image resolution and artefacts

what principles are imaging modalities based on

variation in impedance in the tissue causes acoustic waves to be reflected back to the transducer

• needs to be enough impedance for the beam to be reflected and not just pass through the tissue

• but not too much as to stop the wave from reaching deeper structures

why is sound speed estimated in US machines and what is it a failure of?

• to calculate the depths of different structures the sound speed need to be known

• in the body sound speed is not constant

• US machines average the sound speed → 1540m/s

  • close enough to give estimation of structure depths

  • if actual speed is too different → depth is wrong, location is wrong, artifacting → failure in condition 1

ex:

fat → slower

muscle → faster

what are the 9 ideals conditions for B mode imaging

1. sound speed and absorption is known and constant

born approximation stuff:

2. scattering is much weaker than incident wave

3. wave is only scattered once

array geometry:

4. elevation focusing restricts beam to a thin imaging plane

5. beamforming restricts beam to a thin line within the imaging plane

pulse:

6. probe can emit and detect all frequencies (infinite bandwidth)

7. emitted and received signals are individual pulses with an infinitesimally short pulse duration

misc:

8. there is no scattering from beyond the imaging depth

9. data is noise free

what is spatial resolution

the ability to distinguish between 2 features located close to each other

there is lateral, axial and elevation resolutions

what is lateral resolution

what affects it

what is it a failure of?

how is it improved?

→ ability to distinguish between two objects at the same depth

affected by → beam width $=\frac{1.41\lambda F}{aperture~diameter}$

failure of condition 5 → beam is not a perfectly thin line

• even if you move the center of the A line the sides of the beam still hits adjacent structures → lateral artefacting

improved using:

• multi-zone focusing

• receive beamforming

what is axial resolution?

what affects it

what is it a failure of

→ ability to distinguish between objects at different depths

affected by pulse duration → depths information is provided by their separation in time

separation>\frac{FWHM}{2}

failure in condition 6 and 7 → pulse not infinitesimally small nor does it have infinite bandwidth (not instant)

axial resolution equation

$$axial~resolution=\frac{c_o\tau}{2}$$

elevation resolution

what affects it

what is it a failure of

affected by the element length b → usually 20 - $30\lambda$ which is a compromise between elevation resolution and depth of field

• larger length → narrower focus → better resolution

• however → beam diverges quickly → poorer resolution after focus → poorer depth of field

failure of condition 4

what artefacting does elevation resolution lead to

out of plane artefacting → reflections from objects outside you slice → speckle

what does failure in condition 3 produce

waves can be scattered multiple times → multiple strong reflections between specular scatterers

• b-lines

• comet-tails

mirror-image artifact:

• beam hit large specular surface causing beam to reflect

• reflected beam hits another surface and bounces back to the specular surface

• produces a mirror image across the first specular boundary

failure in condition 1

enhancement artefact:

• Enhancement beyond a fluid-filled (low attenuation) region

  • US passes through a region of low attenuation and is absorbed less than other US waves

  • The time gain compensation in the area after it then enhances it above surrounding levels (it was not needed as it was less attenuated)

• Time-gain-compensation is wrong as attenuation is not uniform

the reverse can occur producing shadow artefacts:

• strong absorber prevents US beam from reaching area behind it

• time gain compensation is insufficient leaving a dark shadowy area

refraction artefacts:

• different medium densities causes sound speed to change → beam bends

• causes objects to appear in the wrong place or missed

failur in condition 5, how is it minimes?

grating/side-lobe artefacts:

• echoes received from the side/grating lobe are interpreted as an object originating from the main beam field of view

reduced by changing transducer design and apodisation

failure in condition 8

scattering from beyond the imaging plane → range ambiguity artefacts

• scattering from beyond the image appears at shallower depth inside the image

• if you change the imaging depth the artefact will move

failure in 4 and 7

→ instant pulse + thin slice = speckle (texture in US image)

• medium contains many diffusive scatterer (<$\lambda$ ) that are randomly distributed

• may have different impedances

→ scattered wave reach the transducer and constructively + destructively interfere with each other

• does not give any position or structural information as the speckle is dependent on transducer and processing

• but it is deterministic (always appears the same)

how do we reduce speckling

using image compounding:

• average several ultrasound images taken at different conditions → non-linear processing

• this lead to images with uncorrelated speckle patterns

• averaging removes this varying component

what is the drawback of compounding?

reduced frame rate

what are types of non-linear processing

1. persistence (temporal compounding): several subsequent image frames are averaged

2. spatial compounding: images taken from different scan directions are averaged

3. frequency compounding: images taken at different frequencies are taken

choosing transducer frequencies

transducers only have a finite bandwidth on transmit or receive (failure of 6) but this can be chosen depending on:

• depth → lower frequencies can travel deeper as they are absorbed less

• resolution → higher frequencies have better spatial resolution

  • higher frequencies = shorter wavelength → narrower focus width + shorter pulse length (better axial resolution)

what determines frame rate?

pulse repetition frequency: time required for pulse to travel to the maximum depth $(L_{max})$ and then return to the transducer

$PRF=\frac{1}{round~trip~time}=\frac{C_0}{2L_{max}}$ remember that it is a frequency not time period

$frame~rate=\frac{PRF}{n}$ where n is the number of scan lines

what reduces frame rate

1. compounding

2. b-mode imaging

3. multi-zone transmitting