A pulse coupled neural network segmentation algorithm for reflectance confocal images of epithelial tissue.

Meagan A Harris,Bilal H Malik,Andrew N Van,Joey M Jabbour,Kristen C Maitland,Thomas Abraham

doi:10.1371/journal.pone.0122368

Abstract

Automatic segmentation of nuclei in reflectance confocal microscopy images is critical for visualization and rapid quantification of nuclear-to-cytoplasmic ratio, a useful indicator of epithelial precancer. Reflectance confocal microscopy can provide three-dimensional imaging of epithelial tissue in vivo with sub-cellular resolution. Changes in nuclear density or nuclear-to-cytoplasmic ratio as a function of depth obtained from confocal images can be used to determine the presence or stage of epithelial cancers. However, low nuclear to background contrast, low resolution at greater imaging depths, and significant variation in reflectance signal of nuclei complicate segmentation required for quantification of nuclear-to-cytoplasmic ratio. Here, we present an automated segmentation method to segment nuclei in reflectance confocal images using a pulse coupled neural network algorithm, specifically a spiking cortical model, and an artificial neural network classifier. The segmentation algorithm was applied to an image model of nuclei with varying nuclear to background contrast. Greater than 90% of simulated nuclei were detected for contrast of 2.0 or greater. Confocal images of porcine and human oral mucosa were used to evaluate application to epithelial tissue. Segmentation accuracy was assessed using manual segmentation of nuclei as the gold standard.

Highlights

Carcinomas, cancers of epithelial tissues that cover the external and internal surfaces of the body, account for more than 80% of all cancers [1]
The spiking cortical model (SCM) automated segmentation algorithm, including the artificial neural network (ANN) classifier trained on objects in confocal images of epithelium, was first applied to the image model of epithelial tissue to evaluate the limitations of the algorithm
The original simulated images shown in the first column are 1000 × 1000 px2 field of view (FOV) containing 750 objects distributed randomly without overlap

Summary

Introduction

Carcinomas, cancers of epithelial tissues that cover the external and internal surfaces of the body, account for more than 80% of all cancers [1]. Visual or endoscopic examination followed by invasive tissue biopsy and histopathology is the current standard of care for detection and diagnosis of carcinoma. The tissue may not fill the entire imaging FOV. A threshold algorithm is utilized to remove the background of the image leaving the active FOV. This was accomplished by assuming the background is composed of a large contiguous area of dark pixels distinct from the foreground (i.e. the tissue). An area filter was applied to remove the background from the image. After removal of the background pixels, the area of the foreground was calculated for use in the NCR calculation

Methods

Results

Conclusion