# Kidney segmentation in CT sequences using SKFCM and improved GrowCut algorithm

- Hong Song
^{1}, - Wei Kang
^{1}, - Qian Zhang
^{2}and - Shuliang Wang
^{1}Email author

**Published: **1 September 2015

## Abstract

### Background

Organ segmentation is an important step in computer-aided diagnosis and pathology detection. Accurate kidney segmentation in abdominal computed tomography (CT) sequences is an essential and crucial task for surgical planning and navigation in kidney tumor ablation. However, kidney segmentation in CT is a substantially challenging work because the intensity values of kidney parenchyma are similar to those of adjacent structures.

### Results

In this paper, a coarse-to-fine method was applied to segment kidney from CT images, which consists two stages including rough segmentation and refined segmentation. The rough segmentation is based on a kernel fuzzy C-means algorithm with spatial information (SKFCM) algorithm and the refined segmentation is implemented with improved GrowCut (IGC) algorithm. The SKFCM algorithm introduces a kernel function and spatial constraint into fuzzy c-means clustering (FCM) algorithm. The IGC algorithm makes good use of the continuity of CT sequences in space which can automatically generate the seed labels and improve the efficiency of segmentation. The experimental results performed on the whole dataset of abdominal CT images have shown that the proposed method is accurate and efficient. The method provides a sensitivity of 95.46% with specificity of 99.82% and performs better than other related methods.

### Conclusions

Our method achieves high accuracy in kidney segmentation and considerably reduces the time and labor required for contour delineation. In addition, the method can be expanded to 3D segmentation directly without modification.

## Keywords

## Background

Image segmentation is one of most important issues in medical technology, which assists physicians in various aspects, such as analysis and diagnosis of different diseases, the study of anatomical structure, making treatment planning [1]. With the increase of CT images in the diagnosis and treatment of diseases, segmentation of human organs from CT images is a prerequisite step in the precise treatment planning. However, different tissues has different sizes and shapes across individuals and the gray scale similarity between kidney and its neighboring tissues, such as liver and spleen. Therefore kidney segmentation is a challenging work.

Many approaches of kidney segmentation have been developed over the recent years, including deformable model, clustering-based method, region growing and knowledge-based method. Tsagaan and Shimizu proposed a deformable model for automatic kidney segmentation which is represented by the grey level appearance of kidney and its statistical information of the shape [2, 3]. Clustering method is a kind of unsupervised learning. So the segmentation methods based on it do not need training sample data, they form clusters of data by grouping pixels [4]. Lin developed an automatic method based on an adaptive region growing method to extract kidney within a region of interest (ROI). However this method mainly depended on the assumption of homogeneity of image intensity, so it is not suitable for the images that have large variation of intensity in the kidney region. Knowledge-based method makes use of the sample data for computing the extracting region, which needs computationally intensive work. Spiegel developed an algorithm based on 3D active shape model (ASM) [5]. Khalifa proposed a level-set method which combined a probabilistic shape prior and a novel stochastic function [6]. Region growing method is sensitive to the seed point location.

In the last decades, fuzzy segmentation methods, especially the fuzzy c-means algorithm (FCM) [7], have been widely used in the field of image segmentation [8]. There are many improved algorithms based on FCM. Zhang et al [9] proposed a kernel-based fuzzy c-means (KFCM) algorithm which has stronger noise immunity and clustering ability. In KFCM algorithm, a kernel-induced metric replaces the original Euclidean norm metric of FCM. In [10, 11], FCM with spatial contextual information (FCM_S) is an effective image segmentation algorithm. Although the contextual information can raise its insensitivity to noise, it still lacks enough robustness to noise and outliers. To overcome these problems, S. Chen et al [12] proposed a novel KFCM algorithm which introduces a spatial constraint (SKFCM). The SKFCM algorithm was used to segment brain and tumor from MR images successfully [12–14].

Cellular automaton (CA) [15, 16] is a nonlinear dynamic model which discrete in time and space and realizes a complex calculation by simple rules. The image processing methods based on cellular automata were used widely, including edge detection, segmentation and denoising. In 2006, Vladimir and Vadim [17] proposed the "GrowCut" algorithm which is an interactive segmentation method and solves pixel labeling task based on cellular automaton. Given some user-labeled points, the rest of the image is segmented automatically by a cellular automaton. The labeling process is iterative. Users can observe the segmentation evolution and guide the algorithm with human input where the segmentation is difficult to compute. The most common application of GrowCut algorithm is segmentation of brain tumors from MR images [17–19].

In this paper, a new coarse-to-fine method is proposed for kidney segmentation. It is a hierarchical segmentation framework combining SKFCM and IGC algorithm for the kidneys segmentation from abdominal CT images. In rough segmentation stage, SKFCM algorithm is better in segmenting images corrupted by noise than FCM algorithm. SKFCM adopts a kernel-induced metric in the data space to replace the original Euclidean norm metric in FCM, so it is a more robust clustering approach. The proposed IGC algorithm is used to refine the rough segmentation result. Due to the IGC algorithm makes good use of the continuity of CT sequences in space; it can generate both object and background seed labels automatically. The IGC algorithm can reduce a lot of interactive time and improve the efficiency of segmentation.

## Methods

### A. Preprocessing

### B. Rough Segmentation with SKFCM

The above algorithm can be summarized in the following steps.

Step 1: Fix the number c of these centroids and select initial class centroids and set $\epsilon >0$ to a very small value.

Step 2: Compute the mean filtered image.

Step 3: Update the partition matrix using (3).

Step 4: Update the centroids using (4)

The purpose of this subsection is to get the rough contour of kidney in the CT images. Owing to the gray scales of kidney is similar to its neighboring tissues, it is important to identify which part belongs to the kidney. To solve this problem, a slice which has the largest contour in the whole dataset is cropped by a rectangle manually. The rectangle must enclose the kidney and its size should be as small as possible, so that it can increase the processing speed and the segmentation accuracy. Other slices are automatically cropped as described in the following. The optimal cluster number is 4 which is determined by experiments. The rough segmentation includes six steps.

Step 1: The cropped image is the input of SKFCM algorithm, and then each pixel in the cropped region is clustered into different clusters.

Step 2: The number of pixels in each cluster is calculated and the cluster which contains maximum pixels is extracted.

Step 3: The largest connected region is extracted to be the candidate kidney region.

Step 4: There are some holes inside the kidney because some vessels are rejected in the processing of fuzzy clustering. Therefore this step is to fill holes.

Step 5: The kidney contour is smoothed by morphological operations.

Step 6: Through the above steps, the mask of candidate kidney region is gotten. In order to realize the continuous segmentation, the minimum bounding rectangle (MBR) of the mask is calculated. Then the MBR is extended about 10 pixels so that we can get a new rectangle. This new rectangle is used to crop the next slice of CT sequeces.

*n*is the slice number in the whole datasets.

### C. Refined Segmentation with IGC Algorithm

#### 1) The traditional GrowCut

GrowCut algorithm is an interactive segmentation method and solves pixel labeling task based on cellular automaton.

A cellular automaton (CA) is defined as a triplet $\mathsf{\text{A}}=\left(\mathsf{\text{S}},\mathsf{\text{N}},\delta \right)$, where S is a set of non-empty state, N is the neighborhood system and $\delta :{\mathsf{\text{S}}}^{\mathsf{\text{N}}}\to \mathsf{\text{S}}$ defines the state transition rule of cells at time t+1 based on the states of neighbor cells at time t. The Moore von (8-connected) and Neumann (4-connected) neighborhoods are two commonly used neighbor systems. The state of each cell is also a tri-plet ${\mathsf{\text{S}}}_{\mathsf{\text{p}}}=\left({\mathsf{\text{l}}}_{\mathsf{\text{p}}},{\theta}_{\mathsf{\text{p}}},{\stackrel{\u20d7}{\mathsf{\text{C}}}}_{\mathsf{\text{p}}}\right)$, where ${\mathsf{\text{l}}}_{\mathsf{\text{p}}}$ is the label of this cell, ${\theta}_{\mathsf{\text{p}}}$ is the strength of this cell which ranges from 0 to 1, and ${\stackrel{\u20d7}{\mathsf{\text{C}}}}_{\mathsf{\text{p}}}$ is the feature vector that its value is image intensity.

where ${\mathsf{\text{l}}}_{\mathsf{\text{p}}}$ is the intensity value of each pixel. As the segmentation algorithm proceeds, all pixels in this image are assigned to one of possible labels.

Before starting the segmentation, user should input an initial label matrix manually. The label matrix has a same size with the original image. In the label matrix, there are two kinds of marked points, one is foreground seed point whose label is ${\mathsf{\text{l}}}_{\mathsf{\text{p}}}=1$, and the other is background seed point whose label is ${\mathsf{\text{l}}}_{\mathsf{\text{p}}}=-1$. The original strength of these two kinds of marked points is ${\theta}_{\mathsf{\text{p}}}=1$. Apart from these two kinds of marked points, the label of the remainder of points is ${\mathsf{\text{l}}}_{\mathsf{\text{p}}}=0$. After all the initial operations have been done, the iteration segmentation runs until the label matrix does not change. Finally, the label value of object region is 1 and the label value of background is -1. The iterative process of labels ${\mathsf{\text{l}}}_{\mathsf{\text{p}}}$ and strength ${\theta}_{\mathsf{\text{p}}}$ at time t+1 is shown as follows,

State transition of CA

// For each cell...

for $\forall \mathsf{\text{p}}\in \mathsf{\text{P}}$

// Copy the previous state

${\mathsf{\text{l}}}_{\mathsf{\text{p}}}^{\mathsf{\text{t}}+1}={\mathsf{\text{l}}}_{\mathsf{\text{p}}}^{\mathsf{\text{t}}}$;

${\theta}_{\mathsf{\text{p}}}^{\mathsf{\text{t}}+1}={\theta}_{\mathsf{\text{p}}}^{\mathsf{\text{t}}}$;

// Neighbors try to attack current cell

for $\forall \mathsf{\text{q}}\in \mathsf{\text{N}}\left(\mathsf{\text{p}}\right)$

if $\mathsf{\text{g}}\left(\left|\right|{\stackrel{\u20d7}{\mathsf{\text{C}}}}_{\mathsf{\text{p}}}-{\stackrel{\u20d7}{\mathsf{\text{C}}}}_{\mathsf{\text{q}}}|{|}_{2}\right)\cdot {\theta}_{\mathsf{\text{q}}}^{\mathsf{\text{t}}}>{\theta}_{\mathsf{\text{p}}}^{\mathsf{\text{t}}}$

${\mathsf{\text{l}}}_{\mathsf{\text{p}}}^{\mathsf{\text{t}}+1}={\mathsf{\text{l}}}_{\mathsf{\text{q}}}^{\mathsf{\text{t}}}$;

${\theta}_{\mathsf{\text{p}}}^{\mathsf{\text{t}}+1}=\mathsf{\text{g}}\left(\mathsf{\text{||}}{\stackrel{\u20d7}{\mathsf{\text{C}}}}_{\mathsf{\text{p}}}-{\stackrel{\u20d7}{\mathsf{\text{C}}}}_{\mathsf{\text{q}}}||{}_{2}\right)\cdot {\theta}_{\mathsf{\text{q}}}^{\mathsf{\text{t}}}$;

end if

end for

end for

#### 2) The Improved GrowCut

Although GrowCut algorithm is simple and precise, it is an interactive segmentation algorithm. For some complex images, just once interactive operation cannot achieve satisfactory results, and both foreground seed points and background seed points also should be selected carefully. In order to avoid multi-interaction, we propose an improved GrowCut algorithm which can generate seed labels automatically.

Firstly, we should get the edge points of kidney contour.

Secondly, calculate the altitude difference between two edge points. Figure 6(b) and 6(c) show the altitude difference (d_{y}) at point P_{m} and P_{n} respectively.

Thirdly, there are three threshold values T_{h}, T_{1}, T_{2} to control the process of generating seed points. The seed points will be located in the vertical direction if d_{y} is less than T_{h}, otherwise they will be located in the horizontal direction. Taking Figure 6(e) as an example, the foreground seed point is located at the bottom of P_{m} and the background seed point is located at the top of P_{m} , because d_{y} is less than T_{h}. In Figure 6(f), the foreground seed point is located at the right of P_{n} and the background seed point is located at the left of P_{n}, because d_{y} is greater than T_{h}. All foreground seed points are located inside of the kidney contour and the distance between them and edge points are T_{1} pixel. All background seed points are located outside of the kidney contour and the distance between them and edge points are T_{2} pixel. In Figure 6(e) and 6(f), the red points are denoted as foreground seed points and the green points are denoted as background seed points.

Finally, according to step 2 and 3, we can get both foreground and background seed points of each edge point

### D. Post-processing

Some segmentation results of IGC and SKFCM have rough boundaries. To achieve a smoother contour of kidney, a post-processing method based on morphological operations is needed. The most common morphological operations are dilation and erosion.

## Results and evaluation

The segmentation experiments and performance evaluation were carried on three groups of abdominal CT images. The parameters of abdominal CT images for scanning were 120.0 KV and 297.0 mA. The pixel spacing was 0.683594 mm, the slice thickness was 1.0 mm and the spacing between slices was 0.5 mm. The number of slices ranged from 217 to 320. Each slice of these three datasets had a spatial resolution of 512 × 512 pixels. Both SKFCM and IGC algorithm were implemented on MATLAB R2013b. All experiments were implemented on the computer with Pentium Dual - Core CPU (2.80GHz) and 2GB memory.

The definition of TP FP FN and TN.

Result by manual segmentation | |||
---|---|---|---|

Positive | Negative | ||

Result by segmentation algorithm | Positive | TP | FP |

Negative | FN | TN |

The evaluation of different algorithms.

Methods | Accuracy (%) | Overlap (%) | NOI | TOGSP (s) | |
---|---|---|---|---|---|

Data 1 | TGC | 99.69 | 86.61 | 1 | 29.14 |

IGC | 99.64 | 85.11 | 0 | 0.50 | |

Data 2 | TGC | 99.59 | 80.71 | 2 | 25.23 |

IGC | 99.62 | 82.57 | 0 | 0.51 | |

Data 3 | TGC | 99.69 | 86.59 | 1 | 38.28 |

IGC | 99.72 | 88.08 | 0 | 0.50 |

## Conclusion

In this paper, we proposed a new coarse-to-fine method that combines SKFCM and the improved GrowCut algorithm to extract the kidneys for the abdominal CT images. The method was tested on the whole dataset of abdominal CT images. Experimental results have been shown visually and achieve reasonable consistency. The performance evaluation of segmentation results demonstrates that our kidney segmentation method is accurate and efficient. There are two key contributions. First, SKFCM algorithm is used to implement rough kidney segmentation successfully due to its strong clustering ability and robust noise immunity. Second, the traditional GrowCut algorithm has been improved. The improved GrowCut algorithm can generate seed labels automatically instead of inputting seed labels by users, so that it can reduce interactive time and improve the efficiency of segmentation. The segmentation results of our method can be used to diagnose the kidney diseases and make treatment planning. They are also the foundation of 3D visualization.

## Publication funding from grants

The publication charges for this article were funded by the National Natural Science Foundation of China grant 61240010.

## Declarations

### Acknowledgements

This work has been funded by the National Natural Science Foundation of China grants 61240010, 61173061 and 71201120.

This article has been published as part of *BMC Systems Biology* Volume 9 Supplement 5, 2015: Selected articles from the IEE International Conference on Bioinformatics and Biomedicine (BIBM 2014): Bioinformatics. The full contents of the supplement are available online at http://www.biomedcentral.com/bmcsystbiol/supplements/9/S5.

## Authors’ Affiliations

## References

- Pham DL, Xu C, Prince JL: Current methods in medical image segmentation. Annu Rev Biomed Eng. 2000, 2: 315-338. 10.1146/annurev.bioeng.2.1.315.View ArticlePubMedGoogle Scholar
- Tsagaan B, Shimizu A, Kobatake H, Miyakawa K, Hanzawa Y: Segmentation of kidney by using a deformable model. in Int Conf Image Processing. 2001, Thessaloniki, Greece, 3: 1059-1062.Google Scholar
- Tsagaan B, Shimizu A, Kobatake H, Miyakawa K: An automated segmentation method of kidney using statistical information. in Proc Medical Image Computing and Computer Assisted Intervention. 2002, 1: 556-563.Google Scholar
- Pham DL, Xu C, Prince JL: Current methods in medical image segmentation 1[J]. Annual review of biomedical engineering. 2000, 2 (1): 315-337. 10.1146/annurev.bioeng.2.1.315.View ArticlePubMedGoogle Scholar
- Spiegel M, Hahn DA, Daum V, Wasza J, Hornegger J: Segmentation of kidneys using a new active shape model generation technique based on non-rigid image registration. Comput Med Imag Grap. 2009, 33 (1): 29-39. 10.1016/j.compmedimag.2008.10.002.View ArticleGoogle Scholar
- Khalifa F, Elnakib A, Beache GM, et al: 3D kidney segmentation from CT images using a level set approach guided by a novel stochastic speed function. in Proc MICCAI 2011. 2011, Springer Berlin Heidelberg, 587-594.Google Scholar
- Bezdek JC: Pattern Recognition with Fuzzy Objective Function Algorithms. 1981, New York: PlenumView ArticleGoogle Scholar
- Udupa JK, Samarasekera S: Fuzzy connectedness and objectdefinition: theory, algorithm and applications in image segmentation. Graph Models Image Process. 1996, 58 (3): 246-261. 10.1006/gmip.1996.0021.View ArticleGoogle Scholar
- Zhang DQ, Chen S, et al: A novel kernelized fuzzy c-means algorithm with application in medical image segmentation. artificial intelligence in medicine. 2004, 32: 37-50. 10.1016/j.artmed.2004.01.012.View ArticlePubMedGoogle Scholar
- Li BN, Chui C, Chang S, et al: Integrating spatial fuzzy clustering with level set methods for automated medical image segmentation. Computers in Biology and Medicine. 2011, 41: 1-10. 10.1016/j.compbiomed.2010.10.007.View ArticlePubMedGoogle Scholar
- UmaDevi N, Poongodi R: Integration of Spatial Fuzzy Clustering with Level Set for Efficient Image Segmentation. International Journal of Computer Science Communication Networks. 3: 296-301.Google Scholar
- Chen S, Zhang D: Robust image segmentation using FCM with spatial constraints based on new kernel-induced distance measure. Systems, Man, and Cybernetics, Part B: Cybernetics, IEEE Transactions. 2004, 34: 1907-1916. 10.1109/TSMCB.2004.831165.View ArticleGoogle Scholar
- Kannan SR, Ramathilagam S, Sathya A: Effective fuzzy c-means based kernel function in segmenting medical images. computers in biology and medicine. 2010, 40: 572-579. 10.1016/j.compbiomed.2010.04.001.View ArticlePubMedGoogle Scholar
- Wu Y, Hou W, Wu S: Brain MRI segmentation using KFCM and Chan-Vese model. Transactions of Tianjin University. 2011, 17: 215-219. 10.1007/s12209-011-1578-4.View ArticleGoogle Scholar
- Vezhnevets V, Konouchine V: GrowCut: Interactive multi-label ND image segmentation by cellular automata. Proc of Graphicon. 2005, 150-156.Google Scholar
- Ghosh P, Antani SK, Long LR, et al: Unsupervised Grow-Cut: cellular automata-based medical image segmentation. Healthcare Informatics, Imaging and Systems Biology (HISB), 2011 First IEEE International Conference on IEEE. 2011, 40-47.View ArticleGoogle Scholar
- Hamamci A, Unal G, Kucuk N, et al: Cellular automata segmentation of brain tumors on post contrast MR images. Medical Image Computing and Computer-Assisted Intervention-MICCAI 2010. 2010, Springer Berlin Heidelberg, 137-146.View ArticleGoogle Scholar
- Yamasaki T, Chen T, Yagi M, et al: GrowCut-based fast tumor segmentation for 3D magnetic resonance images. SPIE Medical Imaging. International Society for Optics and Photonics. 2012, 831434-831434-8.Google Scholar
- Kikinis R, Pieper S: 3D Slicer as a tool for interactive brain tumor segmentation. Engineering in Medicine and Biology Society, EMBC, 2011 Annual International Conference of the IEEE. IEEE. 2011, 6982-6984.View ArticleGoogle Scholar
- Hu YC, Grossberg MD, Wu A, et al: Interactive semiautomatic contour delineation using statistical conditional random fields framework[J]. Medical physics. 2012, 39 (7): 4547-4558. 10.1118/1.4728979.PubMed CentralView ArticlePubMedGoogle Scholar
- Belgherbi A, Hadjidj I, Bessaid A: Morphological segmentation of the kidneys from abdominal CT images[J]. Journal of Mechanics in Medicine and Biology. 2014, 14 (5):Google Scholar

## Copyright

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.