Multi-cell GPS: a multi-cell tracking algorithm using positron emission tomography.
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Kim, Yeseul | - |
dc.contributor.author | Lee, Hyun Woo | - |
dc.contributor.author | Hong, Seongje | - |
dc.contributor.author | Lee, Hoyeon | - |
dc.contributor.author | Jung, Kyung Oh | - |
dc.contributor.author | Sung, Wonmo | - |
dc.date.accessioned | 2024-03-19T01:00:21Z | - |
dc.date.available | 2024-03-19T01:00:21Z | - |
dc.date.issued | 2023-06 | - |
dc.identifier.issn | 0161-5505 | - |
dc.identifier.issn | 1535-5667 | - |
dc.identifier.uri | https://scholarworks.bwise.kr/cau/handle/2019.sw.cau/72910 | - |
dc.description.abstract | Introduction: Although in vivo molecular imaging can measure the average movement of injected cells throughout the body, the accuracy of the cell distribution is hindered by the non-specific accumulation of the contrast agent. The recently proposed single-cell tracking method, which concentrates radioisotopes in nanoparticles, is less accurate in the presence of multiple radiolabeled cells. In order to track multiple cells from the injection site to the initial site of the arrest, we proposed an extended algorithm, called "multi-cell GPS" to investigate cell-to-cell interactions in cellular therapies.Methods: Multi-cell GPS algorithm reconstructs the trajectory of a cell through three steps: 1) Line-Of-Response (LOR) sorting; 2) optimizing the points M (a surrogate for an actual annihilation point); and 3) applying 1-dimensional B-Spline interpolation to the set of M, as described in Figure 1. We utilized Gadolinium-68 as the beta emitter with different activities, we could sort out LORs came from low-activity static sources. Tracking accuracy computed as the average Euclidian distance between the ground truth track and the reconstructed track, a quantitative performance metric of algorithm. Using TOPAS Monte Carlo simulation, we demonstrated a PET device and a cylindrical water phantom simulating small animals with a diameter of 2cm. We used moving and static sources positioned inside the phantom to validate our algorithm by three experiments, as summarized in Figure 2. The primary objective of our algorithm is to reconstruct the moving source’s trajectory even with the existence of static source. In the first single-moving source experiment, we examined how much radioactivity was required to guarantee an accurate enough reconstruction of a single-moving source. In the following moving and static source experiments, we conservatively validated our proposed algorithm by assigning a high enough radioactivity to the static source obtained in the first experiment. Among various experimental parameters, we adjusted the radioactivity differences (100~900Bq) and distance (1~5cm) between two sources, while the velocity of moving source set as 0.04cm/s and the diameter of water phantom fixed as 2cm. In the last two moving sources experiments, we changed the radioactivity differences (300~700Bq) and distance (2~3cm) for one clockwise and the other counter clock-wise moving source. We tuned model’s hyperparameters such as lambda and knots using grid search method.Results: In the first single moving source experiment, the tracking accuracy decreased exponentially as the activity of the sources decreased due to the lower number of LOR generated as shown in Figure 3. When we normalize all the tracking accuracy with respect to the highest tracking accuracy obtained with 10Bq, the tracking accuracy started to improve by over 75% with activities above 100Bq. In the following two sources experiment, we computed heatmap of tracking accuracy from the original cell GPS, multi-cell GPS, and percentage differences between two algorithms (refer to Figure 4). Multi-cell GPS, outperformed the original single cell tracking algorithm. The maximum tracking accuracy difference % = 56.93 % and the average % = 25.59 % (95% CI [28.98 – 22.21]). As shown in Figure 5, in the two moving source experiments' constraint search results also demonstrated that multi-cell GPS works better than the original cell-GPS algorithm. The maximum tracking accuracy difference % = 70.84 % and the average was 55.34 % (range: [70.84 – 34.03]).Conclusions: In this study, we propose a multi-cell tracking algorithm, multi-cell GPS, a trajectory reconstruction algorithm for use in the biomedical field. We confirmed that LOR sorting step reduces LORs unnecessary for trajectory reconstruction and noise signals, and trade of between activity difference and distance difference. In-phantom validation using small animal PET-CT device with helical track acryl phantoms is currently under investigation. | - |
dc.language | 영어 | - |
dc.language.iso | ENG | - |
dc.publisher | SOC NUCLEAR MEDICINE INC | - |
dc.title | Multi-cell GPS: a multi-cell tracking algorithm using positron emission tomography. | - |
dc.type | Article | - |
dc.identifier.bibliographicCitation | JOURNAL OF NUCLEAR MEDICINE, v.64, no.S1 | - |
dc.description.isOpenAccess | N | - |
dc.identifier.wosid | 001109210203084 | - |
dc.citation.number | S1 | - |
dc.citation.title | JOURNAL OF NUCLEAR MEDICINE | - |
dc.citation.volume | 64 | - |
dc.identifier.url | https://jnm.snmjournals.org/content/64/supplement_1/P1523 | - |
dc.type.docType | Meeting Abstract | - |
dc.publisher.location | 미국 | - |
dc.relation.journalResearchArea | Radiology, Nuclear Medicine & Medical Imaging | - |
dc.relation.journalWebOfScienceCategory | Radiology, Nuclear Medicine & Medical Imaging | - |
dc.description.journalRegisteredClass | scie | - |
dc.description.journalRegisteredClass | scopus | - |
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.
84, Heukseok-ro, Dongjak-gu, Seoul, Republic of Korea (06974)02-820-6194
COPYRIGHT 2019 Chung-Ang University All Rights Reserved.
Certain data included herein are derived from the © Web of Science of Clarivate Analytics. All rights reserved.
You may not copy or re-distribute this material in whole or in part without the prior written consent of Clarivate Analytics.