2383 - Initial Examination of Commercial Photon Counting CT for Prostate Planning

Sunday, September 29, 2024

4:45 PM – 6:00 PM ET

Location: Hall C

Screen: 15

Presenter(s)

RB

Ross Berbeco, PhD

Brigham and Women's Hospital/Dana-Farber
Boston, MA, United States

J. OConnell¹, F. Schwartz², T. Morris³, C. Dunning⁴, T. C. Harris¹, D. Ferguson⁵, M. Myronakis⁶, A. Sodickson², and R. I. Berbeco¹; ¹Department of Radiation Oncology, Brigham and Women’s Hospital, Dana Farber Cancer Institute and Harvard Medical School, Boston, MA, ²Brigham and Womens Hospital, and Harvard Medical School, Boston, MA, ³University of Massachusetts Lowell, Lowell, MA, ⁴Siemens Healthineers, Boston, MA, ⁵Department of Radiation Oncology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute and Harvard Medical School, Boston, MA, ⁶Brigham and Womens Hospital, Boston, MA

Purpose/Objective(s): While radiotherapy depends critically on contouring anatomical structures in CT, poor soft tissue contrast and metal artifacts reduce CT contour accuracy. As a remedy to these constraints, photon-counting CT (PCCT) holds promise to provide improved soft tissue contrast and metal artifact reduction (MAR). In this work, we examine these parameters; quantifying PCCT soft tissue contrast, anatomical visibility, and contouring accuracy in a detailed, anthropomorphic pelvis phantom with interchangeable bone, aluminum (Al), and titanium (Ti) femoral head inserts. Materials/

Methods: Ground truths cannot typically be defined for patient bladder and prostate contours, to avoid this constraint, an anthropomorphic pelvis phantom was used in this study, the phantom was scanned using a high-resolution commercial PCCT system (140 kVp, slice thickness 0.2 mm) and an energy-integrating CT (EICT) system (140 kVp, slice thickness 1.25 mm). PCCT images with bilateral and unilateral metal hip inserts were reconstructed using standard and iterative MAR methods, respectively. The bladder and prostate were contoured manually in all images and the contrast to noise ratio (CNR) was measured in regions of interest (ROIs) in the bladder and prostate. PCCT images were binned to have slice thicknesses of 1.2 mm for CNR measurements. For each set of contours, Dice coefficients, Hausdorff distances, and structural similarity index measures (SSIMs) were calculated relative to the no metal ground truth. Fine detail of the images was analysed through delineation of the border between the prostate and rectum in sagittal slices.
Results: PCCT showed similar soft tissue CNR and a qualitative reduction of metal artifacts compared to EICT, however, the reduction in metal artifacts did not correlate to better contours. Dice coefficients, Hausdorff distances and SSIMs for the bladder and prostate (Table 1.) were increased, remained similar, and decreased using MAR for the bilateral Ti, unilateral Ti, and bilateral Al hip inserts, respectively. With the MAR algorithm removing metal artifacts but also blurring soft tissue boundaries. EICT CNRs in the bladder and prostate ROIs (13.8 and 20.3) were slightly higher than PCCT (13.6 and 17.8). PCCT images showed delineation of the boundary between the prostate and the rectum in three sagittal slices where the boundary was unresolved in EICT.
Conclusion: Overall, we find some evidence that PCCT could provide improvement to current workflows in radiation oncology through delineation of soft-tissue boundaries, while PCCT with the vendor supplied MAR shows benefits for contouring anatomy near bilateral Ti hip implants but detriment for contouring anatomy near unilateral Ti or bilateral Al hip implants. Table 1. MAR results.

	Bilat. Al		Unilat. Ti		Bilat. Ti
	PCCT	EICT	PCCT	EICT	PCCT	EICT
Mean Dice - MAR	0.91	-	0.90	-	0.89	-
Mean Dice - no MAR	0.94	0.94	0.92	0.93	0.69	0.73
SSIM - MAR	0.69	-	0.64	-	0.58	-
SSIM - no MAR	0.71	0.67	0.60	0.64	0.30	0.39