Engineering
Coherence-Selected Delays: Refocus Without a Local Sound-Speed Map
A short engineering review of coherence-selected ultrasound focusing delays, distinguishing average-speed estimation from local tissue-property mapping and proposing an OEM evaluation workflow.
By ALEVA | California, USA
Can ultrasound software correct focus without first reconstructing a local tissue sound-speed map?
The method
Vrålstad and colleagues describe a coherence-based approach: reconstruct candidate coherence images at different assumed sound speeds, then estimate a spatially varying average speed for focusing. This avoids the separate local-map inversion. The authors identify the peer-reviewed publication as IEEE TMI, July 2026; its openly readable methodological precursor is a December 2024 preprint. These are not two independent studies.
The trade-off
The preprint uses thresholding, smoothing and rank filtering to address weak coherence and misleading selections near strong scatterers. Its propagation approximation neglects refraction. It also reports examples where correction was not preferred. An average-speed focusing map should not be presented as a validated local tissue-property map.
Our proposed OEM evaluation
Treat this as a software candidate, not an automatic upgrade. Replay one controlled channel-data set through the reference and candidate, retaining a record of acquisition settings and processing versions. Compare point-target location, width and contrast across the field—not just the sharpest selected crop.
Next, test low-coherence regions and strong reflectors deliberately. Record when correction is disabled or rejected. Measure full processing latency and memory use on the intended hardware before promising a frame rate.
Bottom line: evaluate the estimation rule, failure policy and reconstruction together. This is ALEVA's proposed engineering checklist, not measured ALEVA performance, a clinical recommendation or a commercial implementation claim.
Sources
Vrålstad AE et al. Coherence-Based Sound Speed Aberration Correction—With Clinical Validation in Fetal Ultrasound. IEEE Transactions on Medical Imaging. July 2026;45(7):4047–4057. DOI: 10.1109/TMI.2026.3691415. Publication metadata and abstract checked against the author's research page; publisher full text was not accessible in this review.
Open methodological precursor: arXiv:2411.16551v2, 6 December 2024 (preprint, not the peer-reviewed 2026 version). Methods and limitations above are explicitly attributed to this version. No figures or clinical images reproduced.