Room: Davidson Ballroom B
Purpose: To quantitatively measure image quality, specifically resolution, of endorectal digital prostate tomosynthesis (endoDPT) and compare to a typical CT scanner.
Methods: endoDPT is a novel imaging method that combines linear parallel tomosynthesis, an external x-ray source, and an endorectal x-ray sensor to produce high-resolution images of the prostate region. The resolution of endoDPT was assessed in three steps. The physical detector modulation transfer function (MTF) was measured using an opaque edge method (80 kV, 12.5 mAs, 100 cm source-image plane distance [SID]). The reconstruction algorithm MTF was determined by computationally modeling and reconstructing the image of a point impulse using shift-and-add (SAA), backprojection (BP), and filtered BP (FBP) (±40 cm source travel, 41 equispatial projections, 100 cm SID). The product of the detector and reconstruction algorithm MTF resulted in the total MTF of endoDPT. The MTF of a CT scanner (composite CT MTF) was measured by imaging the thin wire (0.009� diameter) in a CT performance phantom (helical, 120 kV, auto-mAs, 1.25 mm slice thickness).
Results: The 10% total endoDPT MTF 4.18 mm�¹ for SAA, 9.55 mm�¹ for BP, and 10.27 mm�¹ for FBP. The 10% composite CT MTF was 0.72 mm�¹. endoDPT exhibited significantly higher resolution than CT at each MTF values (p < 0.0001 for all endoDPT reconstructions compared to CT). A previous study showed that this resolution of endoDPT is achievable at ionizing radiation doses that are less than CT for regions outside the field of view and comparable to CT for regions inside the field of view.
Conclusion: endoDPT demonstrated significantly higher image resolution than CT. This indicates endoDPT may have utility for aspects of prostate cancer management that benefit from improved image resolution. These areas may include post-implant imaging of permanent brachytherapy implants or perfusion imaging of small cancerous lesions using iodine contrast enhancement.
Funding Support, Disclosures, and Conflict of Interest: This work was supported by the Louisiana State University Leveraging Innovation for Technology Transfer (LIFT2) Grant.