Literature DB >> 7612133

Structure analysis of breast lesions using neighborhood graphs.

C A Beltrami1, V Della Mea, N Finato.   

Abstract

The College of American Pathologists divides proliferative lesions of the breast into three categories, but the diagnostic criteria for classification are not easy to apply. Therefore, many morphologic, immunohistochemical and morphometric features were studied in order to clarify this subdivision; however, the reported results are not very satisfactory. We also attempted to identify and differentiate the architectural features useful for the diagnosis of proliferative breast lesions. The morphologic characteristics of proliferative lesions of the breast were considered amenable to a graph theoretical approach. From a set of images of mammary ducts acquired from a hematoxylin-eosin-stained section, a hierarchy of graphs was built. The main graphs representing the structure are: (1) neighborhood graph, (2) planar graph, derived from (1) and (3) a dual graph. A simplified prototype of a system for structure analysis of the lesions was implemented and tested. Forty duct images classified by two expert pathologists were acquired and subjected to the system. The data obtained were analyzed, and some significant graph features validating our approach to the structure representation of the lesions were discovered. This approach can also be applied to other, similar classification problems and may help in understanding the biologic significance of proliferative lesions of the breast.

Entities:  

Mesh:

Year:  1995        PMID: 7612133

Source DB:  PubMed          Journal:  Anal Quant Cytol Histol        ISSN: 0884-6812            Impact factor:   0.302


  1 in total

1.  25 years of telepathology research: a bibliometric analysis.

Authors:  Vincenzo Della Mea
Journal:  Diagn Pathol       Date:  2011-03-30       Impact factor: 2.644

  1 in total

北京卡尤迪生物科技股份有限公司 © 2022-2023.