Optimization Days 2019

HEC Montréal, May 13-15, 2019

JOPT2019

HEC Montréal, 13 — 15 May 2019

Schedule Authors My Schedule

WA4 Healthcare Optimization IV

May 15, 2019 09:00 AM – 10:15 AM

Location: Hélène-Desmarais

Chaired by Marc-Andre Renaud

3 Presentations

  • 09:00 AM - 09:25 AM

    CyberKnife radiation therapy treatment planning with column generation

    • Kafaei Peyman, presenter,
    • Louis-Martin Rousseau, Polytechnique Montréal
    • Nicolas Chapados, Element AI

    Stereotactic body radiation therapy (SBRS) is a non-surgical radiation therapy used to deliver precisely targeted radiation towards tumor in a patient's body. The main purpose of this work is to develop a mathematical programming model producing clinically accepted treatment plan which saves normal tissues. Column generation has proven to be useful for solving large-scale optimization problems in the field of Radiation Therapy, especially for Direct Aperture Optimization. We represent an algorithm based on Column Generation for CyberKnife delivery system, which finds the shape of the apertures by solving the sub-problems sequentially as well as the intensity of related radiation beams in the master problem.

  • 09:25 AM - 09:50 AM

    Robust mixed electron-photon radiation therapy treatment planning using the column generation method

    • Marc-Andre Renaud, presenter, McGill University
    • Monica Serban, McGill University
    • Jan Seuntjens, McGill University

    Mixed beam electron-photon radiation therapy (MBRT) is emerging as a potential technique to reduce radiation dose delivered to normal tissue without compromising target coverage for some superficial tumours. However, dose distributions delivered from charged particle beams are known to be non-robust to patient positioning errors. In this work, we develop a robust stochastic optimization model for MBRT planning based on the column generation method. The model is used to produce robust MBRT treatment plans for chest wall and soft tissue sarcoma treatment patients, and differences between robust and non-robust plans are analysed.

  • 09:50 AM - 10:15 AM

    Treatment planning in VMAT using column generation : comparing arc versus control-point based approaches

    • Nina Dissler, presenter, Polytechnique Montreal
    • Marc-Andre Renaud, McGill University
    • Nadia Lahrichi, Polytechnique Montréal
    • Louis-Martin Rousseau, Polytechnique Montréal

    VMAT is a sophisticated form of radiotherapy with a continuous rotation of the collimator around the patient. Optimisation methods are used to determine gantry speed, beam intensity and collimator leaf motions to target the tumor while avoiding surrounding sane tissues. Two approaches of column generation for the treatment planning problem are compared. On the one hand, generating beam aperture shapes in the pricing and then solving the master problem while ensuring that each aperture is compatible with the previous one. On the other hand optimizing simultaneously the gantry speed, dose intensity and leaf motions to generate arcs in the pricing before solving the master problem. The problem is solved under technical constraints of the machine, while medical suggestions are taken into account in the objective functions. The treatment efficiency and computational time are compared between the two methods.

Back