Radiation and Cancer Physics

PV 03 - Poster Viewing Q&A - Session 3

TU_43_3817 - Improving Efficiency in Radiation Therapy Treatment Planning Using Automated Organ-at-Risk Prioritization

Tuesday, September 17
1:00 PM - 2:15 PM
Location: ASTRO Innovation Hub

Improving Efficiency in Radiation Therapy Treatment Planning Using Automated Organ-at-Risk Prioritization
E. Aliotta, H. Nourzadeh, W. Choi, and J. V. Siebers; University of Virginia, Charlottesville, VA

Purpose/Objective(s): The time-intensiveness of organ-at-risk (OAR) manual delineation (MD) and/or manual review (MR) limits clinical efficiency, particularly in adaptive therapy. Autodelineation (AD) can reduce the need for MD, but MR is still typically utilized to ensure accuracy. Recognizing that some OARs are sufficiently far from dose tolerances that they do not pose a realistic risk, we hypothesize that some OARs can be excluded from MD or MR with no clinical impact. To improve delineation efficiency, this work develops an OAR prioritization method that combines AD with a dose estimate to identify OARs that can be excluded from MD / MR without affecting plan quality.

Materials/Methods: Five head and neck cases were retrospectively evaluated. For each, an initial dose estimate (Dest) was optimized using only target volumes and target-based ring structures. AD OARs (OARAD) were then generated using a SPICE algorithm. Conservative dose metric estimates for each OAR accounted for potential delineation errors as follows: Let M be the maximum delineation uncertainty for OARi. Dmax for OARi is less than or equal to the max dose in OARAD,i plus an M mm surrounding envelope. We conservatively assign Dmax(OARi)=Dmax(OARAD,i+M). To conservatively estimate Dmean, OARs contracted by M mm (to underestimate the volume) were shifted towards the Dest high-dose region up to the edge of OARAD,i+M (to over-estimate Dmean). M=5mm was used for this study. OARs with estimated Dmax or Dmean below their clinical objective value were deemed low-priority and thus excluded from MD or MR. All other OARs were deemed high-priority. Technique evaluation utilized four auto-generated plans for each case. Optimizations used:1) clinically used manual OARs (OARMD); 2) only the high-priority OARs from OARMD (excluding low priority OARs); 3) a mixed OAR set with high-priority OARs from OARMD and low-priority OARs from OARAD; and 4) only OARAD. Post-optimization, dose objectives were evaluated on OARMD for each plan. Each Dobj (Dmean or Dmax) from plans 2-4 was compared with plan 1.

Results: Across 5 cases, OAR prioritization reduced the total number of MD OARs by 35% from 116 to 76 with per-case reductions ranging from 8% to 67% (Table). Ignoring low-priority OARs (plan 2) increased Dobj (non-significantly) across all OARs by 1±9% compared with plan 1(max=68%, p=0.149). Including low-priority AD OARs (plan 3) reduced this change to 0±4% (max=20%, p=0.785). In comparison, using only AD contours (plan 4) increased Dobj significantly by 8±22% (max=85%, p=0.001). No low-priority OARs exceeded clinical objectives in plan 3 (mixed OARs), while one did in plan 2 (low-priority OARs ignored).

Conclusion: Prioritizing OARs using estimated dose distributions allowed a 35% reduction in manual delineation or review without affecting clinically relevant dosimetry.
Site

1 Supraglottis

3 Base of Tongue 3 Softpallate 4 Nasopharynx 5 Oropharynx Total
Total OARs 23 18 28 25 22 116
Low-priority OARs 14 12 5 2 7 40
OAR Reduction 61% 67% 18% 8% 32% 35%

Author Disclosure: E. Aliotta: None. H. Nourzadeh: None. W. Choi: None. J.V. Siebers: Consultant; Reflexion Medical. Research Grant; Philips Medical Systems, Varian Medical Systems. Independent Contractor; Reflexion Medical.

Eric Aliotta, PhD

University of Virginia

Presentation(s):

Send Email for Eric Aliotta


Assets

TU_43_3817 - Improving Efficiency in Radiation Therapy Treatment Planning Using Automated Organ-at-Risk Prioritization



Attendees who have favorited this

Please enter your access key

The asset you are trying to access is locked. Please enter your access key to unlock.

Send Email for Improving Efficiency in Radiation Therapy Treatment Planning Using Automated Organ-at-Risk Prioritization