Encrypted login | home

Program Information

Prospective Patient Evaluation of Iterative Reconstruction in Radiation Oncology


R Price

R Price1,2*, S Vance2 , R Cattaneo2 , L Schultz2 , M Elshaikh2 , I Chetty2 , C Glide-Hurst2 , (1) Wayne State University School of Medicine, (2) Henry Ford Health Systems, Detroit, MI,

Presentations

SU-F-18C-6 Sunday 4:00PM - 6:00PM Room: 18C

Purpose: This work incorporates iterative reconstruction (IR) into a dose reduction study to characterize image quality metrics, delineation, and dosimetric assessment, with the goal of reducing imaging dose in Radiation Oncology.

Methods: Three-dimensional noise power spectrum (NPS) analysis characterized noise magnitude/texture (120 kVp, 50-200 mAs, IR levels 1-6 yielding noise reduction of 0.89-0.55 compared to filtered backprojection (FBP)). Task-specific Modulation Transfer Functions (MTFtask) were characterized across varied subject contrasts. A prospective dose reduction study (500 to 150 mAs) was conducted for 12 patients (43 inter-fraction CTs) for high-dose rate brachytherapy. Three physicians performed qualitative image assessment between full-dose FBP (FD-FBP, 500 mAs), low-dose FBP (LD-FBP, 150-250 mAs), and low-dose IRL5-6 (LD-IR) scans for image noise, cuff/bladder interface detectability, spatial resolution, texture, and segmentation confidence. Comparisons between LD-FBP and LD-IR were conducted for the following metrics: delineation (bladder and rectum evaluated via overlap indices (OI) and Dice similarity coefficients (DSC)), noise, boundary changes, dose calculation, and digitally reconstructed radiographs (DRRs).

Results: NPS showed ~50% reduction in noise magnitude and ~0.1 1/mm spatial frequency shift with IRL6. The largest MTFtask decrease between FBP and IR was 0.08 A.U. Qualitative patient image evaluation revealed LD-IR was equivalent or slightly worse than FD-FBP, and superior to LD-FBP for all metrics except low contrast interface and texture. The largest CT number discrepancy from FBP occurred at a bone/tissue interface using IRL6 (-1.2 ± 4.9 HU (range: -17.6 – 12.5 HU)). No significant contour differences (OIs and DSCs = 0.85 – 0.95) and dose calculation discrepancy (<0.02%) were observed. DRRs preserved anatomical detail and demonstrated <2% difference in intensity between LD-FBP and LD-IRL6.

Conclusion: While phantom analysis showed slight noise texture differences with IR, patient results revealed that image quality, contouring ability, and dosimetric parameters were not adversely affected, thus support integrating IR into treatment planning.

Funding Support, Disclosures, and Conflict of Interest: Research supported in part by a grant from Philips HealthCare.


Contact Email: