Classification of anatomic structures in head and neck by CT-based radiomics

Background and purpose: Radiomics features are used to identify disease types predict therapy outcomes. However, how the radiomics different among anatomical structures has never been investigated. Hence, we analyzed of 22 in head neck area CT images. Furthermore, studied whether can classify using unsupervised machine-learning techniques. Materials methods: We obtained IMRT/VMAT treatment planning data from 36 patients treated for cancers a single institution. There were 1357 contours more than drawn on CTs. calculated 174 SIBEX program. First, tested unique enough all into groups. then developed two-stage clustering technique anatomic sub-groups with similar physiological or biological characteristics. Results: The heatmap showed distinct difference tumors other healthy structures. have allowed us eyes, lens, submandibular, pituitary glands, thyroids over 90% accuracy. resulted six subgroups, which shared common characteristics such as fatty bony tissues. Conclusions: shown that distinguishable features. could observe similarities subgroups results suggest help distinguish lesions.