Paper # 018

Versión en Español

Morphometry on cellular aspirates of infiltrant ductal carcinoma of the breast. Multivariate statistics model.

Dra. Inna Antunez Potashkina(1), Dra. Rosa María Coro Antich(2), Dra. Soraya Rodríguez Ceballos(3).

(1) Centro de Neurotrasplante, (2) Instituto de Cardiología,( 3) Hospital Hermanos Ameijeiras. La Habana, Cuba

Address: Instituto de Cardiología. La Habana, Cuba

[Introduction] [Materials & Methods] [Results] [Pictures] [Discussion] [Bibliography]

Summary

High interobserver variabilities affect the reliability of nuclear grades determined from cellular aspirates. This work explores a way to diminish subjective elements in nuclear grading by means of computerized image analysis together with the Bayesian method of multivariate statistics. Ten slides from each of three nuclear grades, diagnosed according to traditional criteria (visual diagnosis), were used as a training set to determine class descriptors (computerized diagnosis). Because of the low coincidence between visual and computerized diagnoses (66.7%), a new classification of the slides was made considering only low and high nuclear grades; the coincidence then increased up to 90%. The usefulness of this method was shown by the application of the canonical functions equations to 12 cases aside from the training set. The possibility of obtaining quick morphometric results represents a definite advance in pathologic diagnosis, and may be a decisive factor in hard to classify cases.

Inna Antunez Potashkina has recently graduated in Biological Sciences (1996). This paper was her graduating thesis. She is a fellow researcher at Centro Internacional de Restauración Neurológica, better know as Neurotrasplant Center. She works at Morphology Laboratory.

Rosa María Coro Antich PhD is Biologist, specialzed in Computerized Image Analisis applied to Pathology. She belongs to the Cardiology Institute, and she also works in the Department of Pathology at Hospital Ameijeiras (directed by Prof. Israel Borrajero); she is Assistant Professor in Cellular Biology at Biology School in the Universidad de La Habana. She belongs to the development equipment authoring Digipat system in a software firm. She is Secretary of the Cuban Society of Morphological Sciences.

Soraya Rodriguez Ceballos is pathologist

Key Words: Morphometry, Image analysis, Breast cancer, Nuclear grade

• Introduction
• Materials & Methods
• Results
• Pictures
• Discussion
• Bibliography