Advances in the histopathological characterization of breast tissue using combined X‐ray fluorescence and X‐ray diffraction data in a multivariate analysis approach

• Published in: X-ray spectrometry Vol. 48; no. 5; pp. 432 - 437
• Main Authors: Johnston, Eric M., Dao, Erica, Farquharson, Michael J.
• Format: Journal Article
• Language: English
• Published: Bognor Regis Wiley Subscription Services, Inc 01-09-2019
• Subjects: Breast; Dispersion; Energy; Fluorescence; Multivariate analysis; Principal components analysis; Soft tissues; Tissues; X-ray diffraction; Zinc
• ISSN: 0049-8246; 1097-4539
• DOI: 10.1002/xrs.3018

Previous studies have shown that the combination of X‐ray fluorescence and X‐ray diffraction data can be used as a histopathological characterization tool for breast tissue. Recent advances in energy‐dispersive X‐ray fluorescence techniques have allowed for benchtop systems to produce useful results in a reasonable time frame, allowing for clinical implementation to be realized. Using a polarized energy‐dispersive X‐ray fluorescence and energy‐dispersive X‐ray diffraction system optimized for measuring soft tissues, 38 breast tissue samples (19 normal and 19 diseased) were interrogated. The measured elemental concentrations and adipose and fibrous tissue contents were used in a principal component analysis study to determine the variables that produced the most differentiation between the normal and diseased tissues. For each sample, a soft independent modeling of class analogy technique was utilized to create classification models using the K, Fe, and Zn concentration and adipose and fibrous tissue content of all other breast samples. The class model produced from both X‐ray fluorescence and X‐ray diffraction data correctly classified 31 of 38 samples with no false positives or false negatives, showing improvement from solely X‐ray fluorescence models or X‐ray diffraction models alone, and demonstrates the usefulness of such a technique.