Article
Computer Science, Artificial Intelligence
Feng Chen, Fei Wu, Jing Xu, Guangwei Gao, Qi Ge, Xiao-Yuan Jing
Summary: Deformable Convolutional Networks (DCNs) are proposed to solve the limited geometric transformation in CNNs, with reformulation of convolution and pooling modules to enhance focus ability, and empirical study on various arrangements leading to significant findings. Achieving state-of-the-art performance on semantic segmentation and object detection benchmarks.
Article
Engineering, Biomedical
Xiao Liang, Howard Morgan, Ti Bai, Michael Dohopolski, Dan Nguyen, Steve Jiang
Summary: CBCT-based online adaptive radiotherapy requires accurate auto-segmentation, but DL-based direct segmentation of CBCT images is challenging due to poor quality and lack of well-labelled datasets. This study proposes a method using DIR and pseudo labels derived from deformed pCT contours for initial training, influencer volumes for defining the region of interest, and fine-tuning with a smaller set of true labels. Evaluation on nine patients shows that DL-based direct segmentation with influencer volumes improves performance to reach the level of DIR-based segmentation.
PHYSICS IN MEDICINE AND BIOLOGY
(2023)
Article
Engineering, Biomedical
Ruurd J. A. Kuiper, Marijn van Stralen, Ralph J. B. Sakkers, Rick H. J. Bergmans, Frank Zijlstra, Max A. Viergever, Harrie Weinans, Peter R. Seevinck
Summary: The study introduced a new registration method QIR for complex deformations in rotating joints like the knee, showing significant improvements over conventional methods. Results demonstrated that QIR achieved higher registration accuracy and bone rigidity compared to traditional approaches.
PHYSICS IN MEDICINE AND BIOLOGY
(2021)
Article
Automation & Control Systems
Liang Qiu, Hongliang Ren
Summary: This article presents a joint learning framework named RSegNet for concurrent deformable registration and segmentation, achieving improved accuracy of both tasks. By minimizing an integrated loss function, utilizing data augmentation and dual-consistency supervision, the method demonstrates better anatomical consistency and deformation regularity, resulting in increased segmentation and registration accuracy.
IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING
(2022)
Article
Multidisciplinary Sciences
E. Erdem Tuna, Dominique Franson, Nicole Seiberlich, M. Cenk Cavusoglu
Summary: This study proposes a particle filter based framework for tracking the cardiac surface from a time sequence of MRI slices, aiming to use it for interventional cardiovascular magnetic resonance procedures. The framework uses a low-order parametric deformable model and a stochastic dynamic system to represent the cardiac surface motion. Deformable models and adaptive filters are employed to control the deformations and model the complex cardiac motion. The proposed method is validated using numerical and real cardiac MRI datasets.
SCIENTIFIC REPORTS
(2023)
Article
Radiology, Nuclear Medicine & Medical Imaging
Kaicong Sun, Sven Simon
Summary: The proposed FDRN outperforms existing state-of-the-art methods in brain MR image registration by utilizing a compact autoencoder structure and efficient learning, and is a generalized framework not confined to specific types of medical images or anatomy.
Article
Engineering, Biomedical
Yamina Yahia Lahssene, Lila Meddeber, Tarik Zouagui, Rachid Jennane
Summary: The paper introduces a new method for medical image segmentation, utilizing the Selective Binary Level Set function and a new variant of the Topology Preserving Selective Binary Level Set model to achieve global or local segmentations. By evaluating and comparing the results of ventricle segmentation, the effectiveness of the proposed method is demonstrated.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2021)
Article
Engineering, Biomedical
Jia Mi, Wenhao Yin, Lei Zhao, Yangfan Chen, Yujia Zhou, Qianjin Feng
Summary: This study proposes a reconstruction-based 3D/2D registration method for spine surgery navigation. The method utilizes a segmentation network and a multi-scale pose estimation module to establish dimensional correspondence and estimate the pose parameters in 3D space. The experimental results demonstrate that the proposed method achieves considerable improvement in registration performance compared to other methods.
PHYSICS IN MEDICINE AND BIOLOGY
(2023)
Article
Radiology, Nuclear Medicine & Medical Imaging
Marica Masi, Valeria Landoni, Adriana Faiella, Alessia Farneti, Simona Marzi, Maria Guerrisi, Giuseppe Sanguineti
Summary: Rigid image coregistration is sufficiently accurate for radiotherapy of prostate bed cancer recurrence. Deformable registration tends to shrink the voxels and dislocate the region of interest, but this can be mitigated by controlling organ filling.
PHYSICA MEDICA-EUROPEAN JOURNAL OF MEDICAL PHYSICS
(2021)
Article
Medicine, General & Internal
Taeyong Park, Seungwoo Khang, Heeryeol Jeong, Kyoyeong Koo, Jeongjin Lee, Juneseuk Shin, Ho Chul Kang
Summary: This study proposes a novel segmentation and nonrigid registration method to improve the diagnosis and treatment process of X-ray angiography. By using a convolutional neural network for segmentation and CT angiography for topological analysis, accurate registration and segmentation are achieved. The evaluation results show that the method has high accuracy and feasibility.
Article
Computer Science, Interdisciplinary Applications
Dongming Wei, Sahar Ahmad, Yuyu Guo, Liyun Chen, Yunzhi Huang, Lei Ma, Zhengwang Wu, Gang Li, Li Wang, Weili Lin, Pew-Thian Yap, Dinggang Shen, Qian Wang
Summary: In this paper, a recurrently usable deep neural network is proposed for the registration of infant brain MR images. By using brain tissue segmentation maps for registration and training a single registration network that is recurrently applied in inference, the proposed method overcomes the challenge of fast brain development in infants. Experimental results show that the method achieves the highest registration accuracy while preserving the smoothness of the deformation field.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2022)
Article
Computer Science, Artificial Intelligence
Cheolhong An, Yiqian Wang, Junkang Zhang, Truong Q. Nguyen
Summary: This paper proposes a self-supervised multimodal retina registration method to alleviate the burden of preparing training data, achieving comparable accuracy to the state-of-the-art supervised learning method in terms of registration accuracy and Dice coefficient.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2022)
Article
Geochemistry & Geophysics
Leping He, Shuaiqing Wang, Qijun Hu, Qijie Cai, Muyao Li, Yu Bai, Kai Wu, Bo Xiang
Summary: This article proposes a new method for three-dimensional point cloud registration, optimizing the geometric features to improve accuracy and speed. Experimental results demonstrate that this method significantly outperforms traditional ICP and other state-of-the-art registration methods in terms of accuracy and speed.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Engineering, Biomedical
Zhiyong Zhou, Ben Hong, Xusheng Qian, Jisu Hu, Minglei Shen, Jiansong Ji, Yakang Dai
Summary: Deformable multimodal image registration is crucial in medical image analysis. macJNet is a weakly-supervised method that uses a joint learning framework and a cascaded modality independent neighborhood descriptor (macMIND) to align multimodal medical images. The proposed macMIND provides dense correspondence and enhances the representation ability of cross-modal features. The registration network and segmentation networks mutually improve the performance of multimodal image segmentation.
BIOMEDICAL ENGINEERING ONLINE
(2023)
Article
Environmental Sciences
Xiang Cheng, Hong Lei
Summary: This paper proposes a combination model of a modified multiscale deformable convolutional neural network (mmsDCNN) and dense conditional random field (DenseCRF) for the semantic segmentation of remote sensing images. The model consists of a lightweight multiscale deformable convolutional network to generate preliminary prediction probability maps at each pixel, and a multi-level DenseCRF model to optimize the coarse segmentation results. The model also utilizes edge contour features and achieves significant advantages compared to state-of-the-art models.
Article
Biology
Myungeun Lee, Wanhyun Cho, Sunworl Kim, Soonyoung Park, Jong Hyo Kim
COMPUTERS IN BIOLOGY AND MEDICINE
(2012)
Proceedings Paper
Engineering, Electrical & Electronic
Wanhyun Cho, Sunworl Kim, Sangcheol Park
MULTIMEDIA ON MOBILE DEVICES 2012 AND MULTIMEDIA CONTENT ACCESS: ALGORITHMS AND SYSTEMS VI
(2012)