71. Contrast-enhanced spectral mammography. Analysis of region of interest based on gray levels: a preliminary multi-center study in Puglia (Italy). (December 2018)

• Record Type: Journal Article
• Authors: Losurdo, L.; Tangaro, S.; Didonna, V.; Basile, T.M.A.; Bellotti, R.; Carbonara, R.; Cinelli, F.; Fanizzi, A.; Fiorentino, F.; Gorgoglione, A.; Moschetta, M.; Tamborra, P.; Forgia, D. La; Massafra, R.
• Is Part Of: Physica medica. Volume 56(2018)Supplement 2
• Page Start: 107
• Publication Date: 2018-12
• Subjects: Medical physics -- Periodicals; Biophysics -- Periodicals; Biophysics -- Periodicals; Imagerie médicale -- Périodiques; Radiothérapie -- Périodiques; Rayons X -- Sécurité -- Mesures -- Périodiques; Physique -- Périodiques; Médecine -- Périodiques; 610.153
• DOI: 10.1016/j.ejmp.2018.04.081
• Languages: English
• ISSNs: 1120-1797

Abstract

Abstract : Purpose: Contrast-Enhanced Spectral Mammography (CESM) allows the acquisition of low- and high-energy digital mammograms with Dual-Energy technique after a single injection of iodinated contrast medium (CM) obtaining the production of recombined mammographic images[1, 2] . The aim of this paper is to characterize ROI and measure the gray-level parenchymal background in order to improve the lesion detection in CESM. Methods: We have selected 42 patients positive to the methodic for the presence of at least one findings after histological examination.We have evaluated 65 lesions (52 malignant and 13 benign) from 0.5 to 13.5 cm. Primary and secondary (if present) lesions have been identified and their diameters have been measured in axial projection. Regions including the parenchymal background and the non-contrasted background were also marked in both low-energy and recombined images.For each region the statistical parameters of gray levels were recorded by CESM instrumentation and have been elaborated by multivariate statistics and machine learning techniques in order to characterize each region and define a background intensity indicator, not yet defined objectively. Results: The analysis on our available dataset seems to underline significant differences between ROIs with lesions and regions marked on the parenchymal background with respect to the coefficient of variation, i.e. standard deviation of the gray levels of the marked area compared to the mean value of the same) in recombined images.We have found an increase of the coefficient of ROIs greater than 20% with respect to the coefficient of the regions marked on parenchymal background in 57 lesions. Conclusions: As preliminary results, an indicator to detect the parenchymal background can be identified as the coefficient of variation. Data collection is currently ongoing because a more massive data sample will be useful to increase the sensitivity in the lesion detection by CESM device, even in patients with a medium-high value of background.