Breast tumor classification of ultrasound images using wavelet-based channel energy and imageJ

Hsieh Wei Lee, Bin Da Liu, King Chu Hung, Sheau Feng Lei, Po Chin Wang, Tsung Lung Yang

Research output: Contribution to journalArticlepeer-review

16 Citations (Scopus)

Abstract

The infiltrative nature of lesions is a significant feature that implies a malignant breast lesion in ultrasound images. Characterizing the infiltrative nature of lesions with computationally inexpensive and highly efficacious features is crucial for the realization of a computer-aided diagnosis system. In this study, the infiltrative nature of lesions is regarded as an energy that produces irregular and considerably local variances in a 1-D signal. The local variances can be characterized by a few high octave energies (i.e., the channel energies close to low frequency bands) in 1-D discrete periodized wavelet transform (DPWT). To reduce computation cost, high octave decomposition is performed by a reversible round-off 1-D nonrecursive DPWT (1-D RRO-NRDPWT). A test dataset of breast sonograms with the lesion contour delineated by an experienced physician and three datasets of breast sonograms with the lesion contour delineated by a Java-based image processing program, ImageJ, are built for feature efficacy evaluation. Evaluation with the receiver operating characteristic (ROC) parameters, the area under ROC curve (Az), accuracy (Ac), sensitivity (Se), specificity (Sp), and positive (ppv) and negative predictive values (npv), shows that the proposed feature has an individual performance of (Az, Ac, Se, Sp, ppv, npv) = (0.991, 0.951, 0.985, 0.933, 0.973, 0.992) and (0.934, 0.844, 0.933, 0.795, 0.714, 0.956) for manual and ImageJ-generated datasets, respectively. The performance differences in the three ImageJ-generated datasets derived by variant setting parameters are not significant. Experimental results also reveal that the proposed feature is suitable for combination with some morphometric parameters for performance improvement.

Original languageEnglish
Pages (from-to)81-93
Number of pages13
JournalIEEE Journal on Selected Topics in Signal Processing
Volume3
Issue number1
DOIs
Publication statusPublished - 2009

All Science Journal Classification (ASJC) codes

  • Signal Processing
  • Electrical and Electronic Engineering

Fingerprint Dive into the research topics of 'Breast tumor classification of ultrasound images using wavelet-based channel energy and imageJ'. Together they form a unique fingerprint.

Cite this