Region-scalable Fitting Research Articles

Region-scalable fitting-assisted medical image segmentation with noisy labels

To aid in diagnosis and analysis, deep learning has been applied to medical image analysis by many researchers. Most existing methods train models using many images with precise labels. However, medical images are inherently complex and noisy, making it difficult for medical image segmentation to obtain many accurate labels. That means the study of learning with noisy labels is particularly important. In order to make use of all the samples, some methods correct the noisy labels during the training process. However, in the correction process, a label correction error may occur according to the set policy. Therefore, we propose a method that incorporates image features based on region-scalable fitting (RSF) to assist in the label correction process, called RSF-assisted. We tested the performance of our RSF-assisted method on two learning frameworks of learning with noisy labels based on segmentation: mean-teacher-assisted confident learning (MTCL) and adaptive early-learning correction (ADELE). The experimental results indicate that the performance is improved by using RSF-assisted. To be specific, compared with the original methods, the mean intersection over union (mIOU) is improved by 2.41% after RSF-assisted on thoracic organs (SegTHOR) dataset. Similarly, the 95% Hausdorff distance (95HD) is reduced by up to 2.96 mm after RSF-assisted based on left atrium (LA) segmentation dataset.

Improved Quantitative Analysis Method for Magnetic Particle Imaging Based on Deblurring and Region Scalable Fitting.

Magnetic particle imaging (MPI) is a technique for imaging magnetic particle concentration distribution. It has the advantages of high sensitivity, no signal attenuation with depth, and no ionizing radiation. Although MPI has been widely used in the biomedical field, accurate image analysis has been challenging due to its anisotropic point spread function (PSF). The purpose of this study is to propose an MPI image restoring and segmentation method to facilitate a more precise quantitative evaluation of the magnetic particle imaging in vivo. We proposed a DeRSF method that combined deblurring and region scalable fitting (RSF)to determine the imaging tracer distribution. Then a uniform erosion and scaling criterion was established based on simulation experiments to correct the segmentation results, which was further validated on phantom imaging. Finally, we imaged the MPI tracer at gradient concentrations to establish the calibration curve between the MPI signal and iron mass for iron quantification in phantom and in vivo imaging. The phantom imaging experiments showed that our method achieved improved segmentation performance. The mean value of the dice coefficients for segmentation was up to 0.86, demonstrating that our method can accurately map and quantify the distribution of the tracer. Moreover, the iron quantification on both phantom and in vivo mouse imaging was realized with the minimal error of 5.50%, by our established calibration curve. Our proposed DeRSF method was successfully used for improved MPI quantitative analysis. More importantly, this method also showed accurate quantitative results on images with different shapes and tracer concentrations in both phantom and in vivo data, which laid the foundation for the biomedical study of MPI.

Biological Image Segmentation Using Region-Scalable Fitting Energy with B-Spline Level Set Implementation and Watershed

ObjectivesImage segmentation plays an important role in the analysis and understanding of the cellular process. However, this task becomes difficult when there is intensity inhomogeneity between regions, and it is more challenging in the presence of the noise and clustered cells. The goal of the paper is propose an image segmentation framework that tackles the above cited problems. Material and methodsA new method composed of two steps is proposed: First, segment the image using B-spline level set with Region-Scalable Fitting (RSF) active contour model, second apply the Watershed algorithm based on new object markers to refine the segmentation and separate clustered cells. The major contributions of the paper are: 1) Use of a continuous formulation of the level set in the B-spline basis, 2) Develop the energy function and its derivative by introducing the RSF model to deal with intensity inhomogeneity, 3) For the Watershed, propose a relevant choice of markers that considers the cell properties. ResultsExperimental results are performed on widely used synthetic images, in addition to simulated and real biological images, without and with additive noise. They attest the high quality of segmentation of the proposed method in terms of quantitative and qualitative evaluation. ConclusionThe proposed method is able to tackle many difficulties at the same time: overlapped intensities, noise, different cell sizes and clustered cells. It provides an efficient tool for image segmentation especially biological ones.

Image segmentation using active contour model driven by RSF and difference of Gaussian energy

As we all know, the region scalable fitting method is sensitive to initialisations and suffers from bad results in images with complex scene. In our article, we put forward a new framework by integrating region scalable fitting (RSF) term and difference of Gaussian (DOG) term for segmenting images. We first propose a DOG function which can enhance the contrast at the edges of objects. Then, the RSF energy term is introduced to drive the curve closer to the edge. In the next step, the regularisation term is established which can avoid of the process of reinitialisation. Compared with traditional classical methods, the proposed technique is more flexibility with initialisation and has better segmentation performance.

Active contour model based on local intensity fitting and atlas correcting information for medical image segmentation

Intensity inhomogeneity and noises often occur in real medical images, which present a large degree of challenge to image segmentation. At the same time, most of the existing image segmentation algorithms are sensitive to initial conditions and model parameters. This paper presents an accurate and robust active contour model to solve the above problems. Inspired by the idea of the region-scalable fitting (RSF) model, we first define a local atlas fitting term transformed by the segmentation contour of the coherent local intensity clustering (CLIC) model. Then, we define a new energy functional by merging the atlas term into the energy functional of the RSF model. The advantage of this operation is that it makes full use of the existing segmentation features and advantages of the two models and avoids cumbersome adjustment of model parameters and initial contours. The experimental results clearly show that the improved model not only has better segmentation results than the RSF model and other active contour models such as the LINC, REGAC and SMAP models, but also solves the problem of sensitivity to initial contours, parameters adjustment and noise.

Level set framework with transcendental constraint for robust and fast image segmentation

Though image segmentation models are plentiful and have many applications nowadays, it can be difficult to segment images with complex boundaries and serious intensity inhomogeneity. To some extent, the region-scalable fitting energy model can segment images suffering from intensity inhomogeneity since it considers image intensity as a function, but it relies on initial conditions dramatically. Nowadays, prior knowledge has been widely applied in image segmentation models, which can integrate automatic method and experts experience in one robust and fast segmentation model. In this paper we present a new model that can segment various images accurately by taking the advantages of the region-scalable fitting energy model and the advanced transcendental constraint from artificial experience. The proposed energy functional consists of a smooth length term, a target image data term and a transcendental constraint term. The transcendental constraint term plays a key role in the proposed model, which not only gives the accurate segmentation results but also provides us the chance to carry out the parallel computation. In the proposed-parallel model, the efficiency is improved a lot and the results become more precise compared with other methods. The split Bregman method is applied to minimize the energy functional. Furthermore, we present the convergence analysis and the time complexity analysis of our algorithm. Multiple experimental results and comparisons including parameters sensitivity discussion are shown to demonstrate the superiority of the proposed model such as high accuracy, robustness and efficiency.

Active contour model driven by optimized energy functionals for MR brain tumor segmentation with intensity inhomogeneity correction

Segmentation of brain tumors is important for medical diagnosis, treatment planning, and disease development. However, the presence of image artifacts such as noise, intensity inhomogeneity, and partial volume effect in real-world images can aggressively affect the segmentation task. The manual segmentation of the tumor is highly error-prone and is a time-consuming task. Hence, in this work, we propose a methodology that can identify and segment the brain tumor slices in magnetic resonance (MRI) images. The work is composed of three main stages namely pre-processing, segmentation, and post-processing. In the pre-processing stage, N4ITK is applied to correct the intensity inhomogeneity and an Anisotropic diffusion filter is used to remove the noise present in MR images. In the segmentation stage, an active contour model that combines the region scalable fitting energy (RSF) and optimized laplacian of gaussian energy (OLoG) is proposed to segment the tumor region from MRI. The proposed segmentation method first presents an LoG energy term optimized by an energy functional that can smooth the homogeneous regions and enhance edge information simultaneously. Next, the optimized LoG energy term is combined with the RSF energy term which utilizes the local region information to drive the curve towards the boundaries. With the addition of the LoG term, the proposed model is insensitive to the positions of the initial contour and produces an accurate segmentation result. Finally, morphological operations and thresholding are used to extract the tumor region from the segmented image which is computed previously. The performance of the proposed segmentation method is evaluated using the BRATS, and J.Cheng dataset. The obtained experimental results demonstrate that our proposed method significantly outperforms the manual process and state of the art methods in brain tumor segmentation.

A prostate MRI segmentation method based on prior-shape-information combined with gradient and regional information

With the increasing and younger incidence of prostate cancer, the extraction of prostate contour from magnetic resonance imaging (MRI) plays an increasingly important role in clinical diagnosis and related medical research. In order to accurately segment the prostate region in MR images, we propose a level set segmentation model, which integrates the prior shape information of the prostate and the gradient information into the region-scalable fitting (RSF) model proposed by Li et al. to achieve high-precision segmentation of the prostate. In our model, we use the Gaussian probability model to establish the statistical learning of the prior shape, expressed as the external energy term of the evolution curve, and use the local grey information and the improved gradient information as the internal energy term to construct the final level set segmentation model. This method uses the information of the image itself to modify the evolution curve under the constraints of the prior shape, which can better deal with the small difference between the background and the target area in the medical image, and improve the accuracy and robustness of the segmentation. The final experimental results and analysis illustrate the feasibility of the method used in prostate MRI segmentation.

Active contour regularized semi-supervised learning for COVID-19 CT infection segmentation with limited annotations

Infection segmentation on chest CT plays an important role in the quantitative analysis of COVID-19. Developing automatic segmentation tools in a short period with limited labelled images has become an urgent need. Pseudo label-based semi-supervised method is a promising way to leverage unlabelled data to improve segmentation performance. Existing methods usually obtain pseudo labels by first training a network with limited labelled images and then inferring unlabelled images. However, these methods may generate obviously inaccurate labels and degrade the subsequent training process. To address these challenges, in this paper, an active contour regularized semi-supervised learning framework was proposed to automatically segment infections with few labelled images. The active contour regularization was realized by the region-scalable fitting (RSF) model which is embedded to the loss function of the network to regularize and refine the pseudo labels of the unlabelled images. We further designed a splitting method to separately optimize the RSF regularization term and the segmentation loss term with iterative convolution-thresholding method and stochastic gradient descent, respectively, which enable fast optimization of each term. Furthermore, we built a statistical atlas to show the infection spatial distribution. Extensive experiments on a small public dataset and a large scale dataset showed that the proposed method outperforms state-of-the-art methods with up to 5% in dice similarity coefficient and normalized surface dice, 10% in relative absolute volume difference and 8 mm in 95% Hausdorff distance. Moreover, we observed that the infections tend to occur at the dorsal subpleural lung and posterior basal segments that are not mentioned in current radiology reports and are meaningful to advance our understanding of COVID-19.

Active contour model for inhomogenous image segmentation based on Jeffreys divergence

Inhomogenous image segmentation has been a research challenge in recent years. To deal with this difficulty, we propose a new local and global active contour model based on Jeffreys divergence. First, unlike the local data fitting energy of the region-scalable fitting model, a new local data fitting energy based on Jeffreys divergence is proposed instead of Euclidean distance, which achieves relatively better segmentation. Second, to improve the versatility of the model, a new global data fitting energy based on Jeffreys divergence is proposed. Finally, the adaptive weights of the local and global data fitting energies are developed to increase the robustness to the initial curve. Experiments on real-world and medical images with inhomogeneities indicate that the proposed model can obtain accurate segmentation results efficiently and is not strictly dependent on setting up initial curves.

A Level Set Method with Region-Scalable Fitting Energy for Retinal Layer Segmentation in Spectral-Domain Optical Coherence Tomography Images

Retinal layer segmentation of spectral-domain optical coherence tomography images plays an important role during diagnosis and analysis of ophthalmic diseases. In this paper, a novel variational level set framework with region-scalable fitting energy is proposed for automated retinal layer segmentation in SD-OCT. To the best of our knowledge, it is the first time that level set based method succeeds in ten retinal layers segmentation. The proposed framework consists of three steps. First, an anisotropic nonlinear diffusion filter is applied for speckle noise reduction and ROI contrast enhancement. Second, Canny edge detectors are used to extract initial layers: nerve fiber layer, connecting cilia and retinal pigment epithelium. Finally, the rest retinal layers are segmented by means of level set model combined with prior knowledge of retinal thickness and morphology, for which the energy function consists of region-scalable fitting energy data term, area constraint term, regularization term and length penalty term. The proposed method was tested on 50 retinal SD-OCT B-scans from 50 normal subjects. The overall unsigned border position error is 5.92 ± 4.72 μm. The result showed that data terms with border weight terms can keep layer segmentation results in strong border while retaining its fitting capability in weak border. The proposed method achieves better segmentation result than other active contour models.

RSF model with SCE‐based global constraint for image segmentation

A novel region-scalable fitting (RSF) active contour model with symmetric cross entropy (SCE)-based global constraint is proposed to segment various kinds of images. The energy functional is mainly composed of two parts, namely the local and global constraints. The local constraint is the weighted region-scalable fitting term, which computes the local intensity information. The global constraint is constructed by measuring the difference between the intensity distribution of the object image and the intensity distribution of the original image using SCE. Moreover, some regularised terms are incorporated into the energy functional. Experimental results for synthetic and real images show that the proposed model achieves better segmentation performance and is insensitive to initial contour and noise.

Implementation of Image Segmentation Techniques to Detect MRI Glioma Tumour

Image identification to detect a tumour needs several stages of image processing along with identifying analysis. To get an accurate segmentation of the tumour contour and to identify brain tumour based on brain magnetic resonance imaging (MRI), a suitable techniques and stages of image processing are required to be applied. One technique of mid-level image processing became an objective this work. The objective of the study is to segment the boundary of tumour by applying the Modification of Region Fitting (MRF) method in term of data fitting. The performance of the Region Scalable Fitting (RSF) method and Modified Region Scalable Fitting (MRSF) is evaluated by comparing the number of iterations. As the result, the MRF method has successfully segmented the initial region of brain tumour images.

Evaluation of Image Features Within and Surrounding Lesion Region for Risk Stratification in Breast Ultrasound Images

Feature extraction and classification plays a crucial role in the automated analysis of breast ultrasound (BUS) images. Due to varying sonographic characteristics of benign and malignant lesions, the texture and shape features are mostly employed for designing computer-aided diagnosis (CAD) systems of BUS images. The existing CAD systems use features that are extracted either from the lesion segmented area obtained through segmentation techniques or a rectangular region of interest (ROI) extracted under the guidance of expert Radiologists. However, the significance of features extracted from region comprising only the lesion area is still little explored. This paper investigates the significance of features extracted from the lesion area, lesion surrounding area and rectangular ROI for classification of BUS images. The experiments were conducted on the database of 294 BUS images (104 benign and 190 malignant). Initially, the acquired BUS images were preprocessed through speckle reducing anisotropic diffusion (SRAD) for speckle noise removal. The preprocessed images are segmented using a hybrid segmentation approach including a combination of region-based active contour driven by region-scalable fitting (RBACM-RSF) model and multi-scale Gaussian kernel fuzzy c-means clustering with spatial bias correction (MsGKFCM_S) for getting ROI confined area. The segmented images were further partitioned into two parts (lesion area and lesion surrounding area). Subsequently, a total of 457 texture and shape attributes were extracted from within the lesion area, lesion surrounding area and rectangular ROI comprising of both lesion and its surrounding area. The significance of these features is evaluated using different classifiers (i.e. support vector machine (SVM), Back-propagation artificial neural network (BPANN), Random Forest, AdaBoost). The results indicate that features extracted from within lesion area achieve a maximum classification accuracy of 98.980% with the lowest computational time when linear kernel-based SVM is used.

Deep learning-based visual ensemble method for high-speed railway catenary clevis fracture detection

This paper proposes an automatic visual inspection method for the fracture detection of clevises in the catenary systems of high-speed railways, using images of catenaries captured by an inspection vehicle. First, the clevises are extracted from the catenary image using a convolutional neural network based algorithm, known as the faster region-based convolutional neural network. Because the structure of catenary systems does not have many variations and the contextual information near a catenary fitting may have strong correlation with its category, the architecture of the original faster region-based convolutional neural network is modified to make use of the contextual information of the regions of interest in the images for object recognition. A crack detection process is then used to recognize the fractures of clevises. To detect the cracks, the edge map of the clevis sub-image is generated using a region-scalable fitting model. Areas where the cracks are most likely to occur are projected from a standard clevis image to the clevis sub-image by shape context matching and affine transformation matrix computation. The cracks are then recognized by calculating the wavelet entropy inside these areas followed by morphological filtering. Experimental results show that the modified faster region-based convolutional neural network architecture achieves better results in clevis extraction than the original architecture as well as some other state-of-art object detection models. The detection is not affected by the scaling, texture and grayscale changes of the clevises caused by the variation of shooting distance, shooting angle and illumination variations. The fractures of the clevises can be accurately and reliably detected using the fracture detection method proposed in this paper and the performance of this visual inspection method meets the strict requirements for catenary system maintenance.

M‐WRSF model for medical image segmentation

In this Letter, multichannel weighted region-scalable fitting segmentation model (M-WRSF) is proposed for medical image segmentation. The authors have utilised a new edge detection function to improve the performance of image segmentation approaches that already exists on weak edges and the grey-scale inhomogeneity of some medical images. The M-WRSF model introduces a novel punishment item to improve numerical stability and augments time interval to boost iterative efficiency. The Gaussian kernel function is added on the basis of original model to enhance robustness. On the other side, the original medical images need to make de-noising operation before the multichannel active contour model segment them, and also level-set initial curve has no effect on the segmentation results. The benefits of the new M-WRSF model, specifically, the increased efficiency and usability, are verified in the simulation experiment.

A Distance Regularized Level-Set Evolution Model Based MRI Dataset Segmentation of Brain’s Caudate Nucleus

The caudate nucleus of the brain is highly correlated to the emotional decision-making of pessimism, which is an important process for improving the understanding and treatment of depression; and the segmentation of the caudate nucleus is the most basic step in the process of analysis and research concerning this region. In this paper, Level Set Method (LSM) is applied for caudate nucleus segmentation. Firstly, Distance Regularized Level Set Evolution (DRLSE), Region-Scalable Fitting (RSF) and Local Image Fitting (LIF) models are proposed for segmentation of the caudate nucleus of Magnetic Resonance Imaging (MRI) images of the brain, and the segmentation results are compared by using selected evaluation indices. The average Dice Similarity Coefficient (DSC) values of the proposed three methods all exceed 85%, and the average Jaccard Similarity (JS) values are over 77%, respectively. The results indicate that all these three models can have good segmentation results for medical images with intensity inhomogeneity and meet the general segmentation requirements, while the proposed DRLSE model performs better in segmentation.

Hybrid segmentation method based on multi‐scale Gaussian kernel fuzzy clustering with spatial bias correction and region‐scalable fitting for breast US images

Automated segmentation of tumors in breast ultrasound (US) images is challenging due to poor image quality, presence of speckle noise, shadowing effects and acoustic enhancement. This paper improves the multi‐scale Gaussian kernel induced fuzzy C‐means clustering method with spatial bias correction (MsGKFCM_S). Furthermore, it presents a hybrid segmentation method, using both the features of the MsGKFCM_S clustering and active contour driven by a region‐scalable fitting energy function. The result obtained from the MsGKFCM_S method is utilised to initialise the contour that spreads to identify the estimated regions. It also helps to estimate the several controlling parameters of the curve evolution process. The proposed approach is evaluated on a database of 127 breast US images consisting of 75 malignant and 52 solid benign cases. The performance of proposed approach is compared with other related techniques, using performance measures such as Jaccard Index, dice similarity, shape similarity, Hausdroff difference, area difference, accuracy and F‐measure. Results indicate that the proposed approach can successfully detect lesions in breast US images, with high accuracy of 97.889 and 97.513%. Moreover, the proposed approach has the capability of handling shadowing effects, acoustic enhancement and multiple lesions.

LATE: A Level-Set Method Based on Local Approximation of Taylor Expansion for Segmenting Intensity Inhomogeneous Images.

Intensity inhomogeneity is common in real-world images and inevitably leads to many difficulties for accurate image segmentation. Numerous level-set methods have been proposed to segment images with intensity inhomogeneity. However, most of these methods are based on linear approximation, such as locally weighted mean, which may cause problems when handling images with severe intensity inhomogeneities. In this paper, we view segmentation of such images as a nonconvex optimization problem, since the intensity variation in such an image follows a nonlinear distribution. Then, we propose a novel level-set method named local approximation of Taylor expansion (LATE), which is a nonlinear approximation method to solve the nonconvex optimization problem. In LATE, we use the statistical information of the local region as a fidelity term and the differentials of intensity inhomogeneity as an adjusting term to model the approximation function. In particular, since the first-order differential is represented by the variation degree of intensity inhomogeneity, LATE can improve the approximation quality and enhance the local intensity contrast of images with severe intensity inhomogeneity. Moreover, LATE solves the optimization of function fitting by relaxing the constraint condition. In addition, LATE can be viewed as a constraint relaxation of classical methods, such as the region-scalable fitting model and the local intensity clustering model. Finally, the level-set energy functional is constructed based on the Taylor expansion approximation. To validate the effectiveness of our method, we conduct thorough experiments on synthetic and real images. Experimental results show that the proposed method clearly outperforms other solutions in comparison.

A variable region scalable fitting energy approach for human Metaspread chromosome image segmentation

The medical images often suffer from intensity inhomogeneity and therefore they are difficult to segment using the conventional image segmentation approaches. In this work an image segmentation approach is being presented, in which the region scalable fitting energy is variable for the regions inside and outside the contours. The energy term is defined along the two sides of the contour, by considering different scale parameters for the foreground and background regions. Two fitting functions are used to approximate the intensities of the foreground and background regions. The proposed method is able to deal with intensity inhomogeneity as a Gaussian kernel function has been used in the energy formulation. The kernel function is used to assign weights to the intensity values during energy formulation. Different scale parameters are used for the kernel function for background and foreground regions. The proposed method proves to be robust as it has the self-regularization capability, partial differential equations have been discretized by approximations and the method does not require computationally expensive re-initialization procedure. The proposed method is capable of efficiently segmenting the human metaspread chromosome images that suffer from intensity inhomogeneity. The chromosomes that have very low contrast with the background or the group of chromosomes that appear to be nearly touching each other are segmented correctly from the metaspread images.