Modeling Respiratory Signals by Deformable Image Registration on 4DCT Lung Images

The lung organ of human anatomy captured by a medical device reveals inhalation and exhalation information for treatment and monitoring. Given a large number of slices covering an area of the lung, we have a set of three-dimensional lung data. And then, by combining additionally with breath-hold measurements, we have a dataset of multigroup CT images (called 4DCT image set) that could show the lung motion and deformation over time. Up to now, it has still been a challenging problem to model a respiratory signal representing patients’ breathing motion as well as simulating inhalation and exhalation process from 4DCT lung images because of its complexity. In this paper, we propose a promising hybrid approach incorporating the local binary pattern (LBP) histogram with entropy comparison to register the lung images. The segmentation process of the left and right lung is completely overcome by the minimum variance quantization and within class variance techniques which help the registration stage. The experiments are conducted on the 4DCT deformable image registration (DIR) public database giving us the overall evaluation on each stage: segmentation, registration, and modeling, to validate the effectiveness of the approach.

Few-shot learning for deformable image registration in 4DCT images

Objectives: To develop a rapid and accurate 4D deformable image registration (DIR) approach for online adaptive radiotherapy. Methods: We propose a deep learning (DL)-based few-shot registration network (FR-Net) to generate deformation vector fields from each respiratory phase to an implicit reference image, thereby mitigating the bias introduced by the selection of reference images. The proposed FR-Net is pretrained with limited unlabeled 4D data and further optimized by maximizing the intensity similarity of one specific four-dimensional computed tomography (4DCT) scan. Because of the learning ability of DL models, the few-shot learning strategy facilitates the generalization of the model to other 4D data sets and the acceleration of the optimization process. Results: The proposed FR-Net is evaluated for 4D groupwise and 3D pairwise registration on thoracic 4DCT data sets DIR-Lab and POPI. FR-Net displays an averaged target registration error of 1.48 mm and 1.16 mm between the maximum inhalation and exhalation phases in the 4DCT of DIR-Lab and POPI, respectively, with approximately 2 min required to optimize one 4DCT. Overall, FR-Net outperforms state-of-the-art methods in terms of registration accuracy and exhibits a low computational time. Conclusion: We develop a few-shot groupwise DIR algorithm for 4DCT images. The promising registration performance and computational efficiency demonstrate the prospective applications of this approach in registration tasks for online adaptive radiotherapy. Advances in knowledge: This work exploits DL models to solve the optimization problem in registering 4DCT scans while combining groupwise registration and few-shot learning strategy to solve the problem of consuming computational time and inferior registration accuracy.