Back to the list
Congress: ECR25
Poster Number: C-10473
Type: Poster: EPOS Radiologist (scientific)
Authorblock: M. H. Zhang, C. Ezenekwe, I. Hussain, D. T. Ginat; Chicago/US
Disclosures:
Michael Hanzhe Zhang: Nothing to disclose
Chime Ezenekwe: Nothing to disclose
Irfan Hussain: Nothing to disclose
Daniel Thomas Ginat: Nothing to disclose
Keywords: Ear / Nose / Throat, CT, Computer Applications-Detection, diagnosis, Tissue characterisation
Methods and materials

51 subjects with history of 4D neck CT, parathyroidectomy, and pathology proven parathyroid adenoma(s) were retrospectively reviewed for this study. Two anatomical structures (parathyroid adenoma and cervical lymph node) were manually segmented on a 4D CT 25 sec arterial phase CT reconstructed to a slice thickness of 3mm. Segmentation was performed by a medical student or diagnostic radiology resident under the supervision of an attending neuroradiologist. Manual segmentation was performed with 3D Slicer, an open source image computing platform. Radiomic data of the shape, first order, and second order classes (107 variables) were extracted from parathyroid adenomas and lymph nodes. Shape features simply analyze the contour of the segmented region of interest. First order radiomic features are relatively simple parameters which describe the CT voxel intensity based on one-by-one voxel. Second order features, also known as texture features, describe voxel values while taking into account the spatial arrangement. Second order features include entropy and variance and for instance can help quantify how uniform a structure is on CT imaging. A t-test was utilized to assess the radiomic differences between the three structures of interest.

GALLERY