Paper: | FR-PM-PS3.12 |
Session: | Image Guided Diagnosis, Surgery and Therapy |
Time: | Friday, April 7, 13:30 - 14:50 |
Presentation: |
Poster
|
Title: |
Hidden Seed Reconstruction from C-ARM Images in Brachytherapy |
Authors: |
Ryan Kon; The Johns Hopkins University | | |
| Ameet Jain; The Johns Hopkins University | | |
| Gabor Fichtinger; The Johns Hopkins University | | |
Abstract: |
There has been a pressing clinical need for adaptive intra-operative dosimetry in the delivery of prostate brachytherapy implants. The missing prerequisite is the robust matching of the seeds across multiple C-arm images. This is further aggravated since seeds are invariably hidden in each image. We present a solution to recover these hidden seeds in this paper. A network flow formulation of the problem is proposed, where the desired solution is obtained (in polynomial time) by computing the flow with minimum cost. Phantom experiments show that using four X-ray images, on an average 99.8% of the seeds are recovered correctly, while simulations indicate that our algorithm is robust to segmentation errors of up to 1 mm and hidden seed rate of at least 8%. The results show strong feasibility and clinical data collection is currently underway. |