arXiv Science⌕ Search

arXiv subjects

Julius Arnold

Publications and source records attributed to Julius Arnold.

2 recordsLinked to original sources

Evaluation of attenuation-based simulated X-ray images for image-guided radiotherapy

In image-guided radiation therapy (IGRT), in-room X-rays are often used for patient positioning and monitoring, however, usually default imaging settings are utilized for each indication. Obtaining patient-specific imaging settings before treatment is achievable using simulated X-rays, which can facilitate IGRT workflow optimization. This study evaluated the image quality of simulated X-rays for thoracic, cranial and pelvic regions. Simulated X-rays were created from computed tomography (CT) scans with indication-specific default settings using a pre-release pilot software, which mimics the geometry of the room-mounted ExacTrac Dynamic (ETD) system (Brainlab SE, Germany). The basis of this algorithm was the Beer-Lambert law. Linear attenuation coefficients were approximated from the CT hounsfield units, while incident intensities were defined by flood images. CT scans were included from three anthropomorphic phantoms and 57 patients (13 lung, 20 glioblastoma, 24 prostate). For evaluation, real X-rays were used as ground truth. Three image quality metrics - Structural Similarity Index (SSIM), Normalized Cross Correlation (NCC), Normalized Mean Absolute Error (NMAE) - complemented by histogram and intensity profile comparisons were used for evaluation. SSIMs for phantom and patient average were 0.87 and 0.80 for the thoracic, 0.94 and 0.88 for the cranial and 0.85 and 0.79 for the pelvic region. Corresponding NCCs yielded values of 0.99 and 0.89, 0.99 and 0.99 and 0.99 and 0.60. NMAEs were 4.1% and 10.7% (lung), 2.6% and 3.2% (brain) and 2.6% and 4.0% (pelvis). Histograms of real X-rays and simulated X-rays (sX-rays) were slightly shifted and showed differences at the background pixel value. Intensity profiles exhibited similar trends with some differences in absolute values. The investigated simulated X-rays demonstrated promising quality for individualized IGRT treatment preparation.

physics.med-ph↗

COBRA2026: a large-scale multicenter pelvic cone-beam computed tomography projection dataset

The COBRA2026 dataset is a large-scale, multicenter resource of raw radiotherapy cone-beam computed tomography (CBCT) acquisitions created for the development and evaluation of conventional and learning-based reconstruction and image-correction methods. It contains data from 867 patients undergoing pelvic radiotherapy at six European centers, acquired using Elekta and Varian imaging systems. For each case, the dataset includes raw projection data, acquisition geometry, calibration and correction information, clinically reconstructed CBCT images, and corresponding planning CT images. Vendor-specific files were anonymized and converted into open formats. Planning CT images were deformably registered to the daily CBCT anatomy, and matched projections were simulated using the corresponding acquisition geometry. All cases underwent visual quality control, and cases with substantial processing or registration errors were excluded. The approximately 950 GB dataset is divided into training, validation, and test sets containing 692, 52, and 123 cases, respectively. Projection stacks and volumetric images are provided as compressed MetaImage files, with geometry and metadata supplied in XML and YAML formats. COBRA2026 supports research on full- and sparse-view reconstruction, low-dose imaging, artifact and scatter correction, motion compensation, and synthetic CT generation. The dataset is released under the CC BY-NC 4.0 license, indexed on Zenodo (doi:10.5281/zenodo.21322350), and accompanied by openly available preprocessing and baseline reconstruction code. It also forms the basis of the COBRA2026 reconstruction challenge.

physics.med-ph↗