Liver CT image processing: A short introduction of the technical elements

Abstract

In this paper, we describe the technical aspects of image analysis for liver diagnosis and treatment, including the state-of-the-art of liver image analysis and its applications. After discussion on modalities for liver image analysis, various technical elements for liver image analysis such as registration, segmentation, modeling, and computer-assisted detection are covered with examples performed with clinical data sets. Perspective in the imaging technologies is also reviewed and discussed.

Introduction

Considerable reports regarding hepatic tumors have been presented in the past and recent radiology articles [1], [2], [3]. Such literatures include evaluation and development of imaging modalities and techniques, mainly focusing on their detectability of hepatic abnormalities. On the other hand, reports on image processing technology applied to medical images have been increasing according to the rapid development of recent computer technology. Such application (or often called as “solution”) oriented technologies attract researchers in the field of medical science and information technology. In addition, such technologies are expected to help radiologists and surgeons indeed, for example, in detecting hepatic neoplasm imaged in a massive amount of CT images, and in resecting tumors while considering hepatic vasculature. These new and interdisciplinary research fields are called as computer-assisted diagnosis (CAD) (or detection) and computer-assisted surgery (CAS) (and therapy).

In the viewpoint of image processing, there exist several challenging aspects remaining in image analysis for liver. One of the major factors for the difficulty is based on the physical characteristics of the organ. For example, in X-ray CT, there are several adjacent organs such as the pancreas and the stomach that have signal levels similar to the liver. It prevents us from segmenting the liver region from the images with simple algorithms such as region-growing or simple thresholding, which are often used in lung and bronchi segmentation. One of the good solutions for that is use of contrast agent, that is, multi-phase data sets of dynamic CT with contrast enhancement. In such CT data sets, however, another problem arises due to the feature of the liver consisting of soft tissues. In most of the multi-phase CT images, the time lag between one phase and the other allows the liver moves and/or deforms mainly due to respiratory motion of patients. These difficulties become very clear if we compare with the brain-neuro image analysis that is one of the most successful areas for CAD and CAS. In this sense, more advanced approaches are needed for abdominal area.

In this paper, we describe those technical aspects of image analysis for liver diagnosis and treatment, including the state-of-the-art of liver image analysis and its applications. First, several important modalities in the viewpoint of liver CAD and CAS are introduced among the available imaging modalities. Next, technical elements used in liver image analysis are categorized and are described by using actual examples based on clinical data. Then finally, perspective in such technologies is reviewed and is discussed.