Program Information
Optimization of a Coded Aperture Coherent Scatter Spectral Imaging System for Breast Cancer Detection
J Carter*, J Spencer , C Buxton , J Fenoli , J Greenberg , A Kapadia , Duke University, Durham, North Carolina
Presentations
MO-RAM-GePD-IT-1 (Monday, July 31, 2017) 9:30 AM - 10:00 AM Room: Imaging ePoster Theater
Purpose: Previously our group demonstrated the ability to classify healthy and malignant breast tissues using coherent scatter spectral imaging. To date spectral data has been acquired using sectioned, lumpectomy specimens obtained from surgical pathology; specimens vary in thickness (up to 3mm). The purpose of this work was to optimize our spectral imaging system to provide accurate and consistent results across specimens of varying thickness while minimizing the power supplied to the X-ray tube (reduce system heating), thereby improving system throughput.
Methods: Healthy and malignant breast tissue specimens were scanned using 36 different acquisition protocols (75-125 kVp, 50-500 mAs) to obtain a coherent-scatter diffraction spectrum for each measurement. A two-step classification algorithm was developed to compare and classify our measured spectra against known, ground truth spectra from literature. The locations of the spectra’s peaks were used to distinguish adipose and normal (50/50 fibroglandular/adipose) tissue from cancer and fibroglandular tissue, followed by a weighted cross correlation method used to distinguish cancer vs. fibroglandular and normal vs. adipose. Classification performance was assessed across all protocols.
Results: The optimal setting was identified as the minimal power supplied to the X-ray tube that resulted in consistent and accurate classification of the spectra in known regions within the palpable tumor. Preliminary results show that the optimal setting is within the 95-105 kVp and 200-320 mAs energy-current range, resulting in a projected increase in efficiency of 300%.
Conclusion: This study successfully demonstrates the optimization of our coherent scatter spectral imaging system based on classification using a two-step weighted cross correlation method. The efficiency improvement allows for enhanced data collection with higher system throughput, thereby allowing shorter scan times and the ability to scan the specimens at a higher resolution.
Contact Email: