Morphometry of the lymphoid cells was performed with a computerized image analyzer. Database formation was performed automatically using our own new software. The initial database of calculated mean values, of constructed histograms, of regression, and of discriminant analysis was transformed into a set of morphometric features. Eight morphometric features were selected for characterizing two separate populations of the lymphoid cells: one from the lymph node tissue with a malignant tumor and the other without any tumor (control). The eight features ensured the least overlap between the two groups of lymphoid cells. The eight features formed the basis of the decision criteria for identifying a patient with a malignant pathology. A weighting coefficient was calculated for each feature. It was ascertained that the presence of the regional papillary thyroid cancer metastases causes significant changes in the relationship between small (diameter </= 7.5 mum) and large (>7.5 mum) lymphoid cells of the cervical lymph nodes. The morphometric changes were assessed by our diagnostic index. The expert diagnostic system was developed on the basis of a set of quantitative features and their boundary values. This system highlights the morphometric changes in the population of the cervical lymph nodes lymphoid cell with metastases versus control. Clinical trials have shown that the expert diagnostic system's conclusions coincided with the histological diagnosis in all the cases. The expert diagnostic system, image processing, database formation, assessment of lymphoid cells morphometric features formed the basis of the rapid method of diagnosing regional metastases of papillary thyroid cancer. The diagnosis was performed in automatic mode. Taking into account that about 30-50% of aspiration smears contain only blood cells (absence of specialized cells such as thyrocytes), the use of our method should greatly improve the cytologic diagnosis of regional metastases of papillary thyroid cancer.
(c) 2006 Wiley-Liss, Inc.