High-resolution imaging for characterizing microcalcification detection performance in breast CT.

Area of Science:

• Medical Imaging
• Radiology
• Image Analysis

Background:

• Accurate characterization of microcalcifications is crucial for breast cancer detection in mammography and breast CT (bCT).
• Evaluating the detection performance of bCT systems requires reliable ground-truth data on microcalcification size and shape.

Purpose of the Study:

• To establish high-resolution micro-computed tomography (micro-CT) as a tool for generating ground-truth data of calcium grains.
• To evaluate the detection performance of a prototype bCT system using micro-CT derived calcium grain properties.
• To assess the utility of micro-CT data for predicting bCT detectability.

Main Methods:

• Calcium carbonate grains were suspended in agar to simulate microcalcifications within a fibroglandular background.
• Ground-truth imaging of calcium grains was performed using a commercial micro-CT scanner to determine size and shape.
• A prototype bCT system was used to image a breast phantom containing the simulated microcalcifications at varying radiation doses (3 and 6 mGy MGD).
• Non-prewhitening detectability was assessed, and micro-CT derived signal profiles were used with bCT system characterization (MTF, NPS) to predict detectability.

Main Results:

• Estimated detectability of calcium grains on the bCT system ranged from 2.5 to 15.3, with many grains meeting the Rose criterion for visibility.
• Micro-CT image segmentation accurately determined calcium grain boundaries and sizes.
• A regression model showed that micro-CT derived grain size and position significantly impacted bCT detectability.
• Detectability predictions using micro-CT data correlated well with direct bCT measurements.

Conclusions:

• Micro-CT derived ground-truth parameters provide valuable characterizations of detectability for bCT systems.
• Signal profiles from micro-CT imaging, combined with system properties, can effectively evaluate bCT system performance.
• This approach aids in assessing the performance of existing or not-yet-available bCT systems or components.