MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

A Deep-Learning Method of Automatic VMAT Planning Via MLC Dynamic Sequence Prediction (AVP-DSP) Using 3D Dose Prediction: A Feasibility Study of Prostate Radiotherapy Application

Y Ni1*, J Zhang2, Y Sheng2, X Li2, J Ye3, Y Ge4, Q Wu2, C Wang2, (1) Duke Kunshan University, Kunshan, 32, CN, (2) Duke University Medical Center, Durham, NC, (3) Swedish Medical Center, Seattle, WA, (4) University of North Carolina at Charlotte, Charlotte, NC,

Presentations

(Tuesday, 7/14/2020) 3:30 PM - 5:30 PM [Eastern Time (GMT-4)]

Room: Track 3

Purpose:
To develop a deep-learning method to automate VMAT radiotherapy planning, via direct MLC sequences prediction (AVP-DSP), using a patient’s 3D structure volumes and dose predictions.


Methods:
AVP-DSP generates 2D projections of the following 3D volumes at 100 control point angles of a single 360º arc: 1)PTV contours on their radiation exit interfaces; 2)OAR contours on isocenter plane and 3)predicted dose distribution on isocenter plane. These projections are used as inputs of a customized U-net implementation. The outputs are binarized MLC aperture shapes at 100 control points, which are subsequently resampled and automatically sent to a commercial treatment planning system (TPS) for plan finalization.
AVP-DSP was developed using 131 prostate patients who received simultaneously-integrated-boost(SIB) treatment (58.8Gy/70Gy to PTV(58.8)/PTV(70) in 28fx). All patients were planned by a 360º single-arc VMAT technique using an in-house knowledge-based planning tool in a commercial TPS. 120 plans were used in AVP-DSP training/validation. As a feasibility study, the TPS-calculated 3D dose distribution was used as 3D dose prediction surrogate. 11 plans were used as independent tests. Key dosimetry metrics achieved by AVP-DSP were compared against the ones planned by the commercial TPS.


Results:
After dose normalization (PTV(70) V70Gy=95%), all 11 AVP-DSP test plans met institutional clinic guidelines of dose distribution outside PTV. Bladder (V70Gy=6.8±3.6cc, V40Gy=19.4±9.2%) and rectum (V70Gy=2.8±1.8cc, V40Gy=26.3±5.9%) results in AVP-DSP plans were comparable with the commercial TPS plan results (bladder V70Gy=4.1±2.0cc, V40Gy=17.7±8.9%; rectum V70Gy=1.4±0.7cc, V40Gy=24.0±5.0%). 3D max dose results in AVP-DSP plans(D1cc=118.9±4.1%) were higher than the commercial TPS plans results(D1cc=106.7±0.8%). On average, AVP-DSP used 30 seconds for a plan generation in contrast to the current clinical practice (>20 minutes).

Conclusion:
Results suggest that AVP-DSP can generate a prostate VMAT plan with clinically-acceptable dosimetric quality. With its high efficiency, AVP-DSP may hold great potentials of real-time planning application after further validation.

Funding Support, Disclosures, and Conflict of Interest: This work was partially supported by NIH R01CA201212

Keywords

Radiation Therapy, Modeling, Treatment Planning

Taxonomy

TH- External Beam- Photons: Treatment planning using machine learning/Knowledge Based Planning/automation

Contact Email