Improving Cancer Treatment Planning by DMH-Based Inverse Optimization

  • Mihaylov, Ivaylo B (PI)

Project: Research project

Project Details


DESCRIPTION (provided by applicant): Cancer patients continue to represent a challenging disease population, which faces rather poor prognosis with current treatment planning and delivery practices. Venues for a potential dose escalation and/or increased healthy tissue sparing, through innovative therapeutic approaches for those patients, are clearly needed. Current state of the art radiotherapy treatment planning relies on the dose-volume-histogram (DVH) paradigm, where doses to fractional (most often) or absolute volumes of anatomical structures are employed in both optimization and plan evaluation process. It has been argued however, that the effects of delivered dose seem to be more closely related to healthy tissue toxicity (and thereby to clinical outcomes) when dose-mass- histograms (DMHs) are considered in treatment plan evaluation. We propose the incorporation of mass and density information explicitly into the cost functions of the inverse optimization process, thereby shifting from DVH t DMH treatment planning paradigm. This novel DMH-based intensity modulated radiotherapy (IMRT) optimization aims in minimization of radiation doses to a certain mass, rather than a volume, of healthy tissue. Our working hypothesis is that DMH- optimization will reduce doses to healthy tissue substantially. In certain cases, with extensive, difficult to treat disease, lower doses to healthy tissue can be used for isotoxic dose escalation, which may result in an approximately two-fold increase in estimated loco-regional tumor control probability. To test this hypothesis we will pursue the following specific aims: (1) Develop the theoretical and computational framework of the DMH-based IMRT optimization. This framework will incorporate 3D and 4D IMRT as well as 3D volumetric modulated arc (VMAT) planning for different anatomical sites. (2) Investigate different parametric forms for DMH-optimization functions. The ultimate goal would be the simultaneous minimization of healthy tissue doses and/or escalation of therapeutic doses, without violating the established dosimetric tolerances for healthy anatomical structures. And (3) Practical implementation and application of this novel optimization paradigm, where virtual clinical trials for cohorts of lung, head-and-neck, and prostate cancer cases will be performed. Statistical significance of the DMH-optimization dosimetric improvements over standard of care DVH-optimization will be quantified. Prospective 3D and 4D CT data collection will be used to study the interactions between tumor time-trending changes and DMH-based optimization results. 4D CT data will also be used to investigate and quantify the correlation between DMH-based end points and the loss of pulmonary function during and after radiotherapy treatment. The deliverability (with the existing radiotherapy treatment equipment) of our 3D VMAT and 3D/4D IMRT plans will be experimentally verified, thereby paving the road for initiation of clinical trials. PUBLIC HEALTH RELEVANCE: The goal of the proposed research is to pave the road for a novel, dose-mass-histogram (DMH) based, inverse optimization for radiotherapy treatment planning. This novel optimization methodology will be applied for treatment planning for different anatomical sites. Successful development and validation of the proposed research will provide a general framework for generation of radiotherapy treatment plans, with substantially lower doses to healthy tissue, compared to current standard of care, realized through DVH- optimization.
Effective start/end date7/9/124/30/19


  • National Institutes of Health: $318,513.00
  • National Institutes of Health: $318,513.00
  • National Institutes of Health: $308,958.00
  • National Institutes of Health: $299,321.00
  • National Institutes of Health: $1.00
  • National Institutes of Health: $302,600.00


Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.