基于人工蜂群优化的MR图像分割算法研究

针对传统图像阈值分割算法在MR图像分割时存在的易受采集图像灰度不均、医学图像易受噪声干扰,因而难以得到准确分割阈值的问题,本文将人工蜂群算法与二维OSTU阈值分割算法相结合,提出一种基于人工蜂群优化的MR图像分割算法。使用医学图像的离散度矩阵的迹作为人工蜂群优化的目标函数,得到二维OSTU的最佳分割阈值;根据得到的最佳阈值,对图像采用二维OSTU分割的方法进行分割。实验结果证明,对于医学MR图像,本文所提出的算法具有精度高和鲁棒性强的特点,能够得到精确的分割后图像。     

Robust automated three-dimensional segmentation of secondary ion mass spectrometry image sets

Three-dimensional (3-D) element distributions generated by scanning secondary ion mass spectrometry (SIMS) are usually noisy and blurred and contain objects which do not usually have sharp edges or may have noise induced boundaries. Additionally, there are local intensity differences due to sensitivity differences of the channelplate. As a result, traditional segmentation techniques become difficult and yield rather poor results. We present a novel methodology which combines a restoration process (using a combination of channelplate sensitivity compensation with a 3-D de-noising technique based on the wavelet transform) with a fuzzy logic 3-D gray level segmentation which can be used to successfully segment 3-D SIMS image sets. The restoration algorithm removes the artifacts produced by the channelplate inhomogeneities as well as noise aberrations from the image sets and the gray level thresholding algorithm segments their features. The algorithm is designed for minimal user interaction to achieve a high automation level. The methodology is discussed and experimental results using real 3-D images are presented.     

On the implementation of spatial constraints in multivariate curve resolution alternating least squares for hyperspectral image analysis

Hyperspectral imaging (HSI) is a method for exploring spatial and spectral information associated with the distribution of the different compounds in a chemical or biological sample. Amongst the multivariate image analysis tools utilized to decompose the raw data into a bilinear model, multivariate curve resolution alternating least squares (MCR‐ALS) can be applied to obtain the distribution maps and pure spectra of the components of the sample image. However, a requirement is to have the data in a two‐way matrix. Thus, a preliminary step consists of unfolding the raw HSI data into a single‐pixel direction. Consequently, through this data manipulation, the information regarding pixel neighboring is lost, and spatial information cannot directly be constrained on the component profiles in the current MCR‐ALS algorithm. In this short communication, we propose an adaptation of the MCR‐ALS framework, enabling the potential implementation of any variation of spatial constraint. This can be achieved by adding, at each least‐squares step, refolding/unfolding of the distribution maps for the components. The implementation of segmentation, shape smoothness, and image modeling as spatial constraints is proposed as a proof of concept. Copyright © 2015 John Wiley & Sons, Ltd.     

Use of local rank‐based spatial information for resolution of spectroscopic images

Spectroscopic images are singular chemical measurements that enclose chemical and spatial information about samples. Resolution of spectroscopic images is focused on the recovery of the pure spectra and distribution maps of the image constituents from the sole raw spectroscopic measurement. In image resolution, constraints are generally limited to non‐negativity and the spatial information is generally not used. Local rank analysis methods have been adapted to describe the local spatial complexity of an image, providing specific pixel information. This local rank information combined with reference spectral information allows the identification of absent compounds in pixels with low compound overlap. The introduction of this information in the resolution process under the form of constraints helps to increase the performance of the resolution method and to decrease the ambiguity linked to the final solutions. Copyright © 2008 John Wiley & Sons, Ltd.     

An image analysis suite for spot detection and spot matching in two-dimensional electrophoresis gels

We propose a suite of novel algorithms for image analysis of protein expression images obtained from 2-D electrophoresis. These algorithms are a segmentation algorithm for protein spot identification, and an algorithm for matching protein spots from two corresponding images for differential expression study. The proposed segmentation algorithm employs the watershed transformation, k-means analysis, and distance transform to locate the centroids and to extract the regions of the proteins spots. The proposed spot matching algorithm is an integration of the hierarchical-based and optimization-based methods. The hierarchical method is first used to find corresponding pairs of protein spots satisfying the local cross-correlation and overlapping constraints. The matching energy function based on local structure similarity, image similarity, and spatial constraints is then formulated and optimized. Our new algorithm suite has been extensively tested on synthetic and actual 2-D gel images from various biological experiments, and in quantitative comparisons with ImageMaster2D Platinum the proposed algorithms exhibit better spot detection and spot matching.     

Lab分通道直方图的彩色图像分割算法及应用

本文针对传统灰度直方图分割法未综合考虑图像色度及纹理特征、对灰度差异不明显或灰度范围重叠的图像出现过分割或欠分割等问题，提出一种新的基于Lab分通道直方图的彩色图像分割算法，引入具有序列不相关性的亮度L通道、红绿a通道及蓝黄b通道3种分割依据，通过Newton插值法进行拟合运算，可针对不同亮度、色度属性图像进行自由选择，并运用邻域灰度值相匹配原则解决相邻目标区域边缘像素的准确匹配问题，分局部、分形态、分区域实现图像中不同目标的提取。经验证，该法对区域亮度差异较大图像及区域色度差异显著于亮度差异图像的分割效果，均优于传统灰度直方图分割法，极大提升了直方图分割算法的适用性。将其与经典Reinhard色彩迁移算法结合，将源图像感兴趣目标区域经分通道分割后分别进行色彩迁移变换，较好解决了经典Reinhard算法对图像非目标区域的干扰、色彩误传及阶调层次损失严重等问题，突破传统迁移算法只能整体着色的局限性，实现分区域精准着色。     

Analysis of scintigrams using fuzzy theory

Textural features for phantom images were extracted. Texture parameters which represent RI distribution--skew, energy, entropy and angular second moment were used. But, it was difficult to analyse the images using discriminant analysis for textural features, because textural features had statistical noise. Therefore fuzzy reasoning was adapted to analyse the images. Textural features for six kinds of images were showed using membership function. The possibility to the image was evaluated using the value of membership function on each images. Fuzzy reasoning could be done easily using max-min composition formula. The reasoning was found more suitable to analyse the images than discriminant analysis and will be considered useful for analysis of clinical scintigrams.     

SpaRef: a clustering algorithm for multispectral images

Multispectral images such as multispectral chemical images or multispectral satellite images provide detailed data with information in both the spatial and spectral domains. Many segmentation methods for multispectral images are based on a per-pixel classification, which uses only spectral information and ignores spatial information. A clustering algorithm based on both spectral and spatial information would produce better results.

In this work, spatial refinement clustering (SpaRef), a new clustering algorithm for multispectral images is presented. Spatial information is integrated with partitional and agglomeration clustering processes. The number of clusters is automatically identified. SpaRef is compared with a set of well-known clustering methods on compact airborne spectrographic imager (CASI) over an area in the Klompenwaard, The Netherlands. The clusters obtained show improved results. Applying SpaRef to multispectral chemical images would be a straight-forward step.     

Resolution and segmentation of hyperspectral biomedical images by multivariate curve resolution-alternating least squares

MCR-ALS is a resolution method that has been applied in many different fields, such as process analysis, environmental data and, recently, hyperspectral image analysis. In this context, the algorithm provides the distribution maps and the pure spectra of the image constituents from the sole information in the raw image measurement. Based on the distribution maps and spectra obtained, additional information can be easily derived, such as identification of constituents when libraries are available or quantitation within the image, expressed as constituent signal contribution. This work summarizes first the protocol followed for the resolution on two examples of kidney calculi, taken as representations of images with major and minor compounds, respectively. Image segmentation allows separating regions of images according to their pixel similarity and is also relevant in the biomedical field to differentiate healthy from non-healthy regions in tissues or to identify sample regions with distinct properties. Information on pixel similarity is enclosed not only in pixel spectra, but also in other smaller pixel representations, such as PCA scores. In this paper, we propose the use of MCR scores (concentration profiles) for segmentation purposes. K-means results obtained from different pixel representations of the data set are compared. The main advantages of the use of MCR scores are the interpretability of the class centroids and the compound-wise selection and preprocessing of the input information in the segmentation scheme.     

Image fusion combining SEM and ToF‐SIMS images

Image fusion allows for the combination of an image containing chemical information but low spatial resolution with a high‐spatial resolution image having little to no chemical information. The resulting hybrid image retains all the information from the chemically relevant original image, with improved spatial resolution allowing for visual inspection of the spatial correlations. In this research, images were obtained from two sample test grids: one of a copper electron microscope grid with a letter ‘A’ in the center (referred to below as the ‘A‐grid’), and the other a Tantalum and Silicon test grid from Cameca that had an inscribed letter ‘C’ (referred to below as the ‘Cameca grid’). These were obtained using scanning electron microscopy (SEM) and time‐of‐flight secondary ion mass spectrometry (ToF‐SIMS). Image fusion was implemented with the Munechika algorithm. The edge resolution of the resulting hybrid image was calculated compared with the edge resolution obtained for both the individual ToF‐SIMS and SEM images. The challenges of combining complimentary datasets from different instrumental analytical methods are discussed as well as the advantages of having a hybrid image. The distance across the edge for hybrid images of the A‐Grid and the Cameca grid were determined to be 21 µm and 8 µm, respectively. When these values were compared to the original ToF‐SIMS, SEM and optical microscopy measurements, the fused image had a spatial resolution nearly equal to that obtained in the SEM image for both samples. Copyright © 2014 John Wiley & Sons, Ltd.     

A density-based segmentation for 3D images,an application for X-ray micro-tomography

Density-based spatial clustering of applications with noise (DBSCAN) is an unsupervised classification algorithm which has been widely used in many areas with its simplicity and its ability to deal with hidden clusters of different sizes and shapes and with noise. However, the computational issue of the distance table and the non-stability in detecting the boundaries of adjacent clusters limit the application of the original algorithm to large datasets such as images. In this paper, the DBSCAN algorithm was revised and improved for image clustering and segmentation. The proposed clustering algorithm presents two major advantages over the original one. Firstly, the revised DBSCAN algorithm made it applicable for large 3D image dataset (often with millions of pixels) by using the coordinate system of the image data. Secondly, the revised algorithm solved the non-stability issue of boundary detection in the original DBSCAN. For broader applications, the image dataset can be ordinary 3D images or in general, it can also be a classification result of other type of image data e.g. a multivariate image.     

A review of the CLIP system for the quantitative analysis of two-dimensional electrophoresis gels

This paper reviews the CLIP image processing system for the complete analysis of two-dimensional electrophoresis images. The analysis problem for two-dimensional gel images can be broken down into three issues: segmentation of individual gel images, alignment and comparison of pairs of gel images, and information storage and retrieval. This paper describes these problems and reviews how the CLIP system handles each of them. Segmentation is the location and isolation of each protein spot on an individual gel image and also the extraction of individual spot data such as position, area and volume. There are three basic stages: background field correction, noise filtering, spot detection and information extraction. Alignment and comparison of gel images involves matching protein spots between two gels. This can be quite difficult because there is not a simple relationship which can transform one gel image onto another. The database issues concern storing all the information which has been obtained from the above operations such that retrieval of this information can be readily performed. The advantage of the CLIP system over others is speed of processing. CLIP series computers use one processor for every pixel of the camera image such that image processing algorithms run in parallel. The main disadvantage is in the cost of these machines. With the declining trend in the cost of parallel processors, these machines will become more and more viable alternatives. This papers reviews the algorithms for the analysis of two-dimensional gels. It is shown that CLIP is flexible enough to perform more than one type of algorithm for a particular operation.     

Denoising and multivariate analysis of time‐of‐flight SIMS images

Time‐of‐flight SIMS (ToF‐SIMS) imaging offers a modality for simultaneously visualizing the spatial distribution of different surface species. However, the utility of ToF‐SIMS datasets may be limited by their large size, degraded mass resolution and low ion counts per pixel. Through denoising and multivariate image analysis, regions of similar chemistries may be differentiated more readily in ToF‐SIMS image data. Three established denoising algorithms—down‐binning, boxcar and wavelet filtering—were applied to ToF‐SIMS images of different surface geometries and chemistries. The effect of these filters on the performance of principal component analysis (PCA) was evaluated in terms of the capture of important chemical image features in the principal component score images, the quality of the principal component score images and the ability of the principal components to explain the chemistries responsible for the image contrast. All filtering methods were found to improve the performance of PCA for all image datasets studied by improving capture of image features and producing principal component score images of higher quality than the unfiltered ion images. The loadings for filtered and unfiltered PCA models described the regions of chemical contrast by identifying peaks defining the regions of different surface chemistry. Down‐binning the images to increase pixel size and signal was the most effective technique to improve PCA performance. Copyright © 2003 John Wiley & Sons, Ltd.     

Automated correlation and classification of secondary ion mass spectrometry images using a k-means cluster method

We present a novel method for correlating and classifying ion-specific time-of-flight secondary ion mass spectrometry (ToF-SIMS) images within a multispectral dataset by grouping images with similar pixel intensity distributions. Binary centroid images are created by employing a k-means-based custom algorithm. Centroid images are compared to grayscale SIMS images using a newly developed correlation method that assigns the SIMS images to classes that have similar spatial (rather than spectral) patterns. Image features of both large and small spatial extent are identified without the need for image pre-processing, such as normalization or fixed-range mass-binning. A subsequent classification step tracks the class assignment of SIMS images over multiple iterations of increasing n classes per iteration, providing information about groups of images that have similar chemistry. Details are discussed while presenting data acquired with ToF-SIMS on a model sample of laser-printed inks. This approach can lead to the identification of distinct ion-specific chemistries for mass spectral imaging by ToF-SIMS, as well as matrix-assisted laser desorption ionization (MALDI), and desorption electrospray ionization (DESI).     

Vector potential photoelectron microscopy: hyperspectral image processing and super‐resolution applied to images of a Ca–Al alloy intermetallic phase distribution

Vector potential photoelectron microscopy (VPPEM) produces four dimensional hyper‐spectral data, and image processing is an integral part of the experimental technique. VPPEM is a new class of instrument, and this is the first discussion of some of the data reduction techniques that have been found effective. The point spread function of VPPEM is multidimensional with a high‐frequency component. Although this high‐frequency component is a small fraction of the total spatial response, images with good signal‐to‐noise can be spatially deconvolved to give super resolution images. The VPPEM data reduction process is illustrated by the analysis of the multiple intermetallic phases in a Ca–Al alloy. These phases have been imaged with better than 0.5‐μ spatial resolution. Not all the problems with the data reduction process have been satisfactorily dealt with, and lessons from this work will influence the design of future instruments. Copyright © 2014 John Wiley & Sons, Ltd.     

The inter‐ and intra‐operator variability in manual spot segmentation and its effect on spot quantitation in two‐dimensional electrophoresis analysis

Separation of complex mixtures of proteins by 2‐DE is a fundamental component of current proteomic technology. Quantitative analysis of the images generated by digitization of such gels is critical for identifying alterations in protein expression within a given biological system. Software packages are designed for this purpose. The accurate definition of protein spot boundaries, using a suitable method of image segmentation, is a key requirement for image analysis. It is often necessary for operators to intervene manually to correct mistakes in spot segmentation; therefore operator subjectivity and differences in ability can weaken the analysis. We estimated the error in spot quantification after manual spot segmentation, which was performed by different operators, using two different software packages. Our results clearly show that this operation was associated with significant inter‐ and intra‐variability and an overestimation of subsequent spot intensity, especially when spots were weak. For comparative studies, we suggest separately analysing spots which have been manually segmented by imposing a requirement for at least a threefold difference in spot intensity in addition to use of statistical tests.     

Constraining shape smoothness in multivariate curve resolution–alternating least squares

The resolution of complex multicomponent hyperspectral images with multivariate curve resolution–alternating least squares is mainly performed by using a limited number of constraints on the pure constituent distribution maps, such as non‐negativity or local constraints. This work proposes a constraint that works with the spatial information of the whole image and has been given the name shape smoothness constraint. Contrary to local constraints, shape smoothness constraint imposes a global character on the distribution map pattern. It uses two‐dimensional P‐splines to enforce smoothness of the global spatial features of the distribution maps generated within the alternating least squares procedure. This allows revealing main pattern(s) in the spatial data leaving high‐frequency signals, corresponding to fine‐scale structures in the image. This approach has been successfully applied on several simulated examples where imposing shape smoothness resulted in the recovery of the correct pattern for the image objects, which in turn provided more accurate distribution maps and spectral profiles. It was also shown that when information about the smoothness of the pattern(s) of the image constituents holds, the constraint can be a flexible and robust alternative for the resolution of real hyperspectral imaging data. Copyright © 2015 John Wiley & Sons, Ltd.     

MD-UNET: Multi-input dilated U-shape neural network for segmentation of bladder cancer

Accurate segmentation of the tumour area is crucial for the treatment and prognosis of patients with bladder cancer. However, the complex information from the MRI image poses an important challenge for us to accurately segment the lesion, for example, the high distinction among people, size of bladder variation and noise interference. Based on the above issues, we propose an MD-Unet network structure, which uses multi-scale images as the input of the network, and combines max-pooling with dilated convolution to increase the receptive field of the convolutional network. The results show that the proposed network can obtain higher precision than the existing models for the bladder cancer dataset. The MD-Unet can achieve state-of-art performance compared with other methods.     

Effects of including spatial information in clustering multivariate image data

Multivariate image data provide detailed information in variable and image space. Most traditional clustering methods are based on variable information only and ignore spatial information. A method based on both variable and spatial information could improve the results substantially.

In this review, we study the benefits and the pitfalls of including spatial information in chemometric clustering techniques. Spatial information is taken into account in initialization of clustering parameters, during cluster iterations by adjusting the similarity measure or at a post-processing step. We illustrate the effect of taking spatial information into account by a univariate synthetic data set and two real-world multivariate data sets. We show that methods that include neighboring pixel information in the clustering procedure improve the performance accuracy of the clustering in most cases. Homogeneous regions in the image are better recognized and the amount of noise is reduced by these methods.