Journal of Visual Communication and Image Representation
Automatic liver segmentation for volume measurement in CT Images
Introduction
The liver cancer is one of the most common internal malignancies worldwide. The hepatocellular carcinoma is common in Asia and metastasis is common in the West. The liver cancer is also one of the leading death causes. Currently, the confirmed diagnosis used widely for the liver cancer is needle biopsy. The needle biopsy, however, is an invasive technique and generally not recommended [1]. Therefore, computed tomography (CT) and magnetic resonance imaging (MRI) have been identified as accurate non-invasive imaging modalities in the diagnosis of the liver cancer. These medical images are interpreted by radiologists. However, image interpretation by human beings is often limited due to the non-systematic search patterns of themselves, the presence of structural noise in the image, and the presentation of complex disease states requiring the integration of vast amount of image data and clinical information.
Recently, computer-aided diagnosis (CAD), defined as a diagnosis introduced by a radiologist who uses the output from a computerized analysis of medical images as a “second opinion” in detecting lesions, assessing extent of disease, and making diagnostic decisions, is being used to improve the interpretation components of medical imaging [2], [3]. In addition, computer-aided surgery (CAS) that is the future technology in surgery is performed on computerized surgical planning and image-guided surgery by analyzing region-of-interest (ROI) in the medical image. Volume measurement is also of major importance in different fields of medical imaging where physicians need some quantitative assessments for surgical decisions.
Research in CAD for both mammogram and chest radiographs is rapidly growing; however, CAD research for liver cancer is to be insufficient because the liver segmentation that plays an important role for CAD is difficult. This is mainly due to the two following facts. The first one is the proximity of the liver and other organs or muscles with the similar intensity. It makes difficult to resolve by observation of intensity discontinuity alone since partial-volume effects (PVE) cause the discontinuity to weaken where the structures touch. The second one is the variation in both shape and scale across patients even on the same patient [4].
There are many approaches for image segmentation, such as feature thresholding, contour based techniques, region based techniques, clustering, and template matching. Each of these approaches has its advantages and disadvantages in terms of applicability, suitability, performance, and computational cost [5]. Particularly, no one who did not consider above characteristics of the abdominal CT image can meet desirable results on liver segmentation. In addition, the traditional method of getting volume of the liver is to perform a by-hand 2D segmentation of parallel cross-sectional CT slices and to multiply all voxels of the stacked slices by their size while the procedure is often time consuming and non-systematic [6]. Therefore, to address above problems, we present an automatic liver segmentation algorithm in abdominal CT images using the combination of region-based and contour-based approaches. Our algorithm exploits both medical priori knowledge, for example, the general shape, location, and gray level of the liver, and deformable contour method using labeling-based search algorithm. Finally, total liver volumes were calculated from segmented areas of the liver to evaluate the patients for entire or partial liver transplantation and CAS.
This paper is organized as the following. In Section 2, we propose a new segmentation algorithm applicable to CT image, and we describe volume measurement in Section 3. After experimental results and analysis are presented in Section 4, we conclude the paper in Section 5.
Section snippets
Automatic liver segmentation
Mainly, the liver is approximated to muscle and gastrointestinal tract. Since adjacent organs have similar intensity with the liver as shown in Fig. 1, a direct liver-extraction approach without preprocessing may also extract undesirable boundaries resulting from its adjacent organs as fault positive/negative errors [1]. To cope with the problem, we present a new segmentation scheme with three stages. The first stage is image simplification as preprocessing; the second stage detects a search
Volume measurement based on the segmentation
For volume measurement of the liver, we calculate the volume by using thickness and interval information of the slice and size of the pixel. The following equation is for volume measurement using the previous segmented liver region [12].where N is the number of the slice including the segmented liver region, Si is the slice number, D is the interval of the slice, Lp is the number of the pixel in the segmented liver region, and X,Y are the size
Experimental results and discussion
We experimented several samples with various shapes and irregular texture of 10 patients. All of the used samples are contrast-enhanced abdominal CT images of venous phase with 5 mm interval.
Each sample shows the result of each process of the proposed algorithm, six images in total, such as threshold image, the result of morphological filtering, search range, initial liver boundary, gradient-label map, and the final result through Fig. 14 to Fig. 15.
Fig. 14, Fig. 15B depict the multilevel
Conclusions
In this paper, we have proposed an automatic segmentation algorithm for volume measurement of the liver using a priori knowledge and the deformable contour method based on morphological filtering. The proposed algorithm using multilevel thresholding based on the analysis of the intensity distribution within ROI decrease the needless computation time and efforts by reducing the regions of the other organs and tissues. In addition, multiscale morphological filtering using region-labeling and
Acknowledgments
This work was supported by the Korea Research Foundation Grant (No. M07-2004-000-10140-0), in part by the Ministry of Information and Communication (MIC) through the Realistic Bradcasting Research Center (RBRC) at Gwangju Institute of Science and Technology (GIST), and in part by the Ministry of Education (MOE) through the Brain Korea 21 (BK21) project.
References (12)
- et al.
An automatic diagnosis system for CT liver image classification
IEEE Trans. Biomed. Eng.
(1998) - et al.
Automatic segmentation of the liver in CT images using the watershed algorithm based on morphological filtering
SPIE Proc.
(2004) - et al.
Guest editorial: computer-aided diagnosis in medical imaging
IEEE Trans. Med. Imaging
(2001) - et al.
Medical image segmentation using analysis of isolable-contour maps
IEEE Trans. Med. Imaging
(2000) - et al.
Multiscale morphological segmentation of gray-scale images
IEEE Trans. Image Process.
(2003) - B. Reitinger, A. Bornik, R. Beichel, Efficient volume measurement using voxelization, in: Proceedings of Spring...
Cited by (127)
Deeply self-supervised contour embedded neural network applied to liver segmentation
2020, Computer Methods and Programs in BiomedicineA robust multi-variability model based liver segmentation algorithm for CT-scan and MRI modalities
2019, Computerized Medical Imaging and GraphicsAutomatic segmentation methods for liver and hepatic vessels from CT and MRI volumes, applied to the Couinaud scheme
2019, Computers in Biology and MedicineSd-net: a semi-supervised double-cooperative network for liver segmentation from computed tomography (CT) images
2024, Journal of Cancer Research and Clinical OncologyDeep Federated Machine Learning-Based Optimization Methods for Liver Tumor Diagnosis: A Review
2023, Archives of Computational Methods in Engineering