Program Information
BEST IN PHYSICS (IMAGING): X-Ray Diffraction Spectral Imaging for Breast Cancer Assessment
J Spencer1*, J Carter2 , D Nacouzi3 , C Buxton4 , C Leung5 , S McCall6 , J Greenberg7 , A Kapadia8 , (1) Duke University Medical Center, Durham, NC, (2) Duke University, White Sulphur Springs, WV, (3) Duke University, Durham, NC, (4) Duke University, Durham, North Carolina, (5) Duke University Medical Center, Durham, North Carolina, (6) Duke University Medical Center, Durham, North Carolina, (7) Pratt School of Engineering, Duke University, Durham, North Carolina, (8) Duke University Medical Center, Durham, NC
Presentations
TH-AB-708-11 (Thursday, August 3, 2017) 7:30 AM - 9:30 AM Room: 708
Purpose: Previously, we demonstrated a coherent-scatter X-ray diffraction imaging system to differentiate cancerous and healthy breast tissue in a non-destructive manner in near real time. Our previous pilot results utilized a small number of surgically resected breast specimens to confirm the presence of cancerous regions within the specimen. Here we present clinically validated results from 24 unique specimens validated against histology assessment by a pathologist.
Methods: A previously characterized and validated coherent-scatter imaging system was used to scan 173 locations in 24 lumpectomy specimens known a priori to contain a tumor mass or be fully healthy. The specimens were imaged pre- and post-formalin-fixation in 1-4 mm thick slices. Each specimen was vacuum-sealed, scanned in our imaging system, and then prepared for histology assessment. Spectra were produced for each scanned voxel in the specimen and correlated against literature reference spectra to classify the tissue. System fidelity and robustness were assessed by comparing the tissue classification map with H&E stained microslides evaluated by a pathologist.
Results: Weighted-squared-error and cross-correlation classification schemes were used together to classify the scanned tissue voxels. These schemes distinguished cancer and healthy tissue regions within each specimen. The spectra produced by the imaging system exhibited correlation values from 0.77-0.99 and at least 79% accuracy against pathology assessment. The system’s spatial resolution was approximately 1.5 mm at the scanned objects, and the specimen turnover time was generally 30 minutes.
Conclusion: This work shows the first comparisons of fresh and formalin-fixed specimens to pathologist-assessed H&E histology to detect breast cancer through diffraction imaging. The results exhibit the coherent-scatter diffraction imaging system’s ability to distinguish and identify cancerous tissue using principles of x-ray diffraction.
Contact Email: