News |  Sitemap |  Contact
PDF Export  | 

Systematic characterisation of cellular spatial arrangement in tissues.

Administrative Information

Organisation

The University of Birmingham, U.K.

Presentation Information

Full / Half Time Slot: Full Time Slot (50 min)

Contact / Speaker Name

Gabriel Landini

Presentation Title

Systematic characterisation of cellular spatial arrangement in tissues.

Participant Requirements

Medium to advanced

Biography of Speaker

Dr. Gabriel Landini obtained his Doctor in Odontology degree from the Republic University in Montevideo (Uruguay) in 1983 and his PhD in Oral Pathology from Kagoshima University (Japan) in 1991.
He is currently a Reader in Oral Pathology and Head of the Oral Pathology Teaching Speciality at School of Dentistry, The University of Birmingham, UK.
His main research interests are image processing and fractal geometry applied to the analysis and quantification of biological patterns, in particular, invasive patterns of cancer.

Abstract

This work presents a number of methods to facilitate the characterisation of the spatial arrangement of objects (cells) in a tissue as observed in 2D histological sections. The approach is based on an algorithmic segmentation of tissue space into discrete binary cells associated with the position of the cell nuclei. The segmentation is done by means of greyscale reconstruction to extract cell nuclei markers (as regional minima in the image) followed by the so-called watershed transform to partition the rest of the tissue into exclusive basins of attraction (the cells) associated with those markers. Once the cell elements have been identified, user-defined local neighbourhoods (a local set of cells) are used to determine the extent of the relations to be sought. The relations between the cells within these neighbourhoods are characterised using several Graph Theory networks constructed over nodes defined by the centres of mass of the cells. The cell segmentation also allows to quantify other spatial features in the tissues which are difficult to measure – for example defining and labelling tissue layers and the cell orientation within a layer. These features can conveniently extracted after a minimal amount of user-defined landmarks are provided by the operator. The methodology is illustrated with examples of normal, precancerous and cancerous epithelia of the mouth, and with the epithelial lining in different types of odontogenic cysts.

© Luxembourg Institute of Science and Technology | Legal Notice