By Topic

Realization of High Octave Decomposition for Breast Cancer Feature Extraction on Ultrasound Images

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$33 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

5 Author(s)
Hsieh-Wei Lee ; Department of Computer and Communication Engineering, National Kaohsiung First University of Science and Technology, Taiwan ; King-Chu Hung ; Bin-Da Liu ; Sheau-Fang Lei
more authors

An infiltrative nature on ultrasound images is a significant feature of malignant breast lesion. Characterizing the infiltrative nature with highly efficacious and computationally inexpensive features is crucial for computer-aided diagnosis. The local variance can be characterized by a few high octave energies in the 1-D discrete periodized wavelet transform (DPWT). For the realization of high octave energy extraction, a non-recursive DPWT called 1-D RRO-NRDPWT and a segment accumulation algorithm (SAA) are applied. The 1-D RRO-NRDPWT is used to solve the word-length-growth (WLG) problem existing in high octave decomposition. The SAA is used to overcome the filter-tap-growth (FTG) effect existing in the 1-D NRDPWT. Incorporating these two strategies, a SAA-based VLSI architecture is presented for high octave decomposition. The influence of the finite precision process on feature efficacy is also analyzed for hardware efficiency improvement. Hardware simulation shows that with 7-bit filter coefficient representation, the core size of the octave energy feature (D6E5) extractor is about 335.295*335.295 μm2 where the wavelet transformation will take about 54.87% and 2.875 mW.

Published in:

IEEE Transactions on Circuits and Systems I: Regular Papers  (Volume:58 ,  Issue: 6 )