Purposes: This study aimed to quantitatively evaluate the nasopalatine duct (NPD) and nasopalatine duct cysts (NPDC) using CT texture analysis.
Materials and Methods: This IRB-approved retrospective study included 31 NPDC patients (21 males, 10 females; mean age: 44.2, range: 22-83) and 31 age- and sex-matched patients with normal NPD (21 males, 10 females; mean age: 44.2, range: 22-83) who underwent head and neck CT. The NPDC patients were identified via histopathological examinations of surgical specimens. Patients with maxillary tumors or cysts (excluding NPDC), severe metal artifacts that prevented visualization of the NPD, or that were aged<20 years old were excluded. The radiomic features of NPDC and normal NPD were analyzed using the open-access software MaZda Ver.3.3 (Technical University of Lodz, Institute of Electronics, Poland). Two hundred and ninety-four radiomic features were evaluated in each ROI. Then, 10 textural features were selected using the MaZda feature selection tools, which are based on Fisher’s coefficient. The results were compared with the Mann-Whitney U test.
Results: One histogram feature (the 90th percentile); 6 GLCM features (S (0, 3) InvDfMom, S (3, −3) InvDfMom, S (0, 4) InvDfMom, S (4, −4) InvDfMom, S (0, 5) InvDfMom, and S (5, −5) InvDfMom); and 3 GLRLM features (45° fraction, vertical fraction, 135° fraction) differed significantly between NPDC and normal NPD (
P<0.05).
Conclusion: CT texture analysis was able to quantitatively evaluate normal NPD and NPDC. Our results suggest that CT texture analysis might be useful for differentiating NPDC from normal NPD.
View full abstract