Interpretable Multi-Modal Image Registration Network Based on Disentangled Convolutional Sparse Coding

Deng, Xin; Liu, Enpeng; Li, Shengxi; Duan, Yiping; Xu, Mai

doi:10.1109/tip.2023.3240024

Cited by 57 publications

(18 citation statements)

References 60 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The panoramic image stitching methods have been widely studied over the past few decades [10][11][12][13][14][15]. Image registration is the core of image stitching [16], aiming to determine the geometric relationship between multiple images [17]. The image registration methods can be commonly classified as template-based methods and feature-based methods.…”

Section: Related Workmentioning

confidence: 99%

A panoramic image stitching method for nuclear fuel assemblies

Yang,

Cheng,

Chen

et al. 2023

Meas. Sci. Technol.

View full text Add to dashboard Cite

The fuel assembly (FA) is the core component of the nuclear reactor, and inaccurate inspection of FA bending and deformation can cause severe disasters in nuclear power plant reactors. It is an important task to restore a panoramic image of the FA from the video collected by the orthogonal visual inspection system. It can provide a more intuitive qualitative observation. Meanwhile, it is conducive to the estimation of deformation. This paper proposes a panoramic image stitching method for nuclear FA, which classifies video frames into grid images (GI) and non-grid images (NGI). It utilizes the speed of GI frames to estimate the speed of NGI frames. The horizontal offset and vertical offset between frames are estimated in two steps. The high-quality panoramic image can be obtained by image fusion. The experimental results indicate that the proposed method can achieve the panoramic image stitching of simulated samples and actual FAs with a deformation estimation accuracy of 0.3 mm. The proposed method has been verified and used in actual nuclear power plant reactor maintenance.

show abstract

Section: Related Workmentioning

confidence: 99%

A panoramic image stitching method for nuclear fuel assemblies

Yang,

Cheng,

Chen

et al. 2023

Meas. Sci. Technol.

View full text Add to dashboard Cite

show abstract

“…Deep Learning-Based Techniques: Deep learning has emerged as a significant advancement in various fields, revolutionizing the way complex patterns are learned and interpreted from data. Within the realm of medical imaging, particularly in the analysis of CT scans of lungs, there is a prominent trend where deep learning models are being incorporated for nodule segmentation models [13]- [16]. These models, including convolutional neural networks (CNNs), harness large labeled datasets to discern patterns that differentiate nodules from surrounding tissues.…”

Section: A Related Workmentioning

confidence: 99%

Integrated Deep Learning and Stochastic Models for Accurate Segmentation of Lung Nodules From Computed Tomography Images: A Novel Framework

Youssef,

Alksas,

Shalaby

et al. 2023

IEEE Access

View full text Add to dashboard Cite

This paper introduces an innovative model for precise extraction of lung nodules from 3D computed tomography (CT) scans. Our approach comprises two essential preprocessing stages aimed at refining search accuracy and nodule segmentation. Initially, we leverage a two-level joint Markov-Gibbs random field (MGRF) model to delineate the lung region, effectively distinguishing lung wall nodules from the chest region with shared visual characteristics. Subsequently, employing a deep learning U-net technique, we pinpoint the region of interest (ROI) housing the lung nodule, minimizing the inclusion of surrounding lung tissues. Further enhancement comes from a 3D U-net, trained with a novel loss function to mitigate under-or over-segmentation issues. The resulting segmentation robustly outlines lung nodules in terms of morphology and volume metrics, validated by Dice coefficient (DCE), absolute volume difference (AVD), 95 th -percentile Hausdorff distance (HD), sensitivity, and specificity metrics. To assess our approach, we conducted comprehensive experiments. Our evaluation encompasses in vivo data from 50 patients and employs 679 subjects from the publicly available dataset of the Lung Image Database Consortium and Image Database Resource Initiative (LIDC-IDRI). The LIDC-IDRI dataset, a seminal resource for computer-aided diagnosis (CAD) in lung nodules, offers annotations enabling tasks like detection, segmentation, classification, and quantification. Our experiments showcase our model's superiority over existing deep learning methods, particularly evident in metrics such as the 95 th -percentile HD and DCE. While limited demographic information constrains a comprehensive analysis, our approach's robust performance underlines its potential integration into nodule assessment AI systems.

show abstract

“…Transformation parameters found at a lower scale are given as in input to the next level and the LK algorithm iterates until a certain threshold is reached. In another work, the authors of [13] utilized disentangled convolutional sparse coding to separate domain-specific and shared features of multi-modal images for improved accuracy of registration. Multi-scale Generative Adversarial Networks (GANs) are also used to estimate homography parameters as in [36].…”

Section: Related Workmentioning

confidence: 99%

Improving Equality in the Education System of Turkiye

Bitzinger¹,

Perc²

2022

iusd

View full text Add to dashboard Cite

Success breeds more success, the rich get richer, and the powerful get even more powerful. These are often true in human societies and describe the situation where whatever advantages one has in life frequently beget further advantages over time. The social sciences often describe this phenomenon as the Matthew effect of accumulated advantage, and it is known as preferential attachment in network science. The starting point of the process is often due to the first-mover advantage, under which the first person that achieves something will receive recognition and praise at a rate significantly higher than a person who achieves the same at a later point in time. All this boils down to the fact that unequal opportunities early on in education can translate into greater inequalities later in life. A major goal in the Turkish education system is therefore to ensure equal opportunities as well as equal quality of educational processes starting in preschool and beyond. This study explores the thought processed that have led to the implementation of procedures toward this goal, compares the current state in Turkiye with other OECD countries, and outlines future steps that are in the making for reaching targets throughout the educational system.

show abstract

Interpretable Multi-Modal Image Registration Network Based on Disentangled Convolutional Sparse Coding

Cited by 57 publications

References 60 publications

A panoramic image stitching method for nuclear fuel assemblies

A panoramic image stitching method for nuclear fuel assemblies

Integrated Deep Learning and Stochastic Models for Accurate Segmentation of Lung Nodules From Computed Tomography Images: A Novel Framework

Improving Equality in the Education System of Turkiye

Contact Info

Product

Resources

About