Available Technology

Software for Accurate Segmentation of Cell Nuclei in Breast Tissue

Automatic segmentation of cell nuclei is critical in several high-throughput cytometry and pathology applications (1), such as spatial analysis of genetic loci by fluorescencein situhybridization ("FISH"), whereas manual segmentation is laborious (2). Current automated segmentation methods have varying performance in the presence of distortions introduced during sample preparation, non-uniform illumination, clustering of the individual objects of interest (cells or cell nuclei), and seldom assess boundary accuracy.

Researchers at the National Cancer Institute-Frederick, NIH, have developed an automatic algorithm to segment cell nuclei (3) and FISH signals from two-dimensional images of breast tissue. This automated system integrates a series of advanced image processing methods to overcome the delays inherent to current manual methods for segmenting (delineating) individual cell nuclei in tissue samples. The system automatically selects a subset of nuclei that with high likelihood are accurately segmented. This system has been validated using both simulated and actual datasets that have been accurately analyzed by manual methods. The system generalizes to independent analysis of many spatial parameters useful for studying spatial gene positioning in interphase nuclei, and potentially has a wide range of diagnostic pathology, cytological and high throughput screening applications.

Patent Abstract: 

Automatic segmentation of cell nuclei is critical in several high-throughput cytometry and pathology applications (1), such as spatial analysis of genetic loci by fluorescencein situhybridization ("FISH"), whereas manual segmentation is laborious (2). Current automated segmentation methods have varying performance in the presence of distortions introduced during sample preparation, non-uniform illumination, clustering of the individual objects of interest (cells or cell nuclei), and seldom assess boundary accuracy.

Researchers at the National Cancer Institute-Frederick, NIH, have developed an automatic algorithm to segment cell nuclei (3) and FISH signals from two-dimensional images of breast tissue. This automated system integrates a series of advanced image processing methods to overcome the delays inherent to current manual methods for segmenting (delineating) individual cell nuclei in tissue samples. The system automatically selects a subset of nuclei that with high likelihood are accurately segmented. This system has been validated using both simulated and actual datasets that have been accurately analyzed by manual methods. The system generalizes to independent analysis of many spatial parameters useful for studying spatial gene positioning in interphase nuclei, and potentially has a wide range of diagnostic pathology, cytological and high throughput screening applications.

Benefits 
Automatic -Efficient, robust and effective in extracting individual nuclei with FISH labels -Facilitates reproducible and unbiased spatial analysis of DNA sequences in interphase nuclei
Inventors: 

Kaustav Nandy

Lab Representatives
Share to Facebook Share to Twitter Share to Google Plus Share to Linkedin