Ultrasound Physics (Principles, Transducers, Doppler & Artefacts)

• Acoustic impedance. Z = tissue density (ρ) × the speed of sound (c). The proportion of sound reflected at an interface depends on the mismatch in Z: a large mismatch (soft tissue↔bone or ↔gas) reflects almost all the sound (a strong echo, with shadowing beyond), while a small mismatch (soft tissue↔soft tissue) reflects a little and transmits the rest - which is what allows imaging.
• The assumed speed of sound. The machine assumes a fixed soft-tissue speed of about 1540 m/s to convert the echo time-of-flight into depth; real tissues differ (fat slower, bone far faster), a source of some artefacts.
• Attenuation. Sound weakens with depth (absorption/scatter/reflection), roughly 0.5-1 dB/cm per MHz - so higher frequencies attenuate more (the physics behind their poorer penetration); time-gain compensation (TGC) corrects for this depth-dependent loss.
• The two resolutions. Axial resolution (along the beam) improves with higher frequency (shorter pulse); lateral resolution (across the beam) improves with focusing and is best at the focal zone.

• The Doppler equation - angle matters. The frequency shift is proportional to blood velocity × cos(angle) between the beam and flow, so the shift is maximal when the beam is parallel to flow and zero at ninety degrees - for accurate velocity keep the insonation angle low (ideally under 60 degrees).
• Aliasing. If the Doppler shift exceeds the Nyquist limit (half the pulse-repetition frequency), it aliases - the signal wraps around (colour reversal/spectral wrap); correct by raising the velocity scale/PRF, shifting the baseline, using a lower frequency or a smaller angle.
• Colour vs power vs spectral. Colour Doppler shows mean velocity and direction (angle-dependent, aliases); power Doppler shows the signal amplitude - more sensitive to slow flow, non-directional, less angle-dependent and does not alias (best for detecting hyperaemia); spectral/pulsed-wave gives a velocity waveform for quantification.
• Safety and bio-effects. Ultrasound is non-ionising but deposits energy, with thermal (heating - the displayed thermal index, TI) and mechanical (cavitation - the mechanical index, MI) bio-effects; the ALARA principle applies, especially in obstetric/ophthalmic use - diagnostic MSK ultrasound is very safe.

"Operator-dependent" is the stock criticism of ultrasound, and it is repeated so often that candidates assume the modality is second-rate. The accuracy data say otherwise for the commonest musculoskeletal indication.

• Ultrasound and MRI are equivalent for rotator cuff tears. A meta-analysis of 65 studies, all requiring an open or arthroscopic surgical reference standard, found no significant difference in either sensitivity or specificity between MRI and ultrasound for partial-thickness or full-thickness tears (P greater than 0.05). The area under the summary ROC curve was in fact numerically higher for ultrasound (0.889) than for MRI (0.878), though that pairwise difference was not significant.
• MR arthrography beat both. It was significantly more sensitive and specific than either (P less than 0.05), with the highest area under the curve at 0.935 - the trade-off being that it is invasive.
• How to use this in a viva. The right answer is not that ultrasound is inferior, but that for cuff tears it performs comparably to MRI at lower cost, in real time, with dynamic assessment and no ionising radiation, and that MR arthrography is the most accurate if the added invasiveness is justified.
• Where the operator dependence genuinely bites. These are pooled published studies, largely from centres with a declared interest in musculoskeletal ultrasound - so the figures describe the modality in experienced hands. That is precisely what operator dependence means: the technique can match MRI, but only if the person holding the probe knows the anatomy, keeps the beam perpendicular and recognises the artefacts. The physics does not transfer with the machine.
• What is not established here. These figures are for the rotator cuff, and should not be assumed to hold for every tendon, nerve or joint - no comparable pooled accuracy figure exists for most other musculoskeletal sites, and no published study measures how often anisotropy specifically produces a false-positive report.