Fine needle aspiration cytology (FNAC) is the final diagnosis of thyroid nodules before surgery. It is important to further improve the indeterminate FNAC diagnosis results using computerized cytological features. This retrospective cross-sectional study included 240 cases, of whom 110 had histologic diagnosis of papillary thyroid cancers (PTC), 100 had nodular/adenomatous goiters/hyperplasia (benign goiters), 10 had follicular/Hurthle cell carcinomas, and 20 had follicular adenomas. Morphological and chromatic features of FNAC were quantified and analyzed. The result showed that six quantified cytological features were found significantly different between patients with a histologic diagnosis of PTC and patients with histologic diagnosis of benign goiters in multivariate analysis. These cytological features were used to estimate the malignancy risk in nodules with indeterminate FNAC results. The Area Under the Receiver Operating Characteristics (AUROC) of the diagnostic accuracy with a benign or malignant nature was 81.3% (p < 0.001), 78.7% (p = 0.014), and 56.8% (p = 0.52) for nodules with FNAC results of atypia, which is suspicious for malignancy and follicular neoplasm, respectively. In conclusion, quantification of cytological features could be used to develop a computer-aided tool for diagnosing PTC in thyroid nodules with indeterminate FNAC results.
Keywords: computer analysis; diagnosis; morphometry; thyroid cytology.