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.

Figure 1

The flowchart of HNCut-CGAC model shown in the context of gland segmentation on prostate histopathology imagery.

Figure 2

(a) Original color image of needle core biopsy histopathology image, and corresponding (b) color gradient and (c) gray scale gradient obtained after converting the color image in (a) to its gray scale representation with the MATLAB function rgb2gray.

Figure 3

The histogram for segmentation accuracy evaluation of HNCut-CGAC model with color swatch S₀ over 196 images are plotted. The plots reflect the number of studies (y-axis) for which (a) Overlap, (b) Sensitivity, (c) Specificity, and (d) Positive Predictive Value (PPV) values were below certain number(x-axis).

Figure 4

A histogram plot showing the distribution in the values of MAD, for the HNCut-CGAC model using swatches (a) S₀ and (b) S₅ across 196 images. Note that there are no significant differences in MAD values for the two different swatches.

Figure 5

The gland segmentation results (boundaries in green) of HNCut-CGAC, HNCut-GAC, CGAC, RD, and HAC models for a whole-slide needle core biopsy (a). (c) and (d) are two different patches (I) and (II) from the segmentation result (b) of the HNCut-CGAC model which have been magnified to show gland details. (e) and (f) are two magnified patches selected from the same location (I, II) (b) and showing the segmentation result of the HNCut-GAC model. (g) and (h) show corresponding results for the CGAC model. (i) and (j) show corresponding results for the RD model. (k) and (l) show corresponding results for the HAC model.

Figure 6

The gland segmentation results (boundaries in green) of HNCut-CGAC, HNCut-GAC, CGAC, RD, and HAC models from a whole-slide needle core biopsy (a) in study 2. (c) and (d) are two different patches (I) and (II) from the segmentation result (b) of the HNCut-CGAC model which have been magnified to show gland details. (e) and (f) are two magnified patches selected from the same location (I, II) (b) and showing the segmentation result of the HNCut-GAC model. (g) and (h) show corresponding results for the CGAC model. (i) and (j) show corresponding results for the RD model. (k) and (l) show corresponding results for the HAC model.