structures are received and displayed First Cardiac trials by Edler and Hertz in 1953 Hans Torp NTNU, Norway

Real-time Ultrasound B-mode 1974 Vis film Hans Torp NTNU, Norway N. Bom & al. “Multiscan EchoCardiograph” Ultrasound in Medicine aug. 74

Doppler blood flow meter Pedof 1976 Blood velocity Mitral inflow Normal relaxation Delayed relaxation

Fourier transform - measure bloodflow Gaussian Random process - ultrasound signal Analog computer diff. equation solver - model of the cardiovascular system Bernouli equation - from blood velocity to pressure

Ultrasound probe Focusing Steering and Focusing 50-200 elements Strålefokusering kan også foregå elektronisk. Ved å sende på de forskjellige elementene til litt forskjellig tid kan man få bidragene fra elementene til å følge en virtuell sirkelbue. Send først på ytterste, så på innerste. Man kan velge dybden på fokus ved å sette opp bestemte forsinkelser. Elektronisk strålestyring som vist tidligere kankombineres med fokusering. På mottaging kan man tilsvarende motta på alle elementer samtidig, men forsinkesignalet på de midterste elementene slik at man får en virtuell krumning også på mottaking. Denne kan forandres under mottaking som vi skal se: 50-200 elements

Received Echoes from close objects 2 1 Objects probe elements

Digital Beam Former Data per scanline: 2*10.000*128 = 2.5 Mb A / D 134 #Channels: 128 # samples per channel: 10.000 Data per scanline: 2*10.000*128 = 2.5 Mb Data per image: 2.5*100 = 250 Mb Data per second: 40* 250 Mb = 10 Gb

1996 System Five 128 channels Elektronic scanning Mekanisk scanning 2000 Vivid 7 128 kanaler Elektronisk scanning 1986 CFM 700 5 channels Mekanisk scanning

Real-time 2D B-mode Wall motion assessment ))) ))) Hans Torp NTNU, Norway

Ultrasound Probes Phased array Linear array Curve-linear array Small footprint 90 deg. sector format Linear array High resolution Limited width Curve-linear array Large image width Large near field Hans Torp NTNU, Norway Ultrasound Probes

Ultrasound imaging can be applied to almost all human organs Heart, 4 chamber view Ultrasound imaging: Measure dimensions, areas, volumes Study anatomical details Assessment of muscle contraction Heart-valve function Kidney Liver Twin fetus Fetus 3 ½ mnd

Image resolution F-number f# = F/D Wavelength: L ”Dot-size”: f#  L Probe-diameter D F: Focal depth L D aperture F-number f# = F/D Wavelength: L ”Dot-size”: f#  L F: Focal depth Camera example L = 0.9 e-3 mm f# = 5.6 Resolution: 0.005 mm ~ 50000 dpi Infrared camera gives lower resolution Ultrasound example L = 0.5 mm (3 MHz) f# = 8cm/2cm= 4 Resolution: 2 mm ~ 125 dpi Larger probe -> improved resolution Higher frequency -> improved resolution

Embryo 7 weeks. Ca 13 mm length Computer-simulated ultrasound image Higher frequency -> better resolution

3D Transvaginal ultrasound The Lancet: In-vivo three-dimensional ultrasound reconstructions in the embryonic and early fetal period Harm-Gerd Blaas 1, Sturla H. Eik-Nes 1, Sevald Berg 2, Hans Torp 2;

Limb development in Norway 20th century 18 weeks 12 weeks

Color Doppler velocity imaging PW Doppler: Velocity from one point Color M-mode: Velocities along a line Color flow imaging: Velocities in the whole image

Tissue Velocity Imaging Systole Early relax. Atrial systole Curved M-mode Moving upward Moving downward Hans Torp NTNU, Norway

Wall motion quantification Strain rate Tissue velocity Strain rate Systole Early relax. Atrial systole Curved M-mode v1 SR L v2 Shortening No change Elongation Adapted from J-U. Voigt and A. Heimdal

Real-time 3D imaging 2D matrix array 32-192 elements in a 1D array 32*32 ... 96*96 elements in a 2D array 1000 - 10000 elements Cable Electronics Beamformer 50 x 1 elements 50 x 50 =2500 elements

Sanntid 3D Azimuth Elevation

4D Volume Imaging Increased elevation width Standard setup No ECG gating Volume rendering / orthogonal slicing Volume size: ~20 x 80º / ~35 x 45º Volume rate: 17-25

4D Color Imaging (ECG Gated) Gated from 7 heart beats High frame rates (17-35 frames / second) High Color sensitivity High Color resolution

TTK4165 Overview Pulse Echo principle Ultrasound beamforming General imaging system Ultrasound imaging system Doppler – blood velocity measurement and imaging Patient safety issues Ultrasound contrast imaging 3D imaging