Matching and reconstruction of brachytherapy seeds using the Hungarian algorithm (MARSHAL)

Abstract

Deterministic loading of single atoms onto arbitrary two-dimensional lattice points has recently been demonstrated, where by dynamically controlling the optical-dipole potential, atoms from a probabilistically loaded lattice were relocated to target lattice points to form a zero-entropy atomic lattice. In this atom rearrangement, how to pair atoms with the target sites is a combinatorial optimization problem: brute-force methods search all possible combinations so the process is slow, while heuristic methods are time-efficient but optimal solutions are not guaranteed. Here, we use the Hun-garian matching algorithm as a fast and rigorous alternative to this problem of defect-free atomic lattice formation. Our approach utilizes an optimization cost function that restricts collision-free guiding paths so that atom loss due to collision is minimized during rearrangement. Experiments were performed with cold rubidium atoms that were trapped and guided with holographically controlled optical-dipole traps. The result of atom relocation from a partially filled 7-by-7 lattice to a 3-by-3 target lattice strongly agrees with the theoretical analysis: using the Hungarian algorithm minimizes the collisional and trespassing paths and results in improved performance, with over 50% higher success probability than the heuristic shortest-move method.

Tomosynthesis-based radioactive seed localization in prostate brachytherapy using modified distance map images

“…Since the exact locations of the seeds were unknown, the correspondence between the projection of the estimated seeds and the actual seeds in the images was visually assessed. The estimated seed locations were also compared with those computed by an existing software, MARSHAL [7] and the differences were computed. In Figure 2 are shown an example patient image and a magnified image onto which the centroids of the estimated seeds are projected.…”

Prostate Brachytherapy Seed Reconstruction Using C-Arm Rotation Measurement and Motion Compensation

“…1(a)). Three dimensional reconstruction of the seeds has several benefits such as intraoperative dosimetry modifications and quantitative assessment, and has been proposed previously [1,2,3].…”

“…Seed reconstruction entails solving a seed matching problem -assigning the shadows of a seed in different images to each other-which has been solved using simulated annealing [1], Hungarian algorithm [2] and linear programing [3]. rotation of the C-arm around the primary axis (PA) and the world coordinate system, and (c) the scaling problem.…”

Towards Intra-operative Prostate Brachytherapy Dosimetry Based on Partial Seed Localization in Ultrasound and Registration to C-arm Fluoroscopy

“…However, accurate seed localization based on ultrasound has proven to be a very difficult task due to clutter from other highly reflecting objects such as calcifications resulting in false positive appearances, seed specularity and shadowing, and limited field of view. Even when hand-segmented, up to 25% of the seeds remain hidden in ultrasound images [2]. Therefore, C-arm fluoroscopy is commonly used for visual assessment of the implanted seeds.…”