Share Email Print
cover

Proceedings Paper

Noninvasive evaluation of metastatic potential of ocular melanoma by analysis of radiofrequency ultrasound data
Author(s): Ronald H. Silverman; Robert Folberg; Frederick L. Lizzi; H. Culver Boldt; Harriet Lloyd; Mark Rondeau; D. Jackson Coleman
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

Purpose: While histopathologic evaluation of ocular melanoma permits assessment of metastatic risk, this is not possible if visual function is to be preserved. In this report, we describe ultrasound methods for non-invasive evaluation of metastatic risk. Methods: Radiofrequency (RF) ultrasound data were acquired prior to enucleation in 117 eyes with untreated malignant melanoma. Extracellular matrix patterns, an indicator of metastatic potential, were identified in histologic sections. We determined calibrated backscatter power spectra, estimates of effective scatterer size and concentration, and the percentage of tumor area comprised of PAS-positive patterns in the anterior, posterior and core regions of the tumors. We compared the spatial correlation of histologic and acoustic properties, performed linear discriminant analysis to define prognostic models, and used receiver operating characteristic (ROC) curves to evaluate models. Results: Statistically significant correlations were found between acoustic parameters and PAS-positive patterns, although their spatial distributions were only weakly related. Stepwise linear discriminant analyses produced models with three to five variables, and ROC areas as high as 0.89. Conclusion: Acoustic spectrum analysis provides information not evident in conventional gray-scale ultrasonograms regarding tissue microstructure. Our results confirm a relationship between spectra and the presence of extracellular matrix patterns associated with metastatic risk.

Paper Details

Date Published: 11 April 2002
PDF: 11 pages
Proc. SPIE 4687, Medical Imaging 2002: Ultrasonic Imaging and Signal Processing, (11 April 2002); doi: 10.1117/12.462150
Show Author Affiliations
Ronald H. Silverman, Weill Medical College of Cornell Univ. (United States)
Robert Folberg, Univ. of Illinois/Chicago (United States)
Frederick L. Lizzi, Riverside Research Institute (United States)
H. Culver Boldt, Univ. of Iowa (United States)
Harriet Lloyd, Weill Medical College of Cornell Univ. (United States)
Mark Rondeau, Weill Medical College of Cornell Univ. (United States)
D. Jackson Coleman, Weill Medical College of Cornell Univ. (United States)


Published in SPIE Proceedings Vol. 4687:
Medical Imaging 2002: Ultrasonic Imaging and Signal Processing
Michael F. Insana; William F. Walker, Editor(s)

© SPIE. Terms of Use
Back to Top