A high-throughput active contour scheme for segmentation of histopathological imagery

Abstract

In this paper a minimally interactive high-throughput system which employs a color gradient based active contour model for rapid and accurate segmentation of multiple target objects on very large images is presented. While geodesic active contours (GAC) have become very popular tools for image segmentation, they tend to be sensitive to model initialization. A second limitation of GAC models is that the edge detector function typically involves use of gray scale gradients; color images usually being converted to gray scale, prior to gradient computation. For color images, however, the gray scale gradient image results in broken edges and weak boundaries, since the other channels are not exploited in the gradient computation. To cope with these limitations, we present a new GAC model that is driven by an accurate and rapid object initialization scheme; hierarchical normalized cuts (HNCut). HNCut draws its strength from the integration of two powerful segmentation strategies—mean shift clustering and normalized cuts. HNCut involves first defining a color swatch (typically a few pixels) from the object of interest. A multi-scale, mean shift coupled normalized cuts algorithm then rapidly yields an initial accurate detection of all objects in the scene corresponding to the colors in the swatch. This detection result provides the initial contour for a GAC model. The edge-detector function of the GAC model employs a local structure tensor based color gradient, obtained by calculating the local min/max variations contributed from each color channel. We show that the color gradient based edge-detector function results in more prominent boundaries compared to the classical gray scale gradient based function. By integrating the HNCut initialization scheme with color gradient based GAC (CGAC), HNCut-CGAC embodies five unique and novel attributes: 1) Efficiency in segmenting multiple target structures; 2) The ability to segment multiple objects from very large images; 3) Minimal human interaction; 4) Accuracy; and 5) Reproducibility. A quantitative and qualitative comparison of the HNCut-CGAC model against other state of the art active contour schemes (including a Hybrid Active Contour model (Paragios-Deriche) and a region based AC model (Rousson-Deriche)), across 196 digitized prostate histopathology images, suggests that HNCut-CGAC is able to outperform state of the art hybrid and region based AC techniques. Our results show that HNCut-CGAC is computationally efficient and may be easily applied to a variety of different problems and applications.

A Novel Segmentation and Contouring Scheme to Assist Accurate Brain Lesion Classification

“…The contour approaches [13] are widely used in practical exploitation of data from images because they effectively use specific prior information about objects and this makes them inherently efficient algorithms. Furthermore, active contours apply processing algorithms selectively to regions of the image, rather than processing the entire image [8].…”

Spectral embedding based active contour (SEAC) for lesion segmentation on breast dynamic contrast enhanced magnetic resonance imaging

“…However, implementations of the AC model have been developed for multidimensional images. [12][13][14][15] Chan et al 13 presented an extension of the original scalar image-based AC model applied to vector-valued images. Rousson and Deriche 14 also presented a vector-valued active contour.…”

“…Rousson and Deriche 14 also presented a vector-valued active contour. In a recent application to a medical imaging problem, Xu et al 15 developed a tensor gradient-based AC for use with histopathological images by computing the gradient from vectorial images. Xu et al 15 showed that the tensor gradient more completely captured the gradient information in a multichannel image than using a single channel of the image, yielding a more accurate AC scheme.…”

“…Tensor gradients derived from the vectorial image provide stronger gradients for driving the AC model than the corresponding gradient derived from a scalar image. 15 Step 3. Manual initialization of the AC model: A point within the OOI is selected manually, which serves as the initialization for the AC model.…”

“…An AC deforms in order to approximate the border between regions of interest and noninterest. 11,[13][14][15]36 In the simplified case, the energy functional is formulated as the integral of an edge detector function…”