Multimodal evaluation for medical image segmentation

This paper is a joint effort between five institutionsthat introduces several novel similarity measures andcombines them to carry out a multimodal segmentationevaluation. The new similarity measures proposed arebased on the location and the intensity values of themisclassified voxels as well as on the connectivity andthe boundaries of the segmented data. We showexperimentally that the combination of these measuresimprove the quality of the evaluation. The study that weshow here has been carried out using four differentsegmentation methods from four different labs applied toa MRI simulated dataset of the brain. We claim that ournew measures improve the robustness of the evaluation andprovides better understanding about the differencebetween segmentation methods.

A multidimensional segmentation evaluation for medical image data.

Evaluation of segmentation methods is a crucial aspect in image processing, especially in the medical imaging field, where small differences between segmented regions in the anatomy can be of paramount importance. Usually, segmentation evaluation is based on a measure that depends on the number of segmented voxels inside and outside of some reference regions that are called gold standards. Although some other measures have been also used, in this work we propose a set of new similarity measures, based on different features, such as the location and intensity values of the misclassified voxels, and the connectivity and the boundaries of the segmented data. Using the multidimensional information provided by these measures, we propose a new evaluation method whose results are visualized applying a Principal Component Analysis of the data, obtaining a simplified graphical method to compare different segmentation results. We have carried out an intensive study using several classic segmentation methods applied to a set of MRI simulated data of the brain with several noise and RF inhomogeneity levels, and also to real data, showing that the new measures proposed here and the results that we have obtained from the multidimensional evaluation, improve the robustness of the evaluation and provides better understanding about the difference between segmentation methods.

MSight: an image analysis software for liquid chromatography-mass spectrometry.

Images obtained from high-throughput mass spectrometry (MS) contain information that remains hidden when looking at a single spectrum at a time. Image processing of liquid chromatography-MS datasets can be extremely useful for quality control, experimental monitoring and knowledge extraction. The importance of imaging in differential analysis of proteomic experiments has already been established through two-dimensional gels and can now be foreseen with MS images. We present MSight, a new software designed to construct and manipulate MS images, as well as to facilitate their analysis and comparison.

A study of the cambial zone and conductive phloem of common beech (Fagus sylvatica L.) using an image analysis method. I. Influence of tree age on the structure

Quantification is a major problem when using histology to study the influence of ecological factors on tree structure. This paper presents a method to prepare and to analyse transverse sections of cambial zone and of conductive phloem in bark samples. The following paper (II) presents the automated measurement procedure. Part I here describes and discusses the preparation method, and the influence of tree age on the observed structure. Highly contrasted images of samples extracted at breast height during dormancy were analysed with an automatic image analyser. Between three young (38 years) and three old (147 years) trees, age-related differences were identified by size and shape parameters, at both cell and tissue levels. In the cambial zone, older trees had larger and more rectangular fusiform initials. In the phloem, sieve tubes were also larger, but their shape did not change and the area for sap conduction was similar in both categories. Nevertheless, alterations were limited, and demanded statistical analysis to be identified and ascertained. The physiological implications of the structural changes are discussed.

Weighted Shape-based Averaging with Neighborhood Prior Model for Multiple Atlas Fusion-based Medical Image Segmentation

In medical imaging, merging automated segmentations obtained from multiple atlases has become a standard practice for improving the accuracy. In this letter, we propose two new fusion methods: "Global Weighted Shape-Based Averaging" (GWSBA) and "Local Weighted Shape-Based Averaging" (LWSBA). These methods extend the well known Shape-Based Averaging (SBA) by additionally incorporating the similarity information between the reference (i.e., atlas) images and the target image to be segmented. We also propose a new spatially-varying similarity-weighted neighborhood prior model, and an edge-preserving smoothness term that can be used with many of the existing fusion methods. We first present our new Markov Random Field (MRF) based fusion framework that models the above mentioned information. The proposed methods are evaluated in the context of segmentation of lymph nodes in the head and neck 3D CT images, and they resulted in more accurate segmentations compared to the existing SBA.

A 2D/3D image analysis system to track fluorescently labeled structures in rod-shaped cells: application to measure spindle pole asymmetry during mitosis.

BACKGROUND: The yeast Schizosaccharomyces pombe is frequently used as a model for studying the cell cycle. The cells are rod-shaped and divide by medial fission. The process of cell division, or cytokinesis, is controlled by a network of signaling proteins called the Septation Initiation Network (SIN); SIN proteins associate with the SPBs during nuclear division (mitosis). Some SIN proteins associate with both SPBs early in mitosis, and then display strongly asymmetric signal intensity at the SPBs in late mitosis, just before cytokinesis. This asymmetry is thought to be important for correct regulation of SIN signaling, and coordination of cytokinesis and mitosis. In order to study the dynamics of organelles or large protein complexes such as the spindle pole body (SPB), which have been labeled with a fluorescent protein tag in living cells, a number of the image analysis problems must be solved; the cell outline must be detected automatically, and the position and signal intensity associated with the structures of interest within the cell must be determined. RESULTS: We present a new 2D and 3D image analysis system that permits versatile and robust analysis of motile, fluorescently labeled structures in rod-shaped cells. We have designed an image analysis system that we have implemented as a user-friendly software package allowing the fast and robust image-analysis of large numbers of rod-shaped cells. We have developed new robust algorithms, which we combined with existing methodologies to facilitate fast and accurate analysis. Our software permits the detection and segmentation of rod-shaped cells in either static or dynamic (i.e. time lapse) multi-channel images. It enables tracking of two structures (for example SPBs) in two different image channels. For 2D or 3D static images, the locations of the structures are identified, and then intensity values are extracted together with several quantitative parameters, such as length, width, cell orientation, background fluorescence and the distance between the structures of interest. Furthermore, two kinds of kymographs of the tracked structures can be established, one representing the migration with respect to their relative position, the other representing their individual trajectories inside the cell. This software package, called "RodCellJ", allowed us to analyze a large number of S. pombe cells to understand the rules that govern SIN protein asymmetry. CONCLUSIONS: "RodCell" is freely available to the community as a package of several ImageJ plugins to simultaneously analyze the behavior of a large number of rod-shaped cells in an extensive manner. The integration of different image-processing techniques in a single package, as well as the development of novel algorithms does not only allow to speed up the analysis with respect to the usage of existing tools, but also accounts for higher accuracy. Its utility was demonstrated on both 2D and 3D static and dynamic images to study the septation initiation network of the yeast Schizosaccharomyces pombe. More generally, it can be used in any kind of biological context where fluorescent-protein labeled structures need to be analyzed in rod-shaped cells. AVAILABILITY: RodCellJ is freely available under http://bigwww.epfl.ch/algorithms.html, (after acceptance of the publication).

On the convergence of EM-like algorithms for image segmentation using Markov random fields.

Inference of Markov random field images segmentation models is usually performed using iterative methods which adapt the well-known expectation-maximization (EM) algorithm for independent mixture models. However, some of these adaptations are ad hoc and may turn out numerically unstable. In this paper, we review three EM-like variants for Markov random field segmentation and compare their convergence properties both at the theoretical and practical levels. We specifically advocate a numerical scheme involving asynchronous voxel updating, for which general convergence results can be established. Our experiments on brain tissue classification in magnetic resonance images provide evidence that this algorithm may achieve significantly faster convergence than its competitors while yielding at least as good segmentation results.

Sparse regularization for fiber ODF reconstruction: from the suboptimality of ℓ2 and ℓ1 priors to ℓ0.

Diffusion MRI is a well established imaging modality providing a powerful way to probe the structure of the white matter non-invasively. Despite its potential, the intrinsic long scan times of these sequences have hampered their use in clinical practice. For this reason, a large variety of methods have been recently proposed to shorten the acquisition times. Among them, spherical deconvolution approaches have gained a lot of interest for their ability to reliably recover the intra-voxel fiber configuration with a relatively small number of data samples. To overcome the intrinsic instabilities of deconvolution, these methods use regularization schemes generally based on the assumption that the fiber orientation distribution (FOD) to be recovered in each voxel is sparse. The well known Constrained Spherical Deconvolution (CSD) approach resorts to Tikhonov regularization, based on an ℓ(2)-norm prior, which promotes a weak version of sparsity. Also, in the last few years compressed sensing has been advocated to further accelerate the acquisitions and ℓ(1)-norm minimization is generally employed as a means to promote sparsity in the recovered FODs. In this paper, we provide evidence that the use of an ℓ(1)-norm prior to regularize this class of problems is somewhat inconsistent with the fact that the fiber compartments all sum up to unity. To overcome this ℓ(1) inconsistency while simultaneously exploiting sparsity more optimally than through an ℓ(2) prior, we reformulate the reconstruction problem as a constrained formulation between a data term and a sparsity prior consisting in an explicit bound on the ℓ(0)norm of the FOD, i.e. on the number of fibers. The method has been tested both on synthetic and real data. Experimental results show that the proposed ℓ(0) formulation significantly reduces modeling errors compared to the state-of-the-art ℓ(2) and ℓ(1) regularization approaches.

Direct three-dimensional myocardial strain tensor quantification and tracking using zHARP.

Images of myocardial strain can be used to diagnose heart disease, plan and monitor treatment, and to learn about cardiac structure and function. Three-dimensional (3D) strain is typically quantified using many magnetic resonance (MR) images obtained in two or three orthogonal planes. Problems with this approach include long scan times, image misregistration, and through-plane motion. This article presents a novel method for calculating cardiac 3D strain using a stack of two or more images acquired in only one orientation. The zHARP pulse sequence encodes in-plane motion using MR tagging and out-of-plane motion using phase encoding, and has been previously shown to be capable of computing 3D displacement within a single image plane. Here, data from two adjacent image planes are combined to yield a 3D strain tensor at each pixel; stacks of zHARP images can be used to derive stacked arrays of 3D strain tensors without imaging multiple orientations and without numerical interpolation. The performance and accuracy of the method is demonstrated in vitro on a phantom and in vivo in four healthy adult human subjects.

Active deformation fields: dense deformation field estimation for atlas-based segmentation using the active contour framework.

This paper presents a new and original variational framework for atlas-based segmentation. The proposed framework integrates both the active contour framework, and the dense deformation fields of optical flow framework. This framework is quite general and encompasses many of the state-of-the-art atlas-based segmentation methods. It also allows to perform the registration of atlas and target images based on only selected structures of interest. The versatility and potentiality of the proposed framework are demonstrated by presenting three diverse applications: In the first application, we show how the proposed framework can be used to simulate the growth of inconsistent structures like a tumor in an atlas. In the second application, we estimate the position of nonvisible brain structures based on the surrounding structures and validate the results by comparing with other methods. In the final application, we present the segmentation of lymph nodes in the Head and Neck CT images, and demonstrate how multiple registration forces can be used in this framework in an hierarchical manner.

Towards a diffusion image processing validation and accuracy prediction framework

Validation is the main bottleneck preventing theadoption of many medical image processing algorithms inthe clinical practice. In the classical approach,a-posteriori analysis is performed based on someobjective metrics. In this work, a different approachbased on Petri Nets (PN) is proposed. The basic ideaconsists in predicting the accuracy that will result froma given processing based on the characterization of thesources of inaccuracy of the system. Here we propose aproof of concept in the scenario of a diffusion imaginganalysis pipeline. A PN is built after the detection ofthe possible sources of inaccuracy. By integrating thefirst qualitative insights based on the PN withquantitative measures, it is possible to optimize the PNitself, to predict the inaccuracy of the system in adifferent setting. Results show that the proposed modelprovides a good prediction performance and suggests theoptimal processing approach.

Chronic exercise preserves lean muscle mass in masters athletes.

Aging is commonly associated with a loss of muscle mass and strength, resulting in falls, functional decline, and the subjective feeling of weakness. Exercise modulates the morbidities of muscle aging. Most studies, however, have examined muscle-loss changes in sedentary aging adults. This leaves the question of whether the changes that are commonly associated with muscle aging reflect the true physiology of muscle aging or whether they reflect disuse atrophy. This study evaluated whether high levels of chronic exercise prevents the loss of lean muscle mass and strength experienced in sedentary aging adults. A cross-section of 40 high-level recreational athletes ("masters athletes") who were aged 40 to 81 years and trained 4 to 5 times per week underwent tests of health/activity, body composition, quadriceps peak torque (PT), and magnetic resonance imaging of bilateral quadriceps. Mid-thigh muscle area, quadriceps area (QA), subcutaneous adipose tissue, and intramuscular adipose tissue were quantified in magnetic resonance imaging using medical image processing, analysis, and visualization software. One-way analysis of variance was used to examine age group differences. Relationships were evaluated using Spearman correlations. Mid-thigh muscle area (P = 0.31) and lean mass (P = 0.15) did not increase with age and were significantly related to retention of mid-thigh muscle area (P < 0.0001). This occurred despite an increase in total body fat percentage (P = 0.003) with age. Mid-thigh muscle area (P = 0.12), QA (P = 0.17), and quadriceps PT did not decline with age. Specific strength (strength per QA) did not decline significantly with age (P = 0.06). As muscle area increased, PT increased significantly (P = 0.008). There was not a significant relationship between intramuscular adipose tissue (P = 0.71) or lean mass (P = 0.4) and PT. This study contradicts the common observation that muscle mass and strength decline as a function of aging alone. Instead, these declines may signal the effect of chronic disuse rather than muscle aging. Evaluation of masters athletes removes disuse as a confounding variable in the study of lower-extremity function and loss of lean muscle mass. This maintenance of muscle mass and strength may decrease or eliminate the falls, functional decline, and loss of independence that are commonly seen in aging adults.

Pore space analysis of beech wood: The vessel network

Water transport in wood is vital for the survival of trees. With synchrotron radiation X-ray tomographic microscopy (SRXTM), it has become possible to characterize and quantify the three-dimensional (3D) network formed by vessels that are responsible for longitudinal transport. In the present study, the spatial size dependence of vessels and the organization inside single growth rings in terms of vessel-induced porosity was studied by SRXTM. Network characteristics, such as connectivity, were deduced by digital image analysis from the processed tomographic data and related to known complex network topologies.

Intraobserver and interobserver variability of ascending aorta diameter as assessed with ECG-gated MDCT: automatic versus manual measurements

Purpose: Recently morphometric measurements of the ascending aorta have been done with ECG-gated MDCT to help the development of future endovascular therapies (TCT) [1]. However, the variability of these measurements remains unknown. It will be interesting to know the impact of CAD (computer aided diagnosis) with automated segmentation of the vessel and automatic measurements of diameter on the management of ascending aorta aneurysms. Methods and Materials: Thirty patients referred for ECG-gated CT thoracic angiography (64-row CT scanner) were evaluated. Measurements of the maximum and minimum ascending aorta diameters were obtained automatically with a commercially available CAD and semi-manually by two observers separately. The CAD algorithms segment the iv-enhanced lumen of the ascending aorta into perpendicular planes along the centreline. The CAD then determines the largest and the smallest diameters. Both observers repeated the automatic measurements and the semimanual measurements during a different session at least one month after the first measurements. The Bland and Altman method was used to study the inter/intraobserver variability. A Wilcoxon signed-rank test was also used to analyse differences between observers. Results: Interobserver variability for semi-manual measurements between the first and second observers was between 1.2 to 1.0 mm for maximal and minimal diameter, respectively. Intraobserver variability of each observer ranged from 0.8 to 1.2 mm, the lowest variability being produced by the more experienced observer. CAD variability could be as low as 0.3 mm, showing that it can perform better than human observers. However, when used in nonoptimal conditions (streak artefacts from contrast in the superior vena cava or weak lumen enhancement), CAD has a variability that can be as high as 0.9 mm, reaching variability of semi-manual measurements. Furthermore, there were significant differences between both observers for maximal and minimal diameter measurements (p<0.001). There was also a significant difference between the first observer and CAD for maximal diameter measurements with the former underestimating the diameter compared to the latter (p<0.001). As for minimal diameters, they were higher when measured by the second observer than when measured by CAD (p<0.001). Neither the difference of mean minimal diameter between the first observer and CAD nor the difference of mean maximal diameter between the second observer and CAD was significant (p=0.20 and 0.06, respectively). Conclusion: CAD algorithms can lessen the variability of diameter measurements in the follow-up of ascending aorta aneurysms. Nevertheless, in non-optimal conditions, it may be necessary to correct manually the measurements. Improvements of the algorithms will help to avoid such a situation.

3D Automated Lung Nodule Segmentation in HRCT

A fully-automated 3D image analysis method is proposed to segment lung nodules in HRCT. A specific gray-level mathematical morphology operator, the SMDC-connection cost, acting in the 3D space of the thorax volume is defined in order to discriminate lung nodules from other dense (vascular) structures. Applied to clinical data concerning patients with pulmonary carcinoma, the proposed method detects isolated, juxtavascular and peripheral nodules with sizes ranging from 2 to 20 mm diameter. The segmentation accuracy was objectively evaluated on real and simulated nodules. The method showed a sensitivity and a specificity ranging from 85% to 97% and from 90% to 98%, respectively.