Improving CBCT image quality for daily image guidance of patients with head/neck and prostate cancer.

Purpose or Objective Image quality of on-board CBCT imaging in radiation therapy generally falls short of diagnostic CT in particular in terms of low contrast visibility. This limits the application of CBCT mainly to patient setup based on high contrast structures. We address these limitations by applying advanced preprocessing and reconstruction algorithms to improve patient setup and facilitate advanced applications like adaptive radiotherapy. Material and Methods The commercially available TrueBeam CBCT reconstruction pipeline removes scatter usi ng a kernelbased correction followed by filtered bac k-projectionbased reconstruction (FDK). These reconstruction n pipeline steps are replaced by a physics-based scatter correction (pelvis only) and an iterative reconstruction. We use statistical reconstruction that takes the Poisson distribution of quantum noise into account, an d applies an edge preserving image regularization. The advanced scatter correction is based on a finite-ele ment solver (AcurosCTS) to model the behavior of photons as they pass (and scatter) through the object. Both algorit hms have been implemented on a GPU cluster pla tform, and algorithmic acceleration techniques are utilized to achieve clinically acceptable reconstruction times. The image quality improvements have been an alyzed on TrueBeam kV imaging system phantom scans, as well as on daily CBCT scans of head/neck and prostate cancer patients acquired for image-guided localization. Results Artifacts in head/neck FDK reconstructions (Fig . 1) e.g. resulting from photon starvation in the shoulder region or cone-beam are highly reduced in the iterative reconstructions. The iterative reconstruction s show enhanced soft tissue definition providing better cl arity for boundary definition (see the level 2 lymph node located in the contoured region of the axial view, Fig. 1). The advanced scatter correction applied for pelvis scans removes residual scatter artifacts, increasing the mean homogeneity from 78.2 HU ± 18.0 HU to 20.9 HU ± 10.9 HU within the bladder region of 9 daily CBCT scans of typical prostate patients. Iterative reconstruction provides further benefit by reducing image noise as well as eliminating streak and cone-beam artifacts, thereby significantly improving soft-tissue visualization, as noted in the clinical pelvis CBCT scan (Fig. 2). The noise level was reduced to 45% of the original value. Conclusion Statistical reconstruction in combination with advanced scatter correction substantially improves CBCT image quality by enabling removal of artifacts caused by remaining scatter, projection noise, photon starvation, and cone-beam angle. These artifact reductions improve soft tissue definition that is necessary for accurate visualization, contouring, dose calculation, and deformable image registration in clinical practice. The presented improvements are expected to facilitate soft tissue-based patient setup. Promise has been demonstrated for new applications, such as adaptive radiotherapy. (Figure presented)