Non-invasive classification of breast microcalcifications using x-ray coherent scatter computed tomography

Area of Science:

• Medical Imaging
• Biophysics
• Materials Science

Background:

• Breast calcifications are common indicators of breast cancer.
• Differentiating between type I (weddellite) and type II (hydroxyapatite) calcifications is crucial for diagnosis.
• Current methods may lack specificity in distinguishing calcification types.

Purpose of the Study:

• To investigate energy dispersive x-ray coherent scatter computed tomography (ED-CSCT) for non-invasive differentiation of type I and type II breast calcifications.
• To assess the sensitivity of ED-CSCT to compositional and structural differences in microcalcifications.
• To evaluate the potential of ED-CSCT in improving breast diagnostic accuracy.

Main Methods:

• Simulated a CSCT system with a pencil beam, polychromatic x-ray source, and energy-resolving photon counting detector.
• Reduced multidimensional CSCT data to projection-space distributions of specific components (adipose, glandular tissue, weddellite, hydroxyapatite).
• Employed maximum-likelihood estimation for scatter components and filtered backprojection for image reconstruction; experimentally determined scatter cross-sections.

Main Results:

• The ED-CSCT technique demonstrated good discrimination ability between type I and type II calcifications.
• Classification accuracy was analyzed using receiver operating characteristic methodology.
• Effective differentiation was observed at medium and higher radiation doses.

Conclusions:

• Non-invasive ED-CSCT shows significant potential for enhancing breast cancer diagnosis.
• This technique could lead to a reduction in the number of invasive breast biopsies.
• ED-CSCT offers a promising approach for precise characterization of breast calcifications.