A General Approach to Liver Lesion Segmentation in CT Images
Li Caoa,b, Jayaram K. Udupa*a, Dewey Odhnera, Lidong Huanga,c, Yubing Tonga, Drew A. Torigiana
a
Medical Image Processing Group, Department of Radiology, University of Pennsylvania,
Philadelphia, PA; bScience and Technology on Multi-spectral Information Processing Laboratory,
School of Automation, Huazhong University of Science and Technology, Wuhan 430074, China;
c
College of Information and Communication Engineering, Beihang University, Beijing 100083,
China
ABSTRACT
Lesion segmentation has remained a challenge in different body regions. Generalizability is lacking in published
methods as variability in results is common, even for a given organ and modality, such that it becomes difficult to
establish standardized methods of disease quantification and reporting. This paper makes an attempt at a generalizable
method based on classifying lesions along with their background into groups using clinically used visual attributes.
Using an Iterative Relative Fuzzy Connectedness (IRFC) delineation engine, the ideas are implemented for the task of
liver lesion segmentation in computed tomography (CT) images. For lesion groups with the same background properties,
a few subjects are chosen as the training set to obtain the optimal IRFC parameters for the background tissue
components. For lesion groups with similar foreground properties, optimal foreground parameters for IRFC are set as the
median intensity value of the training lesion subset. To segment liver lesions belonging to a certain group, the devised
method requires manual loading of the corresponding parameters, and correct setting of the foreground and background
seeds. The segmentation is then completed in seconds. Segmentation accuracy and repeatability with respect to seed
specification are evaluated. Accuracy is assessed by the assignment of a delineation quality score (DQS) to each case.
Inter-operator repeatability is assessed by the difference between segmentations carried out independently by two
operators. Experiments on 80 liver lesion cases show that the proposed method achieves a mean DQS score of 4.03 and
inter-operator repeatability of 92.3%.
Keywords: computed tomography (CT), liver lesions, segmentation, fuzzy connectedness
1. INTRODUCTION
As liver cancer is among five cancers causing the most deaths worldwide, and metastatic lesions are common in the liver,
a precise analysis of liver lesions is needed to exactly evaluate lesional properties and burden and to guide possible
applicable therapies1. However, to segment liver lesions in computed tomography (CT) images is a challenging task due
to a number of factors, including the irregularity of lesion shape, the inhomogeneity of lesion tissues, and the similarities
between the image characteristics of lesions with a variable number of surrounding normal tissues. A common approach
is for a trained operator to manually segment the lesions, which is time-consuming and subjective due to the differences
of skill, expertise, and experience among operators2. Since the diversity and complexity of liver lesions make it
inadequate to perform lesion segmentation automatically3, the semi-automatic approach is more feasible and suitable to
accomplish the task.
In the past ten years, many semi-automatic methods1-6 have been developed for segmentation of various types of liver
lesions from CT scans by applying some basic image segmentation algorithms, such as thresholding, region growing,
level set, and watershed. By combining adaptive thresholding with model-based morphological processing, Moltz et al.4
present a semi-automatic algorithm for segmentation of liver metastases in CT scans. By using the watershed algorithm
first for liver contour extraction, and region growing and level-set based surface smoothing methods for segmentation
refinement, a semi-automatic technique based on a minimum cross-entropy multi-thresholding algorithm1 is developed
for liver tumor segmentation. Based on 2D region growing with knowledge-based constraints, Wong et al.2 propose a
semi-automatic method to segment liver tumors from constituent 2D slices obtained from 3D CT images. Based on a
marker-controlled watershed transformation, a semi-automatic method developed by Yan et al.5 can accurately segment
3D liver metastases in volumetric CT images. In addition, for the segmentation of complicated liver lesions, a local
Medical Imaging 2016: Image-Guided Procedures, Robotic Interventions, and Modeling,
edited by Robert J. Webster III, Ziv R. Yaniv, Proc. of SPIE Vol. 9786, 978623
© 2016 SPIE · CCC code: 1605-7422/16/$18 · doi: 10.1117/12.2217778
Proc. of SPIE Vol. 9786 978623-1
