Introduction: Herein, we propose a Systems Biology approach aimed at identifying quantitative morphological parameters useful in discriminating benign from malignant breast microcalcifications at digital mammography.
Materials and methods: The study includes 31 patients in which microcalcifications had been detected during XR mammography and were further confirmed by stereotactic (XR-guided) biopsies. Patients were classified according to the BIRADS (Breast Imaging-Reporting and Data System), along with their parenchyma fractal dimension and biopsy size. A geometrical-topological characterization of microcalcifications was obtained as well.
Results: The 'size of biopsy' was the parameter endowed with the highest discriminant power between malignant and benign lesions thus confirming the reliability of surgeon judgment. The quantitative shape evaluation of both lesions and parenchyma allowed for a promising prediction of the BIRADS score. The area of lesions and parenchyma fractal dimension show a complex distribution for malignant breast calcifications that are consistent with their qualitative morphological pattern. Fractal dimension analysis enables the user to obtain reliable results as proved by its efficiency in the prediction of the morphology of breast cancer.
Conclusion: By reconstructing a phase-space distribution of biophysical parameters, different patterns of aggregation are recognized corresponding to different calcium deposition patterns, while the combination of tissue and microcalcification morphological descriptors provide a statistically significant prediction of tumour grade.
Clinical relevance: The development of an automated morphology evaluation system can help during clinical evaluation while also sketching mechanistic hypotheses of microcalcification generation.
Keywords: Fractal dimension; Mammography; Microcalcification; Quantitative morphology; Radiology; Systems biology.
Copyright © 2017 Elsevier Ltd. All rights reserved.