Image Segmentation, Registration and Characterization in R with SimpleITK

Abstract

Many types of medical and scientific experiments acquire raw data in the form of images. Various forms of image processing and image analysis are used to transform the raw image data into quantitative measures that are the basis of subsequent statistical analysis. In this article we describe the SimpleITK R package. SimpleITK is a simplified interface to the insight segmentation and registration toolkit (ITK). ITK is an open source C++ toolkit that has been actively developed over the past 18 years and is widely used by the medical image analysis community. SimpleITK provides packages for many interpreter environments, including R. Currently, it includes several hundred classes for image analysis including a wide range of image input and output, filtering operations, and higher level components for segmentation and registration. Using SimpleITK, development of complex combinations of image and statistical analysis procedures is feasible. This article includes several examples of computational image analysis tasks implemented using SimpleITK, including spherical marker localization, multi-modal image registration, segmentation evaluation, and cell image analysis.

Keywords: R; image processing; image registration; image segmentation; medical imaging.

Figure 1:

Cropped cone-beam CT volume of a metallic sphere (top row) and the result of performing 3D edge detection on the volumetric data (bottom row). Original image intensity values have been mapped to [0, 255] for display purposes. The filled circle edge images at both ends (left, right) of the sphere highlight the fact that the operation is indeed carried out in 3D. If performed on a slice by slice manner all edge images would result in empty circles.

Figure 2:

Five slices extracted from the center of each volume, from top to bottom: original CT image, our fixed image; original MR image, our moving image; fused image after initial spatial alignment of images; fused image after registration.

Figure 3:

Similarity metric changes during rigid registration.

Figure 4:

Deriving a reference segmentation from multiple raters using the STAPLE algorithm: (top row) manual segmentations performed by three radiologists; (bottom row) two additional segmentations derived from the expert segmentations to illustrate the effects of over-segmentation and segmentation with outliers. Last image is the derived reference segmentation obtained by the STAPLE algorithm.

Figure 5:

Comparison of raters using various overlap measures.

Figure 6:

Confocal microscope image of cells stained with Ph3 (red), Ki67 (green) and DAPI (blue).

Figure 7:

Stages of segmentation for the DAPI channel. Touching cells in the mid-right side of the image are separated by the splitting stage.

Figure 8:

Segmentation using Li thresholding of Ph3 and Ki67 channels.

Figure 9:

Ph3 segmentation refinement using geodesic reconstruction – note the increase in size of nuclei at the top and right of image.

Figure 10:

Histograms of cell nucleus area by stain type.