A full migration BBO algorithm with enhanced population quality bounds for multimodal biomedical image registration
Introduction
Medical imaging is the technique and process of creating visual representations of the interior of a body for diagnosis, disease monitoring and medical intervention, as well as visual representation of the function of some organs or tissues [1], [2], [3]. In the development of image acquisition techniques, the images are acquired from different devices for different representations [4], [5]. For example, computer-assisted tomography (CT) is very helpful for imaging bodily structures and dense tissues, while magnetic resonance imaging (MRI) and ultrasound (US) are powerful tools for the visualization of soft tissue. In addition, functional imaging is becoming more and more important in clinical and medical research. For example, positron emission tomography (PET) and single photon computed tomography (SPECT) imaging provide information about blood flow and metabolic processes. In most cases, many of the applications address images with different modalities that were acquired from different acquisition devices at different times, possibly from different perspectives for image comparison, integration, or the fusion of visualizations. These multimodal images are used in a complementary approach to acquire additional insights into a pathological phenomenon. For these multimodal images to be used, they must be properly combined or fused. Therefore, registration is a key technology in medical imaging, as it enables us to align and integrate different images that have a shared content, which are obtained under different conditions, into a single representation [6].
Medical image registration is becoming increasingly important in diagnosis, treatment planning, functional studies, computer-guided therapies, and medical research. Registration involves the transformation of different images that have a shared content into a common coordinate system. Specifically, the objective of the registration is to estimate the optimal geometric transformation that maps the content of one image to the corresponding area of the other. For image processing [7], [8], [9], [10], [11], one key issue of registration is to find the optimal transformation that is settled by optimization methods guided by a similarity metric [12], [13] that can measure the degree of resemblance between two images.
There are many methods for medical image registration. These methods can be roughly divided into two categories: feature-based and intensity-based. Feature-based methods [6] perform alignments using only parts of the image that are salient and distinctive parts, such as lines, corners and contours. These methods can greatly reduce the complexity of the registration problem, while they can significantly depend on the reliability of the feature extraction algorithm. However, these methods are limited to the images that have salient and distinctive features. On the other hand, much work has recently focused on intensity-based methods, in which all of the image data is utilized. A large amount of data can be processed at the expense of increasing the computational requirements, which does not require the extra preprocessing, such as segmentation or feature extraction, in the intensity-based methods. There are usually three important aspects to be considered [14], [15]. First, the search domain is the group of potential transformations that are used to align the images. Second, the similarity metric is a real valued function that quantifies the degree of resemblance between two images. Finally, the search method explores the domain of possible transformations guided by the similarity metric. There are local and global search methods to optimize the similarity metric.
As is known, the key to obtaining good registration results is to find a suitable search method that can optimize the similarity metric. Therefore, the motivation of our work is to design an optimization algorithm to maximize the similarity metric for aligning medical images that are obtained from different modalities. Many algorithms have been proposed in the medical image registration area. For example, the original Powell’s method and its recent variants [16], discrete optimization of Markov random fields (MRFs) [17] and different gradient-based approaches [18] are distinguished from traditional optimization algorithms. However, many adverse factors, such as noise, image discretization and misalignment affect the process of traditional optimization algorithms, which can easily become stuck at local optima. To overcome some of these drawbacks, many metaheuristics-based methods [19] have been proposed and demonstrated. Metaheuristics-based methods have successful applications in many fields [20], [21], [22], [23] and are often used to appropriately settle complex medical image registration problems, due to their advantages in obtaining a robust global optimum. Many evolutionary methods have been proposed in recent decades with the purpose of improving the results of the different medical image registration problems [24], [25]. Among them, an approach based on the Coral Reef Optimization (CRO) algorithm, which is called CRO-SL, was proposed with outstanding results [26]. As far as we know, CRO-SL is the most notable and represents the state-of-the-art. The biogeography-based optimization algorithm (BBO) [27] is a search algorithm that is based on the law of biological migration [28]. This algorithm has good convergence and stability and is widely used in various fields. In this work, we design a new BBO based optimization that is specifically tailored to the medical image registration problem.
Section snippets
Medical image registration
In general, an image registration problem involves two images with different roles: a reference image and a scene image, known as a model and scene, respectively. The scene is the image that is transformed to reach the geometry of the model. Fig. 1 shows the flowchart of the registration process. The registration process can be regarded as an optimization problem that aims to find a geometric transformation that is applied to the scene to make it as similar as possible to the model [6].
Much
The proposed method
In this section, the proposed method for solving the medical image registration problem is presented. First, we introduce the basic BBO algorithm, which was motivated by a natural process. Afterward, an improved BBO with an elite learning algorithm will be introduced.
Experiments and results
In this section, a series of experiments are carried out to demonstrate the feasibility of the BBO-EL algorithm to deal with a real-world biomedical image registration problem in which the multimodal biomedical image registration situation is a tougher challenge. First, we describe the image dataset that we used in the experiments. Second, we will show the detailed comparison results among the BBO-EL, the established algorithm SS* and the CRO-SL which represent the state-of-the-art.
Conclusions and future work
In this work, we described the design and implementation of a novel nature-inspired technique, known as the BBO-EL algorithm, to solve the problem of multimodal biomedical image registration. BBO-EL is an enhanced version of an algorithm that is based on the law of biological migration. We designed three operators: the hybrid full migration operator, the undo operator and the elite learning operator. These three operators, when integrated, can improve both the lower bound and the upper bound of
CRediT authorship contribution statement
Yilin Chen: Conceptualization, Methodology, Software, Validation, Investigation, Writing - original draft. Fazhi He: Supervision, Project administration, Writing - review & editing, Funding acquisition. Haoran Li: Visualization, Writing - review & editing. Dejun Zhang: Funding acquisition, Writing - review & editing. Yiqi Wu: Funding acquisition, Writing - review & editing.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This work was supported by National Key R&D Program of China (Grant No. 2017YFB0503004), the National Natural Science Foundation of China (Grant No. 61702350 and 61802355), China Postdoctoral Science Foundation (Grant No. 2019M662709). The authors would like to thank the Retrospective Image Registration Evaluation (RIRE) project for providing the medical image volumes that were used in our experiments.
References (76)
- et al.
Deep learning based enhanced tumor segmentation approach for MR brain images
Appl. Soft Comput.
(2019) - et al.
Study on the improved fuzzy clustering algorithm and its application in brain image segmentation
Appl. Soft Comput.
(2019) - et al.
A transform-based fast fuzzy C-means approach for high brain MRI segmentation accuracy
Appl. Soft Comput.
(2019) - et al.
Image registration methods: a survey
Image Vis. Comput.
(2003) - et al.
A survey of medical image registration
Med. Image Anal.
(1998) - et al.
A survey of medical image registration–under review
Med. Image Anal.
(2016) - et al.
Improved optimization for the robust and accurate linear registration and motion correction of brain images
Neuroimage
(2002) - et al.
Linear intensity-based image registration by Markov random fields and discrete optimization
Med. Image Anal.
(2010) - et al.
Economic emission dispatch problems with stochastic wind power using summation based multi-objective evolutionary algorithm
Inform. Sci.
(2016) - et al.
Intensity-based image registration using scatter search
Artif. Intell. Med.
(2014)
A novel evolutionary rigid body docking algorithm for medical image registration
Swarm Evol. Comput.
Coral reef optimization with substrate layers for medical image registration
Swarm Evol. Comput.
Metropolis biogeography-based optimization
Inform. Sci.
An overlap invariant entropy measure of 3D medical image alignment
Pattern Recognit.
Slice-to-volume medical image registration: A survey
Med. Image Anal.
Fully automatic 3D feature-based registration of multi-modality medical images
Image Vis. Comput.
Real-coded genetic algorithms and interval-schemata
Biogeography based hybrid scheme for automatic detection of epileptic seizures from EEG signatures
Appl. Soft Comput.
Biogeography-based optimization for identifying promising compounds in chemical process
Neurocomputing
Virtual machine consolidated placement based on multi-objective biogeography-based optimization
Future Gener. Comput. Syst.
An analysis of the equilibrium of migration models for biogeography-based optimization
Inform. Sci.
Self-adaptive mutation differential evolution algorithm based on particle swarm optimization
Appl. Soft Comput.
A multi-objective differential evolutionary algorithm for constrained multi-objective optimization problems with low feasible ratio
Appl. Soft Comput.
Parallel ant colony optimization on multi-core SIMD CPUs
Future Gener. Comput. Syst.
A novel nature-inspired algorithm for optimization: Squirrel search algorithm
Swarm Evol. Comput.
A novel enhanced collaborative autoencoder with knowledge distillation for top-N recommender systems
Neurocomputing
DRCDN: Learning deep residual convolutional dehazing networks
Vis. Comput.
A multi-phase blending method with incrementalintensity for training detection networks
Vis. Comput.
Joint learning of image detail and transmission map for single image dehazing
Vis. Comput.
A scalable region-based level set method using adaptive bilateral filter for noisy image segmentation
Multimedia Tools Appl.
A parallel and robust object tracking approach synthesizing adaptive Bayesian learning and improved incremental subspace learning
Front. Comput. Sci.
A novel bat algorithm based on cross boundary learning and uniform explosion strategy
Appl. Math.-J. Chin. Univ.
Photo-realistic dehazing via contextual generative adversarial networks
Mach. Vis. Appl.
2-D and 3-D Image Registration: For Medical, Remote Sensing, and Industrial Applications
A dividing-based many-objectives evolutionary algorithm for large-scale feature selection
Soft Comput.
Adaptive stochastic gradient descent optimisation for image registration
Int. J. Comput. Vis.
Hybrid metaheuristic algorithms: past, present, and future
An efficient GPU-based parallel tabu search algorithm for hardware/software co-design
Front. Comput. Sci.
Cited by (131)
Haar-wavelet based texture inpainting for human pose transfer
2024, Information Processing and ManagementPartial reinforcement optimizer: An evolutionary optimization algorithm[Formula presented]
2024, Expert Systems with ApplicationsA metaheuristic-driven physical asset risk management framework for manufacturing system considering continuity measures
2023, Engineering Applications of Artificial IntelligenceHybrid machine learning models for prediction of daily dissolved oxygen
2023, Journal of Water Process Engineering