Detection of Masses in Mammography Through Redundant Expansions of Scale

Abstract

We show that dyadic scales may not be sufficient for the detection of masses in mammograms: a lesion may be too blurred on one scale, and then too fragmented at the next. In this paper, we report on the preliminary evidence of our study using a continuous wavelet transform in two dimensions with arbitrary positioning of a wavelet's center frequency channel tuned to the mass detection problem. Our goal is to detect masses in dense mammograms whose diameter is smaller than 1 cm. The aim is to be able to find the scale where the mass is best represented in terms of analysis.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Oct 25, 2001
Accession Number
ADA410558

Entities

People

  • A. F. Laine
  • F. Durupt
  • M. A. Birgen
  • S. Bauduin
  • V. Laffont

Organizations

  • Duke University Hospital

Tags

Communities of Interest

  • Biomedical

DTIC Thesaurus Topics

  • Abstracts
  • Algorithms
  • Biomedical Engineering
  • Biomedical Research
  • Coefficients
  • Decomposition
  • Detection
  • Engineering
  • Frequency
  • Gaussian Noise
  • Image Processing
  • Military Research
  • New York
  • Noise
  • Personal Information Managers
  • Wavelet Transforms

Fields of Study

  • Physics

Readers

  • Image Processing and Computer Vision.
  • Oncology and Biomarker-Based Cancer Detection.