Development of Methods for Improved Breast Imaging Using Ultrasound.
Abstract
A novel framework was introduced for the deconvolution of B-scan images. A model for the rf image was developed first, and then imaging distortions were reconstructed using higher-order statistics of the measured image lines. Based on the estimated distortions deconvolution of the corresponding image was performed, which for the case of images of tissue mimicking phantom as well as human tissue images, led to significant resolution improvement. In the past, estimation of distortions has been carried out using exclusively second order statistics (autocorrelation). Autocorrelation, however, can recover only the minimum phase equivalent of the true distortions, because it is blind to phase. Imaging distortions were also estimates using second order statistics of the image lines. It was demonstrated that although these estimates did lead to resolution improvement, the amount of improvement was far less significant than the one obtained with the higher-order statistics based estimates.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 26, 1995
- Accession Number
- ADA299280
Entities
People
- Athina P. Petropulu
- John M. Reid
Organizations
- Drexel University