Radiomic Features of Organic Phantoms Using Photon-Counting CT

This dataset consists of radiomic features extracted from CT scans of 16 organic phantoms (four apples, four onions, four limes, and four kiwis). Each phantom was scanned twelve times using a first-generation dual-source photon-counting CT (PCCT) system (NAEOTOM Alpha; Siemens Healthcare GmbH). The scans were performed at various tube currents (10, 50, 100 mAs) and included both test and retest scans, with phantoms in their initial and rotated positions.

The dataset contains radiomic features extracted from both polyenergetic reconstructions (PER) and virtual monoenergetic reconstructions (VMER) of the images. The reconstructions were performed using a Br40 kernel for PER images and a Qr40 kernel for VMER images, with energy levels ranging from 40 to 190 keV.

For each phantom, a total of 105 features were extracted using the PyRadiomics package, including shape-based features, first-order statistics (FO), and texture features such as GLCM, GLSZM, GLRLM, NGTDM, and GLDM. The images were processed with voxel normalization, resampling to a voxel size of 2 × 2 × 2 mm, and a fixed bin width of 25 Hounsfield units.

PER: Polyenergetic Reconstruction
VMR: Virtual Monoenergetic Reconstruction