TH‐E‐BRCD‐02: Automated Treatment Planning Using a Database of Prior Patient Treatment Plans

Todd McNutt, B. wu, J. Moore, S. Petit, M. Kazhdan, Russell H Taylor

Research output: Contribution to journalArticle

Abstract

Intensity modulated RT (IMRT) is used to deliver highly conformal radiation treatments. Its broad adoption has introduced considerable inter‐patient and inter‐institutional variability in plan quality. As part of our Oncospace program, we have built a system that uses patient geometric information and a database of previously treated patients for IMRT treatment plan quality assessment and automated planning. The system uses the relationship between organs at risk (OARs) and target volumes to predict the achievable dose distributions for a given patient from a database of prior patients. It then uses this achievable dose distribution to guide the planning process for the new patient. This system also employs a feedback model to insure the database improves over time through continual improvement in plan quality. The basic premise is the overlap volume histogram (OVH), a shape descriptor we introduced to describe the complex spatial relationship between an OAR and the target volume. The OVH describes the volume of overlap between the OAR and an expanded/contracted target as a function of the expansion/contraction distance. It is determined by expanding/contracting the target volume and calculating the volume of overlap for each distance. The OVH tells us what percentage of the OAR is within a given distance of the target and thus contains information about how hard it is to dosimetrically spare the OAR using IMRT. The OVH allows us to characterize patients based on their OAR‐target relationships, and thus allows us to search into the database of prior patients. We can query for patients with similar OVH characteristics, or we can find a group of patients whose OVHs suggest they are harder to plan.We have used our OVH query for both quality management and automatic treatment planning of head and neck and pancreatic cancers. For quality assessment, we take a new patient's OVHs and dosimetry data; look in to the database to find the best dose distribution achieved from the set of all patients whose OVH indicates they are harder to plan; then compare that queried dose distribution to the new patient's dose to indicate if the current treatment plan can be improved upon. For the automatic planning case, we use the queried doses to initialize the dosimetric objective function used in the inverse planning process of IMRT. The presentation will include an in‐depth discussion of the database and infrastructure design; the research and clinical results of the OVH based auto‐planning, and the software developed for clinical deployment with the dosimetrists.

Original languageEnglish (US)
Pages (from-to)4008
Number of pages1
JournalMedical Physics
Volume39
Issue number6
DOIs
StatePublished - 2012

Fingerprint

Databases
Organs at Risk
Therapeutics
Head and Neck Neoplasms
Pancreatic Neoplasms
Software
Radiation
Research

ASJC Scopus subject areas

  • Biophysics
  • Radiology Nuclear Medicine and imaging

Cite this

TH‐E‐BRCD‐02 : Automated Treatment Planning Using a Database of Prior Patient Treatment Plans. / McNutt, Todd; wu, B.; Moore, J.; Petit, S.; Kazhdan, M.; Taylor, Russell H.

In: Medical Physics, Vol. 39, No. 6, 2012, p. 4008.

Research output: Contribution to journalArticle

@article{2df49f2b12354fd9a2a7163444719c5a,
title = "TH‐E‐BRCD‐02: Automated Treatment Planning Using a Database of Prior Patient Treatment Plans",
abstract = "Intensity modulated RT (IMRT) is used to deliver highly conformal radiation treatments. Its broad adoption has introduced considerable inter‐patient and inter‐institutional variability in plan quality. As part of our Oncospace program, we have built a system that uses patient geometric information and a database of previously treated patients for IMRT treatment plan quality assessment and automated planning. The system uses the relationship between organs at risk (OARs) and target volumes to predict the achievable dose distributions for a given patient from a database of prior patients. It then uses this achievable dose distribution to guide the planning process for the new patient. This system also employs a feedback model to insure the database improves over time through continual improvement in plan quality. The basic premise is the overlap volume histogram (OVH), a shape descriptor we introduced to describe the complex spatial relationship between an OAR and the target volume. The OVH describes the volume of overlap between the OAR and an expanded/contracted target as a function of the expansion/contraction distance. It is determined by expanding/contracting the target volume and calculating the volume of overlap for each distance. The OVH tells us what percentage of the OAR is within a given distance of the target and thus contains information about how hard it is to dosimetrically spare the OAR using IMRT. The OVH allows us to characterize patients based on their OAR‐target relationships, and thus allows us to search into the database of prior patients. We can query for patients with similar OVH characteristics, or we can find a group of patients whose OVHs suggest they are harder to plan.We have used our OVH query for both quality management and automatic treatment planning of head and neck and pancreatic cancers. For quality assessment, we take a new patient's OVHs and dosimetry data; look in to the database to find the best dose distribution achieved from the set of all patients whose OVH indicates they are harder to plan; then compare that queried dose distribution to the new patient's dose to indicate if the current treatment plan can be improved upon. For the automatic planning case, we use the queried doses to initialize the dosimetric objective function used in the inverse planning process of IMRT. The presentation will include an in‐depth discussion of the database and infrastructure design; the research and clinical results of the OVH based auto‐planning, and the software developed for clinical deployment with the dosimetrists.",
author = "Todd McNutt and B. wu and J. Moore and S. Petit and M. Kazhdan and Taylor, {Russell H}",
year = "2012",
doi = "10.1118/1.4736348",
language = "English (US)",
volume = "39",
pages = "4008",
journal = "Medical Physics",
issn = "0094-2405",
publisher = "AAPM - American Association of Physicists in Medicine",
number = "6",

}

TY - JOUR

T1 - TH‐E‐BRCD‐02

T2 - Automated Treatment Planning Using a Database of Prior Patient Treatment Plans

AU - McNutt, Todd

AU - wu, B.

AU - Moore, J.

AU - Petit, S.

AU - Kazhdan, M.

AU - Taylor, Russell H

PY - 2012

Y1 - 2012

N2 - Intensity modulated RT (IMRT) is used to deliver highly conformal radiation treatments. Its broad adoption has introduced considerable inter‐patient and inter‐institutional variability in plan quality. As part of our Oncospace program, we have built a system that uses patient geometric information and a database of previously treated patients for IMRT treatment plan quality assessment and automated planning. The system uses the relationship between organs at risk (OARs) and target volumes to predict the achievable dose distributions for a given patient from a database of prior patients. It then uses this achievable dose distribution to guide the planning process for the new patient. This system also employs a feedback model to insure the database improves over time through continual improvement in plan quality. The basic premise is the overlap volume histogram (OVH), a shape descriptor we introduced to describe the complex spatial relationship between an OAR and the target volume. The OVH describes the volume of overlap between the OAR and an expanded/contracted target as a function of the expansion/contraction distance. It is determined by expanding/contracting the target volume and calculating the volume of overlap for each distance. The OVH tells us what percentage of the OAR is within a given distance of the target and thus contains information about how hard it is to dosimetrically spare the OAR using IMRT. The OVH allows us to characterize patients based on their OAR‐target relationships, and thus allows us to search into the database of prior patients. We can query for patients with similar OVH characteristics, or we can find a group of patients whose OVHs suggest they are harder to plan.We have used our OVH query for both quality management and automatic treatment planning of head and neck and pancreatic cancers. For quality assessment, we take a new patient's OVHs and dosimetry data; look in to the database to find the best dose distribution achieved from the set of all patients whose OVH indicates they are harder to plan; then compare that queried dose distribution to the new patient's dose to indicate if the current treatment plan can be improved upon. For the automatic planning case, we use the queried doses to initialize the dosimetric objective function used in the inverse planning process of IMRT. The presentation will include an in‐depth discussion of the database and infrastructure design; the research and clinical results of the OVH based auto‐planning, and the software developed for clinical deployment with the dosimetrists.

AB - Intensity modulated RT (IMRT) is used to deliver highly conformal radiation treatments. Its broad adoption has introduced considerable inter‐patient and inter‐institutional variability in plan quality. As part of our Oncospace program, we have built a system that uses patient geometric information and a database of previously treated patients for IMRT treatment plan quality assessment and automated planning. The system uses the relationship between organs at risk (OARs) and target volumes to predict the achievable dose distributions for a given patient from a database of prior patients. It then uses this achievable dose distribution to guide the planning process for the new patient. This system also employs a feedback model to insure the database improves over time through continual improvement in plan quality. The basic premise is the overlap volume histogram (OVH), a shape descriptor we introduced to describe the complex spatial relationship between an OAR and the target volume. The OVH describes the volume of overlap between the OAR and an expanded/contracted target as a function of the expansion/contraction distance. It is determined by expanding/contracting the target volume and calculating the volume of overlap for each distance. The OVH tells us what percentage of the OAR is within a given distance of the target and thus contains information about how hard it is to dosimetrically spare the OAR using IMRT. The OVH allows us to characterize patients based on their OAR‐target relationships, and thus allows us to search into the database of prior patients. We can query for patients with similar OVH characteristics, or we can find a group of patients whose OVHs suggest they are harder to plan.We have used our OVH query for both quality management and automatic treatment planning of head and neck and pancreatic cancers. For quality assessment, we take a new patient's OVHs and dosimetry data; look in to the database to find the best dose distribution achieved from the set of all patients whose OVH indicates they are harder to plan; then compare that queried dose distribution to the new patient's dose to indicate if the current treatment plan can be improved upon. For the automatic planning case, we use the queried doses to initialize the dosimetric objective function used in the inverse planning process of IMRT. The presentation will include an in‐depth discussion of the database and infrastructure design; the research and clinical results of the OVH based auto‐planning, and the software developed for clinical deployment with the dosimetrists.

UR - http://www.scopus.com/inward/record.url?scp=85024806307&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85024806307&partnerID=8YFLogxK

U2 - 10.1118/1.4736348

DO - 10.1118/1.4736348

M3 - Article

AN - SCOPUS:85024806307

VL - 39

SP - 4008

JO - Medical Physics

JF - Medical Physics

SN - 0094-2405

IS - 6

ER -