JAPC - Quantitative Assessment of Tumor Malignancy Using Fractal Analysis

Speaker: Dr. Mauro Tambasco, Department of Oncology, University of Calgary

Topic: "Quantitative Assessment of Tumor Malignancy Using Fractal Analysis"

Time: 2:00 PM, Friday, February 23, 2007

Place: P-148 (refreshments will be served at 1:45 PM in P146A)

Abstract: In 2006, prostate and breast cancer continued to be the leading forms of cancer diagnosed in American men and women, and the third and second leading causes of mortality, respectively. A method that has proven useful for guiding the choice of treatment strategy for patients with these cancers is the assessment of histologic tumor grade (a measure of the degree of abnormality of the cancer cells). The standard methods for grading prostate and breast cancers are partially based on how effectively cells can structure themselves into normal regularly shaped glands. Although these methods work reasonably well, there still exists a significant amount of inter-observer variability. This is not surprising since pathologists use subjective visual criteria for identifying a complex continuum of glandular architectural differences that may exist in histological specimens. Since the level of complexity in glandular structures increases with increasing tumor malignancy, fractal dimension (a measure of complexity) may provide a quantitative index of malignancy that can aid pathologists in grading tumor specimens. The success of this method partially depends on the accuracy with which the fractal dimension of tissue samples can be calculated. An important step in computing the fractal dimension of cancer specimens involves optimizing the identification and segmentation of the histologic structures of interest. We found that a commonly used stain for the pathologic assessment of prostate and breast tissues is not optimal for fractal analysis because it highlights extra-cellular structures making it difficult to isolate and segment the glandular structures of interest. Hence, we searched for and found a more optimal stain for fractal analysis. We also developed an optimal procedure, based on luminance threshold, for capturing and segmenting glandular structures. Our results indicate that our optimized staining and fractal analysis procedure can discriminate between low- and high-grade cancer specimens. Such prognostic value could supplement a pathologist's assessment of tumor grade by providing additional objective information that could reduce inter-observer variability and lead to optimal treatment strategies for patients.

You can view JAPC upcoming talks or the archive.