An Unsupervised Learning-Based Multi-Organ Registration Method for 3D Abdominal CT Images.

Shaodi Yang,Fan Zhang,Miao Liao,Yuqian Zhao

doi:10.3390/s21186254

Abstract

Medical image registration is an essential technique to achieve spatial consistency geometric positions of different medical images obtained from single- or multi-sensor, such as computed tomography (CT), magnetic resonance (MR), and ultrasound (US) images. In this paper, an improved unsupervised learning-based framework is proposed for multi-organ registration on 3D abdominal CT images. First, the explored coarse-to-fine recursive cascaded network (RCN) modules are embedded into a basic U-net framework to achieve more accurate multi-organ registration results from 3D abdominal CT images. Then, a topology-preserving loss is added in the total loss function to avoid a distortion of the predicted transformation field. Four public databases are selected to validate the registration performances of the proposed method. The experimental results show that the proposed method is superior to some existing traditional and deep learning-based methods and is promising to meet the real-time and high-precision clinical registration requirements of 3D abdominal CT images.

Highlights

Sensors and sensing systems play important roles in various medical applications, including disease diagnosis, monitoring, preoperative planning, surgical navigation, and so on [1,2,3,4]
Aiming at avoiding the time-consuming pre-processing and maintaining the topologypreserving property of the transformation, we developed an unsupervised learning-based registration framework for the segmented multi-organ from 3D abdominal computed tomography (CT) images
We present an improved unsupervised learning-based framework for multi-organ registration from 3D abdominal CT images

Summary

Introduction

Sensors and sensing systems play important roles in various medical applications, including disease diagnosis, monitoring, preoperative planning, surgical navigation, and so on [1,2,3,4]. Registration is one of the fundamental technologies that enable the sensing systems to be used in the above-mentioned applications [5,6,7,8]. Most existing methods cannot simultaneously meet the clinical requirements of high-accuracy and real-time performance for full abdominal image registration. To solve the above problems, many researchers pay attention to the segmented-based abdominal image registration methods. Li et al [9] proposed a liver MR image registration method based on the respiratory sliding motion segmentation that achieves more accurate registration results. Xie et al [10] proposed a lung and liver 4D-CT image registration method based on tissue features and ROI segmentation, which can be implemented on clinical data

Methods

Results

Conclusion