Congenital aortic disease: 4D magnetic resonance segmentation and quantitative analysis

doi:10.1016/j.media.2009.02.005

Medical Image Analysis

Volume 13, Issue 3, June 2009, Pages 483-493

https://doi.org/10.1016/j.media.2009.02.005 Get rights and content

Abstract

Automated and accurate segmentation of the aorta in 4D (3D+time) cardiovascular magnetic resonance (MR) image data is important for early detection of congenital aortic disease leading to aortic aneurysms and dissections. A computer-aided diagnosis (CAD) method is reported that allows one to objectively identify subjects with connective tissue disorders from 16-phase 4D aortic MR images. Starting with a step of multi-view image registration, our automated segmentation method combines level-set and optimal surface segmentation algorithms in a single optimization process so that the final aortic surfaces in all 16 cardiac phases are determined. The resulting aortic lumen surface is registered with an aortic model followed by calculation of modal indices of aortic shape and motion. The modal indices reflect the differences of any individual aortic shape and motion from an average aortic behavior. A Support Vector Machine (SVM) classifier is used for the discrimination between normal and connective tissue disorder subjects.

4D MR image data sets acquired from 104 normal volunteers and connective tissue disorder patients MR datasets were used for development and performance evaluation of our method. The automated 4D segmentation resulted in accurate aortic surfaces in all 16 cardiac phases, covering the aorta from the aortic annulus to the diaphragm, yielding subvoxel accuracy with signed surface positioning errors of $- 0.07 \pm 1.16$ voxel $(- 0.10 \pm 2.05 mm)$ . The computer-aided diagnosis method distinguished between normal and connective tissue disorder subjects with a classification correctness of 90.4%.

Introduction

Aortic aneurysms and dissections are the 18th leading cause of death in the US, representing 0.6% of all deaths in 2006 (The Centers for Disease Control and Prevention, 2006). Persons with certain congenital connective tissue disorders, such as Marfan Syndrome and Familial Thoracic Aortic Aneurysm Syndrome are at increased risk of developing aortic aneurysm and/or dissection. Therefore, early diagnosis of connective tissue disorders is of a great importance.

During the last decade, cardiovascular magnetic resonance (MR) imaging contributed substantially to cardiovascular disease diagnosis. A 4D cardiovascular MR image, which consists of 3D images along the cardiac cycle, supplies important functional information about aortic motion that is missing in single-phase 3D images. While there are only limited clinical studies to support the usefulness of motion data, early observations in patients with Marfan syndrome already suggested the importance of examining motion information (Atsuchi et al., 1977). The motion information holds promise of improving the diagnosis of the aortic disease. However, 4D MR image analysis is a nontrivial task. When attempting this task manually, obtaining tracings of the aorta in an n-phase 4D image requires expert knowledge and is tedious, time-consuming, and suffers from inter- and intra-observer variability. Fig. 1 shows typical aortic MR data from a connective tissue disorder subject enrolled in this study. Even more importantly, objectively analyzing the small motion and shape differences of the individual aortas throughout the cardiac cycle is difficult if not impossible for a human operator relying solely on qualitative visual means. Therefore, an automated computer-aided diagnosis (CAD) system allowing one to distinguish between normal and abnormal (connective tissue disorder) subjects needs to be developed.

Many aortic segmentation techniques were developed in 3D using computed tomography (CT) or MR images. Rueckert et al. (1997) used Geometric Deformable Models (GDM) to track the ascending and descending aorta. Behrens et al. (2003) obtained a coarse segmentation using Randomized Hough Transform (RHT). de Bruijne et al. (2003) introduced an Adapting Active Shape Model (AASM) for tubular structure segmentation. Subasic et al. (2002) utilized a level-set algorithm for segmentation of abdominal aortic aneurysms. Considerable work has been done to establish the abnormalities in vessel mechanics that exist in patients with connective tissue disorders (Cheng, 2006, Kardesoglu et al., 2007, Vitarelli et al., 2006). Included among the parameters that have proven useful in assessing the risk of aortic disease progression in Marfan patients is the determination of aortic pulse wave velocity (Sandor et al., 2003).

In a closely related research direction, several authors have proposed techniques for tracking the cardiac movement in 4D cardiac images. Bardinet et al. (1996) presented an algorithm for tracking surfaces in 4D cardiac images based on a parametric model. Chandrashekara et al. (2003) built a statistical model derived from the motion fields in the hearts of several healthy volunteers to track the movement of the myocardium. McInerney and Terzopoulos (1995) built a dynamic finite element surface model for segmentation. Montagnat and Delingette (2005) presented a 4D cardiac segmentation by introducing time-dependent constraints to the deformable surface framework.

To our knowledge, 4D segmentation algorithms for automated tracking of the aortic movement in 4D have not been presented. In this study, we report a computer-aided diagnosis method for aortic lumen segmentation and subsequent objective identification of subjects with connective tissue disorder from 16-phase 4D aortic MR images. Starting with a step of multi-view image registration, this automated CAD system allows simultaneous segmentation of a sequence of 3D luminal surfaces in a 4D aortic MR image, followed by calculation of quantitative indices of aortic shape and motion. The descriptive indices of shape and motion are capable of correctly distinguishing the normal and diseased subjects, which may help with early diagnosis of this congenital disorder.

Section snippets

Methods

The reported CAD method consists of two main stages – aortic segmentation and connective tissue disorder assessment. Surface segmentation of the aortic lumen is obtained with an automatic 4D segmentation method. Next, a quantitative method to detect the differences in the aortic 4D function between normal and connective tissue disorder patients is employed to provide quantitative descriptors used in a disease classification step.

Image data acquisition

The algorithm was developed and evaluated on a set of 104 MR image sequences acquired from 104 subjects (52 normal, 52 patients with genetically suspected or established connective tissue disorder aortic disease — the patient status was established based on family history or 2D ultrasound-based clinical diagnosis, physical examination, and genetic testing when available). The MR imaging was performed either on a GE Signa or Siemens Avanto 1.5 T scanners, in both cases using a 2D cine true

Results

In all results referenced below, the following method’s parameters were used (Eqs. (2), (5), (7), (8), (10), (11)): $α = 0.5, β = 0.5, θ = 0.4, σ = 1.2, δ_{u} = 5, δ_{l} = 5, ω = 0.8, Δ P = 100$ . All these values were determined empirically.

Discussion and conclusion

This section will concentrate on four separate topics: (a) utility of the developed methods for diagnosing patients at risk from connective tissue disorder, (b) properties of the aortic segmentation method, (c) limitations of this technique, and (d) performance of the method to distinguish between normal and early-stage disease.

Acknowledgements

The authors would like to thank Drs. Nicholas E. Walker and Steven C. Mitchell, as well as Natalie R. Van Waning for their contribution to the project. This work was supported, in part, by the NIH grants R01HL071809 and R01EB004640.

References (37)

Y. Atsuchi et al.
Echocardiographic manifestation of annuloaortic ectasia: its paradoxical motion of the aorta and premature systolic closure of the aortic valve
American Heart Journal
(1977)
E. Bardinet et al.
Tracking and motion analysis of the left ventricle with deformable superquadrics
Medical Image Analysis
(1996)
T.O. Cheng
Decreased aortic root distensibility rather than increased aortic root diameter as an important cardiovascular risk factor in the Marfan syndrome
The American Journal of Cardiology
(2006)
T. McInerney et al.
A dynamic finite element surface model for segmentation and tracking in multidimensional medical images with application to cardiac 4D image analysis
Computerized Medical Imaging and Graphics
(1995)
J. Montagnat et al.
4D deformable models with temporal constraints: application to 4D cardiac image segmentation
Medical Image Analysis
(2005)
G.G. Sandor et al.
A novel Doppler echocardiographic method of measuring the biophysical properties of the aorta in pediatric patients
J Am Soc Echocardiogr
(2003)
A. Vitarelli et al.
Aortic wall mechanics in the Marfan syndrome assessed by transesophageal tissue Doppler echocardiography
The American Journal of Cardiology
(2006)
T. Behrens et al.
Robust segmentation of tubular structures in 3D medical images by parametric object detection and tracking
IEEE Transactions on Systems Man and Cybernetics
(2003)
P.J. Besl et al.
A method for registration of 3-D shapes
IEEE Transactions on Pattern Analysis and Machine Intelligence
(1992)
Boser, B.E., Guyon, I., Vapnik, V., 1992. A training algorithm for optimal margin classifiers. In: Proc. of ACM...

Chandrashekara, R., Rao, A., Sanchez-Ortiz, G.I., Mohiaddin, R.H., Rueckert, D., 2003. Construction of a statistical...

Chang, C.-C., Lin, C.-J., 2006. LIBSVM: A library for support vector machines. URL...

C. Cortes et al.

Support-vector networks

Machine Learning

(1995)

N. Cristianini et al.

An Introduction to Support Vector Machines and Other Kernel-based Learning Methods

(2000)

de Bruijne, M., van Ginneken, B., Viergever, M.A., Niessen, W.J., 2003. Adapting active shape models for 3D...

Frangi, A., Niessen, W., Vincken, K., Viergever, M., 1998. Multiscale vessel enhancement filtering. In: Proceedings of...

A. Frangi et al.

Automatic construction of multiple-object three-dimensional statistical shape models: Application to cardiac modelling

IEEE Transactions on Medical Imaging

(2002)

X. Gu et al.

Genus zero surface conformal mapping and its application to brain surface mapping

IEEE Transactions on Medical Imaging

(2004)

Cited by (52)

Fully convolutional multi-scale residual DenseNets for cardiac segmentation and automated cardiac diagnosis using ensemble of classifiers
2019, Medical Image Analysis
Citation Excerpt :
A summary of these approaches can be found in Table 1. For automating cardiac diagnosis from cine MR images statistical shape based methods (Sonka et al., 2003; Suinesiaputra et al., 2009; Zhao et al., 2009; Suinesiaputra et al., 2018) have been proposed over the years. In the field of medical image analysis, considerable amount of work has been done in the lines of automating segmentation of various anatomical structures, detection and delineation of lesions and tumors.
Deep fully convolutional neural network (FCN) based architectures have shown great potential in medical image segmentation. However, such architectures usually have millions of parameters and inadequate number of training samples leading to over-fitting and poor generalization. In this paper, we present a novel DenseNet based FCN architecture for cardiac segmentation which is parameter and memory efficient. We propose a novel up-sampling path which incorporates long skip and short-cut connections to overcome the feature map explosion in conventional FCN based architectures. In order to process the input images at multiple scales and view points simultaneously, we propose to incorporate Inception module’s parallel structures. We propose a novel dual loss function whose weighting scheme allows to combine advantages of cross-entropy and Dice loss leading to qualitative improvements in segmentation. We demonstrate computational efficacy of incorporating conventional computer vision techniques for region of interest detection in an end-to-end deep learning based segmentation framework. From the segmentation maps we extract clinically relevant cardiac parameters and hand-craft features which reflect the clinical diagnostic analysis and train an ensemble system for cardiac disease classification. We validate our proposed network architecture on three publicly available datasets, namely: (i) Automated Cardiac Diagnosis Challenge (ACDC-2017), (ii) Left Ventricular segmentation challenge (LV-2011), (iii) 2015 Kaggle Data Science Bowl cardiac challenge data. Our approach in ACDC-2017 challenge stood second place for segmentation and first place in automated cardiac disease diagnosis tasks with an accuracy of 100% on a limited testing set (n=50). In the LV-2011 challenge our approach attained 0.74 Jaccard index, which is so far the highest published result in fully automated algorithms. In the Kaggle challenge our approach for LV volume gave a Continuous Ranked Probability Score (CRPS) of 0.0127, which would have placed us tenth in the original challenge. Our approach combined both cardiac segmentation and disease diagnosis into a fully automated framework which is computationally efficient and hence has the potential to be incorporated in computer-aided diagnosis (CAD) tools for clinical application.
Aortic length measurements for pulse wave velocity calculation: Manual 2D vs automated 3D centreline extraction
2017, Journal of Cardiovascular Magnetic Resonance
Pulse wave velocity (PWV) is a biomarker for the intrinsic stiffness of the aortic wall, and has been shown to be predictive for cardiovascular events. It can be assessed using cardiovascular magnetic resonance (CMR) from the delay between phase-contrast flow waveforms at two or more locations in the aorta, and the distance on CMR images between those locations. This study aimed to investigate the impact of different distance measurement methods on PWV. We present and evaluate an algorithm for automated centreline tracking in 3D images, and compare PWV calculations using distances derived from 3D images to those obtained from a conventional 2D oblique-sagittal image of the aorta.
We included 35 patients from a twin cohort, and 20 post-coarctation repair patients. Phase-contrast flow was acquired in the ascending, descending and diaphragmatic aorta. A 3D centreline tracking algorithm is presented and evaluated on a subset of 30 subjects, on three CMR sequences: balanced steady-state free precession (SSFP), black-blood double inversion recovery turbo spin echo, and contrast-enhanced CMR angiography. Aortic lengths are subsequently compared between measurements from a 2D oblique-sagittal plane, and a 3D geometry.
The error in length of automated 3D centreline tracking compared with manual annotations ranged from 2.4 [1.8-4.3] mm (mean [IQR], black-blood) to 6.4 [4.7-8.9] mm (SSFP). The impact on PWV was below 0.5m/s (<5%). Differences between 2D and 3D centreline length were significant for the majority of our experiments (p < 0.05). Individual differences in PWV were larger than 0.5m/s in 15% of all cases (thoracic aorta) and 37% when studying the aortic arch only. Finally, the difference between end-diastolic and end-systolic 2D centreline lengths was statistically significant (p < 0.01), but resulted in small differences in PWV (0.08 [0.04 - 0.10]m/s).
Automatic aortic centreline tracking in three commonly used CMR sequences is possible with good accuracy. The 3D length obtained from such sequences can differ considerably from lengths obtained from a 2D oblique-sagittal plane, depending on aortic curvature, adequate planning of the oblique-sagittal plane, and patient motion between acquisitions. For accurate PWV measurements we recommend using 3D centrelines.
Automatic Abdominal Aortic Aneurysm segmentation in MR images
2016, Expert Systems with Applications
Citation Excerpt :
Other methods perform a full 3D segmentation in one single step, profiting from the usage of more information simultaneously (Ayyalasomayajula et al., 2010; Lee et al., 2010). It should also be noted that 4D methods (3D throughout the cardiac cycle) have been also proposed (Hameeteman et al., 2013; Zhao et al., 2009). In order to analyze this kind of information (2D, 3D CT or MR images) in a fast and effective way, automatic or semi-automatic computer-assisted segmentation methods become crucial for the diagnosis of AAA (Shang et al., 2015).
Abdominal Aortic Aneurism is a disease related to a weakening in the aortic wall that can cause a break in the aorta and the death. The detection of an unusual dilatation of a section of the aorta is an indicative of this disease. However, it is difficult to diagnose because it is necessary image diagnosis using computed tomography or magnetic resonance. An automatic diagnosis system would allow to analyze abdominal magnetic resonance images and to warn doctors if any anomaly is detected. We focus our research in magnetic resonance images because of the absence of ionizing radiation. Although there are proposals to identify this disease in magnetic resonance images, they need an intervention from clinicians to be precise and some of them are computationally hard. In this paper we develop a novel approach to analyze magnetic resonance abdominal images and detect the lumen and the aortic wall. The method combines different algorithms in two stages to improve the detection and the segmentation so it can be applied to similar problems with other type of images or structures. In a first stage, we use a spatial fuzzy C-means algorithm with morphological image analysis to detect and segment the lumen; and subsequently, in a second stage, we apply a graph cut algorithm to segment the aortic wall. The obtained results in the analyzed images are pretty successful obtaining an average of 79% of overlapping between the automatic segmentation provided by our method and the aortic wall identified by a medical specialist. The main impact of the proposed method is that it works in a completely automatic way with a low computational cost, which is of great significance for any expert and intelligent system.
Deformable models direct supervised guidance: A novel paradigm for automatic image segmentation
2016, Neurocomputing
Citation Excerpt :
Then, the segmentation result is classified as belonging to two or more categories. This process is performed by a classifier trained with features derived from the image itself and/or the shape and position of the final state of the DM evolution [91,51,82,76,65]. We named this category recognition as the role of MLR in these approaches fulfills this higher-level image processing task.
Deformable models are a well-established and vigorously researched approach to tackle a large variety of image segmentation problems. However, they generally use a set of weighting parameters that need to be manually tuned, a task that is both hard and time consuming. More importantly, these techniques assume that the global minimum of the energy functional corresponds to the optimum segmentation result. However, it is difficult to model this condition a priori with a robust mathematical formulation, in particular when high precision of the segmentation results is required.
This contribution aims to establish an alternative approach for traditional optimization-based deformable model adjustment. This involves a general purpose, machine learning-based image segmentation framework that translates the available information into a different type of decision process. An automatically derived model is used to directly drive the deformable model evolution. In particular, we exploit ground truth information represented in a training image set by forcing the deformable model to evolve in a way that, in its final state, will make it perfectly cover the target object. This is done with the generation of a proper dataset of vector-label pairs, which we called the Image Vector-Label Dataset, the key element responsible of the integration of the different components in the framework: the deformable model, the term set, the driver, and the localizer. As opposed to classical optimization approaches, this framework gives the opportunity to automatically generate complex, nonlinear, and data-driven relationships among different sources of information, without any human intervention.
To prove the feasibility of our novel approach, we provide an effective reference implementation of our framework, tailored to the medical imaging field. We test it against a large set of state-of-the-art segmentation algorithms over two well-known image datasets with different image modalities and target structures. Although the proposed framework is not intended for a specific segmentation problem, its implementation is competitive or even outperforms most of the state-of-the-art algorithms specifically designed for the segmentation tasks at hand.
Robust automated bolus tracker positioning for MRI liver scans
2015, Magnetic Resonance Imaging
Citation Excerpt :
Many aortic segmentation techniques employing computed tomography (CT) or MRI have been proposed. For example, geometric deformable models [4], adapting active shape model [5], Hough transform [6], and level set for analyzing four-dimensional (4D) blood flow [7] are used for aortic segmentation. On the other hand, no techniques have yet been developed for automatically adjusting the bolus tracker other than our recently proposed approach [8,9].
To improve the workflow of MRI abdominal scans by reducing the examination time and operator skill dependence related to bolus tracker positioning.
Ten or more axial images of two-dimensional scout scan were analyzed to identify the aorta and detect its center position using the mean shift to allow automated bolus tracker placement. Adaptive boosting (AdaBoost) classifier was used to identify the aorta rotating a sub-window around the cerebrospinal fluid (CSF), the location of which was detected in each axial image in advance. The search region of the aorta in the next inferior axial image was restricted to half to reduce computation time. Tests were conducted using the proposed method with a 1.5 T scanner in 31 volunteers.
The success rate of aorta detection was 98.4%, and the accuracy of center location was around 0 – 5 mm shift from the true center. The computation time was 30 s on MATLAB, which was half that required for non-restrictive search.
The proposed algorithm was able to accurately detect the aorta in all volunteers with practical computation time so that the automated bolus tracker placement improved the workflow of MRI abdominal scans.
Robust segmentation methods with an application to aortic pulse wave velocity calculation
2014, Computerized Medical Imaging and Graphics
Aortic stiffness has proven to be an important diagnostic and prognostic factor of many cardiovascular diseases, as well as an estimate of overall cardiovascular health. Pulse wave velocity (PWV) represents a good measure of the aortic stiffness, while the aortic distensibility is used as an aortic elasticity index. Obtaining the PWV and the aortic distensibility from magnetic resonance imaging (MRI) data requires diverse segmentation tasks, namely the extraction of the aortic center line and the segmentation of aortic regions, combined with signal processing methods for the analysis of the pulse wave. In our study non-contrasted MRI images of abdomen were used in healthy volunteers (22 data sets) for the sake of non-invasive analysis and contrasted magnetic resonance (MR) images were used for the aortic examination of Marfan syndrome patients (8 data sets). In this research we present a novel robust segmentation technique for the PWV and aortic distensibility calculation as a complete image processing toolbox. We introduce a novel graph-based method for the centerline extraction of a thoraco-abdominal aorta for the length calculation from 3-D MRI data, robust to artifacts and noise. Moreover, we design a new projection-based segmentation method for transverse aortic region delineation in cardiac magnetic resonance (CMR) images which is robust to high presence of artifacts. Finally, we propose a novel method for analysis of velocity curves in order to obtain pulse wave propagation times. In order to validate the proposed method we compare the obtained results with manually determined aortic centerlines and a region segmentation by an expert, while the results of the PWV measurement were compared to a validated software (LUMC, Leiden, the Netherlands). The obtained results show high correctness and effectiveness of our method for the aortic PWV and distensibility calculation.

View all citing articles on Scopus

View full text

Congenital aortic disease: 4D magnetic resonance segmentation and quantitative analysis

Abstract

Introduction

Section snippets

Methods

Image data acquisition

Results

Discussion and conclusion

Acknowledgements

American Heart Journal

Medical Image Analysis

The American Journal of Cardiology

Computerized Medical Imaging and Graphics

Medical Image Analysis

J Am Soc Echocardiogr

The American Journal of Cardiology

Robust segmentation of tubular structures in 3D medical images by parametric object detection and tracking

IEEE Transactions on Systems Man and Cybernetics

A method for registration of 3-D shapes

IEEE Transactions on Pattern Analysis and Machine Intelligence

Support-vector networks

Machine Learning

An Introduction to Support Vector Machines and Other Kernel-based Learning Methods

Automatic construction of multiple-object three-dimensional statistical shape models: Application to cardiac modelling

IEEE Transactions on Medical Imaging

Genus zero surface conformal mapping and its application to brain surface mapping

IEEE Transactions on Medical Imaging