A multi-scale algorithm for ultrasonic strain reconstruction under moderate compression

Jing Bai, Chuxiong Ding, Yu Fan

Research output: Contribution to journalArticlepeer-review

9 Scopus citations


Since the waveform of an echo is more distorted under larger compression, elastography can be applied in cases where the compression is only a few percent. To reduce errors due to distortion of the echo waveform, a novel algorithm (i.e. the multi-scale correlation algorithm) for strain profile reconstruction is proposed in this article. The basic idea of this method is to simulate human vision: first, we locate the region of the target, and then match the detail within this region and make a comprehensive criterion. Based on this concept, three different approach algorithms are proposed and investigated: (1) the two-step method, (2) the multi-scale correlation algorithm, and (3) the extended multi-scale correlation algorithm. To evaluate the algorithms proposed in this work, a computer simulation model is used and a sequence of simulation experiments are performed. The correlating specificity is used as a criterion for evaluating the performance characteristics of the correlation estimation. The results show that the extended multi-scale correlation algorithm performs significantly better than the other two methods proposed in this article under moderate compression ratios. The extended multi-scale correlation algorithm is applicable to both small and moderate compression ratios. This novel algorithm may provide a useful tool for the clinical application of elastograms.

Original languageEnglish (US)
Pages (from-to)511-519
Number of pages9
Issue number7
StatePublished - Nov 1999


  • Compression
  • Correlation
  • Elastography
  • Multi-scale
  • Simulation
  • Strain reconstruction
  • Ultrasound

ASJC Scopus subject areas

  • Acoustics and Ultrasonics


Dive into the research topics of 'A multi-scale algorithm for ultrasonic strain reconstruction under moderate compression'. Together they form a unique fingerprint.

Cite this