Potentials and limits of a novel CT reconstruction algorithm (DirectDensity™) developed for radiotherapy treatment planning

Purpose: Computed tomography (CT) represents the standard of treatment planning in radiotherapy. Calculation of dose distribution requires proper energy-dependent curves to convert Hounsfield Units into relative to water electron densities (rED). To possibly reduce the patient dose while maintaining optimal image quality, it could be useful to modulate x-ray tube potential (kVp) based on patient size, but this practice requires the use of several conversion curves in the treatment planning system (TPS). The DirectDensity (TM) (DD) algorithm, recently developed by Siemens Healthcare, uses projection data to reconstruct images directly proportional to rEDs, regardless of the used kVp. The purpose of this study was to evaluate this algorithm for clinical use in radiotherapy. Methods: Image quality was evaluated on Catphan images reconstructed with standard and DD algorithms. The CIRS Electron Density Phantom, with its several tissue-equivalent inserts, was then used to assess the accuracy of reconstructed rEDs against those certified at different tube voltages. Finally, the treatment plans calculated on clinical images reconstructed with standard and DD algorithms were compared. Results: DD showed an image quality comparable to standard algorithms except for the contrast: DD causes a loss in contrast (up to 69%) for materials densities greater than 1.1 g cm(-3). CIRS images analysis proved the rEDs reconstruction accuracy except for the metal inserts: titanium rED was underestimated up to 40% when kVp was reduced. Nevertheless, minimal dosimetric differences were found between treatment plans calculated in different body regions. Conclusions: The DD algorithm allows to plan radiotherapy treatments on CT images acquired at tube potentials optimized both for the anatomical region and patient size, without complicating the clinical workflow with tube potential specific calibration procedures. However, the presence of contrast medium and metal implants can limit the applications of this algorithm in clinical practice.