ACOSA: A Script-Based Algorithm for Multi-Disease Target Crop and Optimization in Radiotherapy

IntroductionCommercial automatic planning modules are currently limited to single, conventional disease types, which severely restricts their utility when dealing with unconventional plans. Given that such unconventional plans are actually the norm in most hospitals, there is an urgent need for an automatic planning algorithm that can be applied to a wide range of clinical situations. To address this issue, we developed an algorithm capable of automatically cropping target areas and setting optimization conditions for multiple diseases, known as the Auto Crop and Optimization Setup Algorithm (ACOSA). This paper presents the principles of ACOSA and conducts a preliminary comparative evaluation of its performance against existing solutions.MethodsThe development of ACOSA utilized the Eclipse Script Application Programming Interface (ESAPI) scripting language provided by Eclipse. Based on the input prescriptions, the algorithm simulates the operations of a physicist, automatically crops the target areas, and sets appropriate optimization parameters. Retrospectively, 20 cases of glioma and head and neck cancers were selected. Without considering organ-at-risk dose limits, dose calculations were performed using both ACOSA and Eclipse's built-in AutoCrop, and a dosimetric comparison was conducted.ResultsIn terms of target volume homogeneity index (<i>HI</i>) and D98, the AutoCrop group demonstrated slight superiority over the ACOSA group. However, the ACOSA group exhibited superior performance in conformity index (<i>CI</i>), gradient index (<i>GI</i>), <i>D2</i>, and particularly in parameters reflecting the rate of low-dose fall-off outside the target volume, including <i>Ratio20</i>, <i>Ratio30</i>, and <i>Ratio40</i>, when compared to the AutoCrop group.ConclusionsACOSA can be reliably applied in clinical settings and demonstrates superiority over the AutoCrop module of the Eclipse planning system.