Frequently Asked Questions

In Oncospace, Predictive Planning makes use of sophisticated machine learning models built to support a wide range of flexible protocols.

For a given anatomical treatment site like prostate, these models support protocols for target volumes that may include the prostate bed, or lymph nodes, or not, and delivery options like SIB, hypofractionation, and SBRT.

For H&N planning these models support protocols with target volumes in various anatomies, in unilateral or bilateral configurations, and deliveries with multiple dose levels.

What might be separate models with alternate knowledge-based approaches, and the curating, training, and validating they entail, is covered out of the box cloud with Predictive Planning.

In addition to Predictive Planning for select indications, Oncospace is also an application for the entire planning workflow with protocol management tools that help confirm data integrity and communicate planning intent, as well as scorecards that allow for evaluation of planning results of all radiation therapy planning indications.
Properly configured, Oncospace will send to a TPS information in the “Physician Directive” (e.g., predicted dose objectives and delivery method) to support the creation of a treatment plan to these treatment planning systems:
 
  • Philips Pinnacle™ 9.8, or later

Oncospace is an International Standards Organization (ISO) 27001 Information Security Management Systems (ISMS) certified company (link) that encompasses all our products, services, and business practices.

To highlight some specific areas of interest,

    • Patient information:  Prior to being uploaded to Oncospace patient data is anonymized and double encrypted via customer and Oncospace key (RSA-2048 implemented in HSM).

    • Data in transit:  Patient data is protected via browser supported SSH/SSL (TLS 1.2).
    • Data at rest:  Once in your organization’s dedicated subscription, patient data protected via Microsoft Azure at-rest encryption for file and database storage.
    • User Authentication:  For a more secure deployment, Oncospace leverages the active directory (AD; Azure or Microsoft) policies and requirements of your organization.

 

We look forward to reviewing our implementation in the context of your information security requirements

For a treatment plan with a defined dose prescription to one or more target volumes, Oncospace Predictive Planning predicts achievable dose volume histograms (DVH) for organs at risk (OAR). The proprietary, domain informed machine learning (ML) model used for dose predictions takes as inputs the prescription and a range of features describing the spatial relationship between the target(s) and each OAR.  For each DVH volume, a dose range is predicted representing the level of OAR dose sparing that should be achievable.

 

Oncospace Predictive Planning uses a regression ensemble ML model, including feature extraction and 5-fold cross-validation for hyperparameter optimization.

 
Refer to the following for more information:
 

 

 

Look here for more information on the rich history of Oncospace related publications.

Oncospace preserves user specified naming while adhering to the following naming convention for Organ at Risk (OAR) structures when creating protocol templates and machine learning models:
 
  • American Association of Medical Physicists (AAPM) Task Group (TG)-263

Access Oncospace from almost anywhere with these supported web browsers:

  • Apple Safari™ 14.1, or later
  • Google Chrome™ 84, or later
  • Microsoft Edge™ 84, or later
  • Mozilla Firefox™ 68, or later

Access Oncospace from almost anywhere with these supported web browsers:

  • Apple Safari™ 14.1, or later
  • Google Chrome™ 91, or later
  • Microsoft Edge™ 91, or later
  • Mozilla Firefox™ 89, or later

All product and company names are trademarks ™ or registered ® trademarks of their respective holders. Use of them does not imply any affiliation with or endorsement by them.