Three-dimensional multi-scale line filter for segmentation and visualization of curvilinear structures in medical images

Open source tools for fluorescent imaging

As microscopy becomes increasingly automated and imaging expands in the spatial and time dimensions, quantitative analysis tools for fluorescent imaging are becoming critical to remove both bottlenecks in throughput as well as fully extract and exploit the information contained in the imaging. In recent years there has been a flurry of activity in the development of bio-image analysis tools and methods with the result that there are now many high-quality, well-documented, and well-supported open source bio-image analysis projects with large user bases that cover essentially every aspect from image capture to publication. These open source solutions are now providing a viable alternative to commercial solutions. More importantly, they are forming an interoperable and interconnected network of tools that allow data and analysis methods to be shared between many of the major projects. Just as researchers build on, transmit, and verify knowledge through publication, open source analysis methods and software are creating a foundation that can be built upon, transmitted, and verified. Here we describe many of the major projects, their capabilities, and features. We also give an overview of the current state of open source software for fluorescent microscopy analysis and the many reasons to use and develop open source methods.

Automated cavity detection of infectious pulmonary tuberculosis in chest radiographs

The presence of cavities in the upper lung zones is an important indicator of highly infectious Tuberculosis (TB). Diagnoses performed by the radiologists are labor intensive and of high inter-reader variation. After analyzing the existing computer-aided detection techniques, we propose an fully automated TB cavity detection system which combines a 2D Gaussian-model-based template matching (GTM) for candidates detection with Hessian-matrix-based image enhancement (HIE) for the following cavity segmentation and feature extraction. Experimental results demonstrate that our approach outperforms the existing TB cavity detection technique with higher accuracy and lower false positive rate.

A mutual information approach to automate identification of neuronal clusters in Drosophila brain images

Mapping neural circuits can be accomplished by labeling a small number of neural structures per brain, and then combining these structures across multiple brains. This sparse labeling method has been particularly effective in Drosophila melanogaster, where clonally related clusters of neurons derived from the same neural stem cell (neuroblast clones) are functionally related and morphologically highly stereotyped across animals. However identifying these neuroblast clones (approximately 180 per central brain hemisphere) manually remains challenging and time consuming. Here, we take advantage of the stereotyped nature of neural circuits in Drosophila to identify clones automatically, requiring manual annotation of only an initial, smaller set of images. Our procedure depends on registration of all images to a common template in conjunction with an image processing pipeline that accentuates and segments neural projections and cell bodies. We then measure how much information the presence of a cell body or projection at a particular location provides about the presence of each clone. This allows us to select a highly informative set of neuronal features as a template that can be used to detect the presence of clones in novel images. The approach is not limited to a specific labeling strategy and can be used to identify partial (e.g., individual neurons) as well as complete matches. Furthermore this approach could be generalized to studies of neural circuits in other organisms.

3D segmentation of keratin intermediate filaments in confocal laser scanning microscopy

In this paper, we propose and compare different methods for the 3D segmentation of keratin intermediate filaments (KFs) in images acquired using confocal laser scanning microscopy (CLSM). KFs are elastic cables forming a complex scaffolding within epithelial cells. They are involved in many basic cell functions. To understand the mechanisms of filament formation and network organisation under physiological and pathological conditions, quantitative measurements of dynamic network alterations are essential. Segmenting KFs is a key component for analyzing their dynamic and biomechanical properties. KFs were labeled with fluorescent keratins to allow high resolution imaging of network dynamics in native cells. Our segmentation methods follow the principle of ridge enhancement filtering and subsequent centerline extraction. The evaluation of the methods is two-fold: (i) We develop synthetic data that exhibit the characteristics of real CLSM data to evaluate the precision of the different methods in terms of centerline localisation and (ii) we perform a connected component analysis on the segmentation results of real KF data to assess whether the connectivity of highly complex networks is being preserved by the segmentation. Our evaluation shows that in the presence of strong noise and despite the highly anisotropic spatial resolution of CLSM images the proposed method is able to accurately localize the centerlines of the KFs and to preserve the KF networks' connectivity. Taken together this is a strong indicator that also the network topology is being preserved.

An automated computational framework for retinal vascular network labeling and branching order analysis

Changes in retinal vascular morphology are well known as predictive clinical signs of many diseases such as hypertension, diabetes and so on. Computer-aid image processing and analysis for retinal vessels in fundus images are effective and efficient in clinical diagnosis instead of tedious manual labeling and measurement. An automated computational framework for retinal vascular network labeling and analysis is presented in this work. The framework includes 1) detecting and locating the optic disc; 2) tracking the vessel centerline from detected seed points and linking the breaks after tracing; 3) extracting all the retinal vascular trees and identifying all the significant points; and 4) classifying terminal points into starting points and ending points based on the information of optic disc location, and finally assigning branch order for each extracted vascular tree in the image. All the modules in the framework are fully automated. Based on the results, morphological analysis is then applied to achieve geometrical and topological features based on branching order for one individual vascular tree or for the vascular network through the retinal vascular network in the images. Validation and experiments on the public DRIVE database have demonstrated that the proposed framework is a novel approach to analyze and study the vascular network pattern, and may offer new insights to the diagnosis of retinopathy.

Automated reconstruction of neuronal morphology based on local geometrical and global structural models

Digital reconstruction of neurons from microscope images is an important and challenging problem in neuroscience. In this paper, we propose a model-based method to tackle this problem. We first formulate a model structure, then develop an algorithm for computing it by carefully taking into account morphological characteristics of neurons, as well as the image properties under typical imaging protocols. The method has been tested on the data sets used in the DIADEM competition and produced promising results for four out of the five data sets.

Contrast-independent curvilinear structure detection in biomedical images

Many biomedical applications require detection of curvilinear structures in images and would benefit from automatic or semiautomatic segmentation to allow high-throughput measurements. Here, we propose a contrast-independent approach to identify curvilinear structures based on oriented phase congruency, i.e., the phase congruency tensor (PCT). We show that the proposed method is largely insensitive to intensity variations along the curve and provides successful detection within noisy regions. The performance of the PCT is evaluated by comparing it with state-of-the-art intensity-based approaches on both synthetic and real biological images.

Diverse populations of intrinsic cholinergic interneurons in the mouse olfactory bulb

Cholinergic activities affect olfactory bulb (OB) information processing and associated learning and memory. However, the presence of intrinsic cholinergic interneurons in the OB remains controversial. As a result, morphological and functional properties of these cells are largely undetermined. We characterized cholinergic interneurons using transgenic mice that selectively mark choline acetyltransferase (ChAT)-expressing cells and immunolabeling. We found a significant number of intrinsic cholinergic interneurons in the OB. These interneurons reside primarily in the glomerular layer (GL) and external plexiform layer (EPL) and exhibit diverse distribution patterns of nerve processes, indicating functional heterogeneity. Further, we found these neurons express ChAT and vesicular acetylcholine transporter (VAChT), but do not immunoreact to glutamatergic, GABAergic or dopaminergic markers and are distinct from calretinin-expressing interneurons. Interestingly, the cholinergic population partially overlaps with the calbindin D28K-expressing interneuron population, revealing the neurotransmitter identity of this sub-population. Additionally, we quantitatively determined the density of VAChT labeled cholinergic nerve fibers in various layers of the OB, as well as the intensity of VAChT immunoreactivity within the GL, suggesting primary sites of cholinergic actions. Taken together, our results provide clear evidence showing the presence of a significant number of cholinergic interneurons and that these morphologically and distributionally diverse interneurons make up complex local cholinergic networks in the OB. Thus, our results suggest that olfactory information processing is modulated by dual cholinergic systems of local interneuron networks and centrifugal projections.

The importance of combinatorial gene expression in early Mammalian thalamic patterning and thalamocortical axonal guidance

The thalamus is essential for sensory perception. In mammals, work on the mouse has taught us most of what we know about how it develops and connects to the cortex. The mature thalamus of all mammalian species comprises numerous anatomically distinct collections of neurons called nuclei that differ in function, connectivity, and molecular constitution. At the time of its initial appearance as a distinct structure following neural tube closure, the thalamus is already patterned by the regional expression of numerous regulatory genes. This patterning, which lays down the blueprint for later development of thalamic nuclei, predates the development of thalamocortical projections. In this review we apply novel analytical methods to gene expression data available in the Allen Developing Mouse Brain Atlas to highlight the complex organized molecular heterogeneity already present among cells in the thalamus from the earliest stages at which it contains differentiating neurons. This early patterning is likely to invest in axons growing from different parts of the thalamus the ability to navigate in an ordered way to their appropriate area in the cerebral cortex. We review the mechanisms and cues that thalamic axons use, encounter, and interpret to attain the cortex. Mechanisms include guidance by previously generated guidepost cells, such as those in the subpallium that maintain thalamic axonal order and direction, and axons such as those of reciprocal projections from intermediate structures or from the cortex itself back toward the thalamus. We show how thalamocortical pathfinding involves numerous guidance cues operating at a series of steps along their route. We stress the importance of the combinatorial actions of multiple genes for the development of the numerous specific identities and functions of cells in this exquisitely complex system and their orderly innervation of the cortex.

Automated image-based colon cleansing for laxative-free CT colonography computer-aided polyp detection

PURPOSE

To evaluate the performance of a computer-aided detection (CAD) system for detecting colonic polyps at noncathartic computed tomography colonography (CTC) in conjunction with an automated image-based colon cleansing algorithm.

METHODS

An automated colon cleansing algorithm was designed to detect and subtract tagged-stool, accounting for heterogeneity and poor tagging, to be used in conjunction with a colon CAD system. The method is locally adaptive and combines intensity, shape, and texture analysis with probabilistic optimization. CTC data from cathartic-free bowel preparation were acquired for testing and training the parameters. Patients underwent various colonic preparations with barium or Gastroview in divided doses over 48 h before scanning. No laxatives were administered and no dietary modifications were required. Cases were selected from a polyp-enriched cohort and included scans in which at least 90% of the solid stool was visually estimated to be tagged and each colonic segment was distended in either the prone or supine view. The CAD system was run comparatively with and without the stool subtraction algorithm.

RESULTS

The dataset comprised 38 CTC scans from prone and/or supine scans of 19 patients containing 44 polyps larger than 10 mm (22 unique polyps, if matched between prone and supine scans). The results are robust on fine details around folds, thin-stool linings on the colonic wall, near polyps and in large fluid/stool pools. The sensitivity of the CAD system is 70.5% per polyp at a rate of 5.75 false positives/scan without using the stool subtraction module. This detection improved significantly (p = 0.009) after automated colon cleansing on cathartic-free data to 86.4% true positive rate at 5.75 false positives/scan.

CONCLUSIONS

An automated image-based colon cleansing algorithm designed to overcome the challenges of the noncathartic colon significantly improves the sensitivity of colon CAD by approximately 15%.

Automated measurement of nerve fiber density using line intensity scan analysis

Quantification of nerve fibers in peripheral and central nervous systems is important for the understanding of neuronal function, organization and pathological changes. However, current methods to quantify nerve fibers are resource-intensive and often provide an indirect measurement of nerve fiber density. Here, we describe an automated and efficient method for nerve fiber quantification, which we developed by making use of widely available software and analytical techniques, including Hessian-based feature extraction in NIH ImageJ and line intensity scan analysis. The combined use of these analytical tools through an automated routine enables reliable detection and quantification of nerve fibers from low magnification, non-uniformly labeled epifluorescence images. This allows for time-efficient determination of nerve density and also comparative analysis in large brain structures, such as hippocampus or between various regions of neural circuitry. Using this method, we have obtained accurate measurements of cholinergic fiber density in hippocampus and a large area of cortex in mouse brain sections immunolabeled with an antibody against the vesicular acetylcholine transporter (VAChT). The density values are comparable among animals tested, showing a high degree of reproducibility. Because our method can be performed at relatively low cost and in large tissue sections where nerve fibers can be labeled by various antibodies or visualized by expression of reporter proteins, such as green fluorescent protein in transgenic mice, we expect our method to be broadly useful in both research and clinical investigation. To our knowledge, this is the first method to reliably quantify nerve fibers through a rapid and automated protocol.

Magnetic Resonance Image Analysis for Brain CAD Systems with Machine Learning

This chapter describes the image analysis for brain Computer-Aided Diagnosis (CAD) systems with machine learning techniques, which could assist radiologists in the detection of such brain diseases as asymptomatic unruptured aneurysms, Alzheimer’s Disease (AD), vascular dementia, and Multiple Sclerosis (MS) by magnetic resonance imaging. Image analysis in CAD systems consists of image enhancement, initial detection, and image feature extraction, including segmentation. In addition, the authors review the classification of true and false positives using machine learning techniques, as well as the evaluation methods and development cycle for CAD systems.

Automated coronary artery tree extraction in coronary CT angiography using a multiscale enhancement and dynamic balloon tracking (MSCAR-DBT) method

RATIONAL AND OBJECTIVES

To evaluate our prototype method for segmentation and tracking of the coronary arterial tree, which is the foundation for a computer-aided detection (CADe) system to be developed to assist radiologists in detecting non-calcified plaques in coronary CT angiography (cCTA) scans.

MATERIALS AND METHODS

The heart region was first extracted by a morphological operation and an adaptive thresholding method based on expectation-maximization (EM) estimation. The vascular structures within the heart region were enhanced and segmented using a multiscale coronary response (MSCAR) method that combined 3D multiscale filtering, analysis of the eigenvalues of Hessian matrices and EM estimation segmentation. After the segmentation of vascular structures, the coronary arteries were tracked by a 3D dynamic balloon tracking (DBT) method. The DBT method started at two manually identified seed points located at the origins of the left and right coronary arteries (LCA and RCA) for extraction of the arterial trees. The coronary arterial trees of a data set containing 20 ECG-gated contrast-enhanced cCTA scans were extracted by our MSCAR-DBT method and a clinical GE Advantage workstation. Two experienced thoracic radiologists visually examined the coronary arteries on the original cCTA scans and the rendered volume of segmented vessels to count the untracked false-negative (FN) segments and false positives (FPs) for both methods.

RESULTS

For the visible coronary arterial segments in the 20 cases, the radiologists identified that 25 segments were missed by our MSCAR-DBT method, ranging from 0 to 5 FN segments in individual cases, and that 55 artery segments were missed by the GE software, ranging from 0 to 7 FN segments in individual cases. 19 and 15 FPs were identified in our and the GE coronary trees, ranging from 0 to 4 FPs for both methods in individual cases, respectively.

CONCLUSION

The preliminary study demonstrates the feasibility of our MSCAR-DBT method for segmentation and tracking coronary artery trees. The results indicated that both our method and GE software can extract coronary artery trees reasonably well and the performance of our method is superior to that of GE software in this small data set. Further studies are underway to develop methods for improvement of the segmentation and tracking accuracy.

Automatic registration between 3D intra-operative ultrasound and pre-operative CT images of the liver based on robust edge matching

The registration of a three-dimensional (3D) ultrasound (US) image with a computed tomography (CT) or magnetic resonance image is beneficial in various clinical applications such as diagnosis and image-guided intervention of the liver. However, conventional methods usually require a time-consuming and inconvenient manual process for pre-alignment, and the success of this process strongly depends on the proper selection of initial transformation parameters. In this paper, we present an automatic feature-based affine registration procedure of 3D intra-operative US and pre-operative CT images of the liver. In the registration procedure, we first segment vessel lumens and the liver surface from a 3D B-mode US image. We then automatically estimate an initial registration transformation by using the proposed edge matching algorithm. The algorithm finds the most likely correspondences between the vessel centerlines of both images in a non-iterative manner based on a modified Viterbi algorithm. Finally, the registration is iteratively refined on the basis of the global affine transformation by jointly using the vessel and liver surface information. The proposed registration algorithm is validated on synthesized datasets and 20 clinical datasets, through both qualitative and quantitative evaluations. Experimental results show that automatic registration can be successfully achieved between 3D B-mode US and CT images even with a large initial misalignment.

A computational tool for quantitative analysis of vascular networks

Angiogenesis is the generation of mature vascular networks from pre-existing vessels. Angiogenesis is crucial during the organism' development, for wound healing and for the female reproductive cycle. Several murine experimental systems are well suited for studying developmental and pathological angiogenesis. They include the embryonic hindbrain, the post-natal retina and allantois explants. In these systems vascular networks are visualised by appropriate staining procedures followed by microscopical analysis. Nevertheless, quantitative assessment of angiogenesis is hampered by the lack of readily available, standardized metrics and software analysis tools. Non-automated protocols are being used widely and they are, in general, time--and labour intensive, prone to human error and do not permit computation of complex spatial metrics. We have developed a light-weight, user friendly software, AngioTool, which allows for quick, hands-off and reproducible quantification of vascular networks in microscopic images. AngioTool computes several morphological and spatial parameters including the area covered by a vascular network, the number of vessels, vessel length, vascular density and lacunarity. In addition, AngioTool calculates the so-called "branching index" (branch points/unit area), providing a measurement of the sprouting activity of a specimen of interest. We have validated AngioTool using images of embryonic murine hindbrains, post-natal retinas and allantois explants. AngioTool is open source and can be downloaded free of charge.

Decreased brain venous vasculature visibility on susceptibility-weighted imaging venography in patients with multiple sclerosis is related to chronic cerebrospinal venous insufficiency

Background

The potential pathogenesis between the presence and severity of chronic cerebrospinal venous insufficiency (CCSVI) and its relation to clinical and imaging outcomes in brain parenchyma of multiple sclerosis (MS) patients has not yet been elucidated. The aim of the study was to investigate the relationship between CCSVI, and altered brain parenchyma venous vasculature visibility (VVV) on susceptibility-weighted imaging (SWI) in patients with MS and in sex- and age-matched healthy controls (HC).

Methods

59 MS patients, 41 relapsing-remitting and 18 secondary-progressive, and 33 HC were imaged on a 3T GE scanner using pre- and post-contrast SWI venography. The presence and severity of CCSVI was determined using extra-cranial and trans-cranial Doppler criteria. Apparent total venous volume (ATVV), venous intracranial fraction (VIF) and average distance-from-vein (DFV) were calculated for various vein mean diameter categories: < .3 mm, .3-.6 mm, .6-.9 mm and > .9 mm.

Results

CCSVI criteria were fulfilled in 79.7% of MS patients and 18.2% of HC (p < .0001). Patients with MS showed decreased overall ATVV, ATVV of veins with a diameter < .3 mm, and increased DFV compared to HC (all p < .0001). Subjects diagnosed with CCSVI had significantly increased DFV (p < .0001), decreased overall ATVV and ATVV of veins with a diameter < .3 mm (p < .003) compared to subjects without CCSVI. The severity of CCSVI was significantly related to decreased VVV in MS (p < .0001) on pre- and post-contrast SWI, but not in HC.

Conclusions

MS patients with higher number of venous stenoses, indicative of CCSVI severity, showed significantly decreased venous vasculature in the brain parenchyma. The pathogenesis of these findings has to be further investigated, but they suggest that reduced metabolism and morphological changes of venous vasculature may be taking place in patients with MS.

Performance of a fully automatic lesion detection system for breast DCE-MRI

PURPOSE

To describe and test a new fully automatic lesion detection system for breast DCE-MRI.

MATERIALS AND METHODS

Studies were collected from two institutions adopting different DCE-MRI sequences, one with and the other one without fat-saturation. The detection pipeline consists of (i) breast segmentation, to identify breast size and location; (ii) registration, to correct for patient movements; (iii) lesion detection, to extract contrast-enhanced regions using a new normalization technique based on the contrast-uptake of mammary vessels; (iv) false positive (FP) reduction, to exclude contrast-enhanced regions other than lesions. Detection rate (number of system-detected malignant and benign lesions over the total number of lesions) and sensitivity (system-detected malignant lesions over the total number of malignant lesions) were assessed. The number of FPs was also assessed.

RESULTS

Forty-eight studies with 12 benign and 53 malignant lesions were evaluated. Median lesion diameter was 6 mm (range, 5-15 mm) for benign and 26 mm (range, 5-75 mm) for malignant lesions. Detection rate was 58/65 (89%; 95% confidence interval [CI] 79%-95%) and sensitivity was 52/53 (98%; 95% CI 90%-99%). Mammary median FPs per breast was 4 (1st-3rd quartiles 3-7.25).

CONCLUSION

The system showed promising results on MR datasets obtained from different scanners producing fat-sat or non-fat-sat images with variable temporal and spatial resolution and could potentially be used for early diagnosis and staging of breast cancer to reduce reading time and to improve lesion detection. Further evaluation is needed before it may be used in clinical practice.

Automated tracing of neurites from light microscopy stacks of images

Automating the process of neural circuit reconstruction on a large-scale is one of the foremost challenges in the field of neuroscience. In this study we examine the methodology for circuit reconstruction from three-dimensional light microscopy (LM) stacks of images. We show how the minimal error-rate of an ideal reconstruction procedure depends on the density of labeled neurites, giving rise to the fundamental limitation of an LM based approach for neural circuit research. Circuit reconstruction procedures typically involve steps related to neuron labeling and imaging, and subsequent image pre-processing and tracing of neurites. In this study, we focus on the last step--detection of traces of neurites from already pre-processed stacks of images. Our automated tracing algorithm, implemented as part of the Neural Circuit Tracer software package, consists of the following main steps. First, image stack is filtered to enhance labeled neurites. Second, centerline of the neurites is detected and optimized. Finally, individual branches of the optimal trace are merged into trees based on a cost minimization approach. The cost function accounts for branch orientations, distances between their end-points, curvature of the merged structure, and its intensity. The algorithm is capable of connecting branches which appear broken due to imperfect labeling and can resolve situations where branches appear to be fused due the limited resolution of light microscopy. The Neural Circuit Tracer software is designed to automatically incorporate ImageJ plug-ins and functions written in MatLab and provides roughly a 10-fold increases in speed in comparison to manual tracing.

Three-dimensional imaging of the mouse neurovasculature with magnetic resonance microscopy

Knowledge of the three-dimensional (3D) architecture of blood vessels in the brain is crucial because the progression of various neuropathologies ranging from Alzheimer's disease to brain tumors involves anomalous blood vessels. The challenges in obtaining such data from patients, in conjunction with development of mouse models of neuropathology, have made the murine brain indispensable for investigating disease induced neurovascular changes. Here we describe a novel method for “whole brain” 3D mapping of murine neurovasculature using magnetic resonance microscopy (μMRI). This approach preserves the vascular and white matter tract architecture, and can be combined with complementary MRI contrast mechanisms such as diffusion tensor imaging (DTI) to examine the interplay between the vasculature and white matter reorganization that often characterizes neuropathologies. Following validation with micro computed tomography (μCT) and optical microscopy, we demonstrate the utility of this method by: (i) combined 3D imaging of angiogenesis and white matter reorganization in both, invasive and non-invasive brain tumor models; (ii) characterizing the morphological heterogeneity of the vascular phenotype in the murine brain; and (iii) conducting “multi-scale” imaging of brain tumor angiogenesis, wherein we directly compared in vivo MRI blood volume measurements with ex vivo vasculature data.

A Hessian-based filter for vascular segmentation of noisy hepatic CT scans

PURPOSE

Extraction and enhancement of tubular structures are important in image processing applications, especially in the analysis of liver CT scans where delineation of vascular structures is needed for surgical planning. Portal vein cross-sections have circular or elliptical shapes, so an algorithm must accommodate both. A vessel segmentation method based on medial-axis points was developed and tested on portal veins in CT images.

METHODS

A medial-axis enhancement filter was developed. Consider a line passing through a point inside a tube and intersecting the edges of the tube. If the point is located on the medial axis, the distance of the point in the direction of the line to the edges of the tube will be equal. This feature was employed in a multi-scale framework to identify liver vessels. Dynamic thresholding was used to reduce noise sensitivity. The isotropic coefficient introduced by Pock et al. was used to reduce the response of the filter for asymmetric cross-sections.

RESULTS

Quantitative and qualitative evaluation of the proposed method were performed using both 2D/3D and synthetic/clinical datasets. Compared to other methods for medial-axis enhancement, our method produces better results in low-resolution CT images. Detection rate of the medial axis by the proposed method in a noisy image of standard deviation equal to 0.3 is 68% higher than prior methods.

CONCLUSION

A new Hessian-based method for medial axis vessel segmentation was developed and tested. This method produced superior results compared to prior methods. This new method has the potential for many applications of medial-axis enhancement.

Simple Neurite Tracer: open source software for reconstruction, visualization and analysis of neuronal processes

MOTIVATION

Advances in techniques to sparsely label neurons unlock the potential to reconstruct connectivity from 3D image stacks acquired by light microscopy. We present an application for semi-automated tracing of neurons to quickly annotate noisy datasets and construct complex neuronal topologies, which we call the Simple Neurite Tracer.

AVAILABILITY

Simple Neurite Tracer is open source software, licensed under the GNU General Public Licence (GPL) and based on the public domain image processing software ImageJ. The software and further documentation are available via http://fiji.sc/Simple_Neurite_Tracer as part of the package Fiji, and can be used on Windows, Mac OS and Linux. Documentation and introductory screencasts are available at the same URL.

CONTACT

longair@ini.phys.ethz.ch; longair@ini.phys.ethz.ch.

VE-LLI-VO: vessel enhancement using local line integrals and variational optimization

Vessel enhancement is a primary preprocessing step for vessel segmentation and visualization of vasculatures. In this paper, a new vessel enhancement technique is proposed in order to produce accurate vesselness measures and vessel direction estimations that are less subject to local intensity abnormalities. The proposed method is called vessel enhancement using local line integrals and variational optimization (VE-LLI-VO). First, vessel enhancement using local line integrals (VE-LLI) is introduced in which a vessel model is embedded by regarding a vessel segment as a straight line based upon the second order information of the local line integrals. Useful quantities similar to the eigenvalues and eigenvectors of the Hessian matrix are produced. Moreover, based upon the local line integrals, junctions can be detected and handled effectively. This can help deal with the bifurcation suppression problem which exists in the Hessian-based enhancement methods. Then a more generic curve model is embedded to model vessels and a variational optimization (VO) framework is introduced to generate optimized vesselness measures. Experiments have been conducted on both synthetic images and retinal images. It is experimentally demonstrated that VE-LLI-VO produces improved performance as compared with the widely used techniques in terms of both vesselness measurement and vessel direction estimation.

Vascular phenotyping of brain tumors using magnetic resonance microscopy (μMRI)

Abnormal vascular phenotypes have been implicated in neuropathologies ranging from Alzheimer's disease to brain tumors. The development of transgenic mouse models of such diseases has created a crucial need for characterizing the murine neurovasculature. Although histologic techniques are excellent for imaging the microvasculature at submicron resolutions, they offer only limited coverage. It is also challenging to reconstruct the three-dimensional (3D) vasculature and other structures, such as white matter tracts, after tissue sectioning. Here, we describe a novel method for 3D whole-brain mapping of the murine vasculature using magnetic resonance microscopy (μMRI), and its application to a preclinical brain tumor model. The 3D vascular architecture was characterized by six morphologic parameters: vessel length, vessel radius, microvessel density, length per unit volume, fractional blood volume, and tortuosity. Region-of-interest analysis showed significant differences in the vascular phenotype between the tumor and the contralateral brain, as well as between postinoculation day 12 and day 17 tumors. These results unequivocally show the feasibility of using μMRI to characterize the vascular phenotype of brain tumors. Finally, we show that combining these vascular data with coregistered images acquired with diffusion-weighted MRI provides a new tool for investigating the relationship between angiogenesis and concomitant changes in the brain tumor microenvironment.

Estimation of local orientations in fibrous structures with applications to the Purkinje system

The extraction of the cardiac Purkinje system (PS) from intensity images is a critical step toward the development of realistic structural models of the heart. Such models are important for uncovering the mechanisms of cardiac disease and improving its treatment and prevention. Unfortunately, the manual extraction of the PS is a challenging and error-prone task due to the presence of image noise and numerous fiber junctions. To deal with these challenges, we propose a framework that estimates local fiber orientations with high accuracy and reconstructs the fibers via tracking. Our key contribution is the development of a descriptor for estimating the orientation distribution function (ODF), a spherical function encoding the local geometry of the fibers at a point of interest. The fiber/branch orientations are identified as the modes of the ODFs via spherical clustering and guide the extraction of the fiber centerlines. Experiments on synthetic data evaluate the sensitivity of our approach to image noise, width of the fiber, and choice of the mode detection strategy, and show its superior performance compared to those of the existing descriptors. Experiments on the free-running PS in an MR image also demonstrate the accuracy of our method in reconstructing such sparse fibrous structures.

Automation of Hessian-Based Tubularity Measure Response Function in 3D Biomedical Images

The blood vessels and nerve trees consist of tubular objects interconnected into a complex tree- or web-like structure that has a range of structural scale 5 μm diameter capillaries to 3 cm aorta. This large-scale range presents two major problems; one is just making the measurements, and the other is the exponential increase of component numbers with decreasing scale. With the remarkable increase in the volume imaged by, and resolution of, modern day 3D imagers, it is almost impossible to make manual tracking of the complex multiscale parameters from those large image data sets. In addition, the manual tracking is quite subjective and unreliable. We propose a solution for automation of an adaptive nonsupervised system for tracking tubular objects based on multiscale framework and use of Hessian-based object shape detector incorporating National Library of Medicine Insight Segmentation and Registration Toolkit (ITK) image processing libraries.

Microscopic image analysis for quantitative characterization of muscle fiber type composition

Skeletal muscles consist of muscle fibers that are responsible for contracting and generating force. Skeletal muscle fibers are categorized into distinct subtypes based on several characteristics such as contraction time, force production and resistance to fatigue. The composition of distinct muscle fibers in terms of their number and cross-sectional areas is characterized by a histological examination. However, manual delineation of individual muscle fibers from digitized muscle histology tissue sections is extremely time-consuming. In this study, we propose an automated image analysis system for quantitative characterization of muscle fiber type composition. The proposed system operates on digitized histological muscle tissue slides and consists of the following steps: segmentation of muscle fibers, registration of successive slides with distinct stains, and classification of muscle fibers into distinct subtypes. The performance of the proposed approach was tested on a dataset consisting of 25 image pairs of successive muscle histological cross-sections with different ATPase stain. Experimental results demonstrate a promising overall segmentation and classification accuracy of 89.1% in identifying muscle fibers of distinct subtypes.

A multi-scale non-linear vessel enhancement technique

We present an enhancement method based on nonlinear diffusion filter and statistical intensity approaches for smoothing and extracting 3-D vascular system from Magnetic Resonance Angiography (MRA) data. Our method distinguishes and enhances the vessels from the other embedded tissues. The Expectation Maximization (EM) technique is employed with non-linear diffusion in order to find the optimal contrast for enhancing vessels; therefore, smoothing while dimming the embedded tissues around the vessels and brightening the vessels. The non-linear diffusion filter smooths the homogeneous regions while preserving edges. The EM technique finds the optimal statistical parameters based on the probability distribution of the classes to discriminate the tissues in the image. Our enhancement technique has been applied to 4 3-D MRA-TOF datasets consisting of around 300 images and has been compared to the regularized Perona and Malik filter. Our experimental results show that the proposed method enhances the image, keeping only the vessels while eliminating the signal from other tissues. In comparison, the conventional non-linear diffusion filter keeps unwanted tissues in addition to the vessels.

Vascular active contour for vessel tree segmentation

In this paper, a novel active contour model is proposed for vessel tree segmentation. First, we introduce a region competition-based active contour model exploiting the gaussian mixture model, which mainly segments thick vessels. Second, we define a vascular vector field to evolve the active contour along its center line into the thin and weak vessels. The vector field is derived from the eigenanalysis of the Hessian matrix of the image intensity in a multiscale framework. Finally, a dual curvature strategy, which uses a vesselness measure-dependent function selecting between a minimal principal curvature and a mean curvature criterion, is added to smoothen the surface of the vessel without changing its shape. The developed model is used to extract the liver and lung vessel tree as well as the coronary artery from high-resolution volumetric computed tomography images. Comparisons are made with several classical active contour models and manual extraction. The experiments show that our model is more accurate and robust than these classical models and is, therefore, more suited for automatic vessel tree extraction.

Automatic segmentation of coronary angiograms based on fuzzy inferring and probabilistic tracking

BACKGROUND

Segmentation of the coronary angiogram is important in computer-assisted artery motion analysis or reconstruction of 3D vascular structures from a single-plan or biplane angiographic system. Developing fully automated and accurate vessel segmentation algorithms is highly challenging, especially when extracting vascular structures with large variations in image intensities and noise, as well as with variable cross-sections or vascular lesions.

METHODS

This paper presents a novel tracking method for automatic segmentation of the coronary artery tree in X-ray angiographic images, based on probabilistic vessel tracking and fuzzy structure pattern inferring. The method is composed of two main steps: preprocessing and tracking. In preprocessing, multiscale Gabor filtering and Hessian matrix analysis were used to enhance and extract vessel features from the original angiographic image, leading to a vessel feature map as well as a vessel direction map. In tracking, a seed point was first automatically detected by analyzing the vessel feature map. Subsequently, two operators [e.g., a probabilistic tracking operator (PTO) and a vessel structure pattern detector (SPD)] worked together based on the detected seed point to extract vessel segments or branches one at a time. The local structure pattern was inferred by a multi-feature based fuzzy inferring function employed in the SPD. The identified structure pattern, such as crossing or bifurcation, was used to control the tracking process, for example, to keep tracking the current segment or start tracking a new one, depending on the detected pattern.

RESULTS

By appropriate integration of these advanced preprocessing and tracking steps, our tracking algorithm is able to extract both vessel axis lines and edge points, as well as measure the arterial diameters in various complicated cases. For example, it can walk across gaps along the longitudinal vessel direction, manage varying vessel curvatures, and adapt to varying vessel widths in situations with arterial stenoses and aneurysms.

CONCLUSIONS

Our algorithm performs well in terms of robustness, automation, adaptability, and applicability. In particular, the successful development of two novel operators, namely, PTO and SPD, ensures the performance of our algorithm in vessel tracking.

An open-source toolkit for the volumetric measurement of CT lung lesions

An open source lesion sizing toolkit has been developed with a general architecture for implementing lesion segmentation algorithms and a reference algorithm for segmenting solid and part-solid lesions from lung CT scans. The CT lung lesion segmentation algorithm detects four three-dimensional features corresponding to the lung wall, vasculature, lesion boundary edges, and low density background lung parenchyma. These features form boundaries and propagation zones that guide the evolution of a subsequent level set algorithm. User input is used to determine an initial seed point for the level set and users may also define a region of interest around the lesion. The methods are validated against 18 nodules using CT scans of an anthropomorphic thorax phantom simulating lung anatomy. The scans were acquired under differing scanner parameters to characterize algorithm behavior under varying acquisition protocols. We also validated repeatability using six clinical cases in which the patient was rescanned on the same day (zero volume change). The source code, data sets, and a running application are all provided under an unrestrictive license to encourage reproducibility and foster scientific exchange.

Steerable features for statistical 3D dendrite detection

Most state-of-the-art algorithms for filament detection in 3-D image-stacks rely on computing the Hessian matrix around individual pixels and labeling these pixels according to its eigenvalues. This approach, while very effective for clean data in which linear structures are nearly cylindrical, loses its effectiveness in the presence of noisy data and irregular structures. In this paper, we show that using steerable filters to create rotationally invariant features that include higher-order derivatives and training a classifier based on these features lets us handle such irregular structures. This can be done reliably and at acceptable computational cost and yields better results than state-of-the-art methods.

Segmentation and quantification of pulmonary artery for noninvasive CT assessment of sickle cell secondary pulmonary hypertension

PURPOSE

Pulmonary arterial hypertension (PAH) is a progressive vascular disease that results in high mortality and morbidity in sickle cell disease (SCD) patients. PAH diagnosis is invasive via right heart catheterization, but manual measurements of the main pulmonary artery (PA) diameters from computed tomography (CT) have shown promise as noninvasive surrogate marker of PAH. The authors propose a semiautomated computer-assisted diagnostic (CAD) tool to quantify the main PA size from pulmonary CT angiography (CTA).

METHODS

A follow-up retrospective study investigated the potential of CT and image analysis to quantify the presence of PAH secondary to SCD based on PA size. The authors segmented the main pulmonary arteries using a combination of fast marching level sets and geodesic active contours from smoothed pulmonary CTA images of 20 SCD patients with proven PAH by right heart catheterization and 20 matched negative controls. From the PA segmentation, a Euclidean distance map was calculated and an algorithm based on fast marching methods was used to compute subvoxel precise centerlines of the PA trunk (PT) and main left/right PA (PM). Maximum distentions of PT and PM were automatically quantified using the centerline and validated with manual measurements from two observers.

RESULTS

The pulmonary trunk and main were significantly larger (p < 0.001) in PAH/SCD patients (33.73 +/- 3.92 mm for PT and 25.17 +/- 2.90 for PM) than controls (27.03 +/- 2.94 mm for PT and 20.62 +/- 3.06 for PM). The discrepancy was qualitatively improved when vessels' diameters were normalized by body surface area (p < 0.001). The validation of the method showed high correlation (mean R=0.9 for PT and R = 0.91 for PM) and Bland-Altman agreement (0.4 +/- 3.6 mm for PT and 0.5 +/- 2.9 mm for PM) between CAD and manual measurements. Quantification errors were comparable to intraobserver and interobserver variability. CAD measurements between two different users were robust and reproducible with correlations of R = 0.99 for both PT and PM and Bland-Altman agreements of -0.13 +/- 1.33 mm for PT and -0.08 +/- 0.84 mm for PM.

CONCLUSION

Results suggest that the semiautomated quantification of pulmonary artery has sufficient accuracy and reproducibility for clinical use. CT with image processing and extraction of PA biomarkers show great potential as a surrogate indicator for diagnosis or quantification of PAH, and could be an important tool for drug discovery and noninvasive clinical surveillance.

Informatics in radiology: Electronic cleansing for noncathartic CT colonography: a structure-analysis scheme

Computed tomographic (CT) colonography performed after tagging of fecal matter but without a cathartic agent, or noncathartic CT colonography (also known as laxative-free CT colonography), is regarded as a promising next-generation technique for reducing or eliminating the discomfort associated with cathartic bowel preparation, which is the major barrier to undergoing colon cancer screening. Electronic cleansing is an emerging technique for the removal of tagged fecal materials from CT colonographic images. Three major electronic cleansing artifacts--soft-tissue degradation, pseudo-soft-tissue structures, and incomplete cleansing--severely impair the quality of electronically cleansed noncathartic CT colonographic images and limit the diagnostic utility of this modality. A structure-analysis electronic cleansing scheme was developed that makes use of local morphologic information to identify submerged colonic soft-tissue structures while removing the tagged material. Combined with other cutting-edge image processing techniques, including local roughness analysis, mosaic decomposition, and level set segmentation, structure-analysis cleansing helps eliminate the aforementioned artifacts, providing diagnostic-quality cleansed CT colonographic images for the detection of colon cancer. Noncathartic CT colonography with the application of structure-analysis cleansing is expected to help promote CT colonography as a patient-friendly method of colorectal cancer screening.