Data mining with eclipse scripting

Abstract

The eclipse treatment planning system scripting interface allows the creation of software scripts to access the patient plans in the eclipse database. These scripts may be “plug-ins”, which are embedded within eclipse and provide access to the current patient. Scripts may also be created as stand-alone programs which can access the entire eclipse database. Data mining is the computational process of data or information discovery through the analysis of databases (“big data”) and the identification of correlations and relationships among the data. These techniques have been used to investigate a number of clinical issues using the facility provided by eclipse scripting. We recently migrated from the use of the Pencil Beam Convolution (PBC) dose calculation algorithm to the Analytic Anisotropic Algorithm (AAA) for all treatment sites. We have investigated: 1. Have the field sizes of lung plans increased when using the AAA compared to PBC to deliver PBC-like dose distributions? 2. Are AAA Breast plans significantly cooler than PBC plans to justify the reduction of our clinical protocol for maximum dose constraints? Eclipse scripts were created to analyse several years of PBC data to establish (1) field size to target volume relationships and (2) maximum dose distributions and to then verify that the AAA plans do not diverge from these relationships. Data mining with eclipse scripting allows rapid access to and the analysis of many years’ worth of patient plans enabling clinical decisions to be made in a timely manner and based on our own data and experience.