Duct detection and wall spacing estimation in breast tissue

L. Huang, K. D. Donohue, V. Genis, F. Forsberg

Research output: Contribution to journalArticlepeer-review

21 Scopus citations

Abstract

The relationship between duct tissue and several types of malignant disease suggests that methods for characterizing duct structures may be useful tools in ultrasonic tissue characterization. This paper presents performance results from ultrasonic phantom experiments and Monte Carlo simulations for detecting and estimating duct wall spacings on the order of those typically found in breast tissue using methods based on the generalized spectrum (GS) and cepstrum. A performance comparison demonstrates the advantages of each method and examines the effects of various signal processing options, including a special normalization technique for the GS that effectively whitens the data spectrum and reduces interfering spectral influences with little overall performance loss. Experimental results (for both simulation and phantom) indicate that the GS typically achieves detection rates of over 90% (at 10% false alarm rates) over a broad range of SNR values (3-21 dB). The GS detection performance exceeds that of the cepstrum and exhibits more robustness to noise and signal processing parameters. Simulation results with fixed system effects indicate better estimation performance for cepstral-based methods, while experimental phantom results show the GS estimation performance to be the same or better than the cepstral-based method.

Original languageEnglish
Pages (from-to)137-152
Number of pages16
JournalUltrasonic Imaging
Volume22
Issue number3
DOIs
StatePublished - 2000

Keywords

  • Breast tissue
  • Cepstrum
  • Duct wall spacing
  • Generalized spectrum
  • Tissue characterization

ASJC Scopus subject areas

  • Radiological and Ultrasound Technology
  • Radiology Nuclear Medicine and imaging

Fingerprint

Dive into the research topics of 'Duct detection and wall spacing estimation in breast tissue'. Together they form a unique fingerprint.

Cite this