Endocardial border detection in cardiac magnetic resonance images using level set method.

Segmentation of the left ventricle in MRI images is a task with important diagnostic power. Currently, the evaluation of cardiac function involves the global measurement of volumes and ejection fraction. This evaluation requires the segmentation of the left ventricle contour. In this paper, we propose a new method for automatic detection of the endocardial border in cardiac magnetic resonance images, by using a level set segmentation-based approach. To initialize this level set segmentation algorithm, we propose to threshold the original image and to use the binary image obtained as initial mask for the level set segmentation method. For the localization of the left ventricular cavity, used to pose the initial binary mask, we propose an automatic approach to detect this spatial position by the evaluation of a metric indicating object's roundness. The segmentation process starts by the initialization of the level set algorithm and ended up through a level set segmentation. The validation process is achieved by comparing the segmentation results, obtained by the automated proposed segmentation process, to manual contours traced by tow experts. The database used was containing one automated and two manual segmentations for each sequence of images. This comparison showed good results with an overall average similarity area of 97.89%.

Deep Learning Models for Coronary Atherosclerosis Detection in Coronary CT Angiography.

BACKGROUND: Patients with atherosclerosis have a rather high risk of showing complications, if not diagnosed quickly and efficiently. OBJECTIVE: In this paper we aim to test and compare different pre-trained deep learning models, to find the best model for atherosclerosis detection in coronary CT angiography. METHODS: We experimented with different pre-trained deep learning models and fine-tuned each model to achieve the best classification accuracy. We then used the Haar wavelet decomposition to improve the model's sensitivity. RESULTS: We found that the Resnet101 architecture had the best performance with an accuracy of 95.2%, 60.8% sensitivity, and 90.48% PPV. Compared to the state of the art which uses a 3D CNN and achieved  90.9% accuracy, 68.9% Sensitivity and 58.8% PPV, sensitivity was quite low. To improve the sensitivity, we chose to use the Haar wavelet decomposition and trained the CNN model with the module of the three details: Low_High, High_Low, and High_High. The best sensitivity reached 80% with the CNN_KNN classifier . CONCLUSIONS: It is possible to perform atherosclerosis detection straight from CCTA images using a pretrained Resnet101, which has good accuracy and PPV. The low sensitivity can be improved using Haar wavelet decomposition and CNN-KNN classifier.

Spine localization in X-ray images using interest point detection.

This study was conducted to evaluate a new method used to calculate vertebra orientation in medical x-ray images. The goal of this work is to develop an x-ray image segmentation approach used to identify the location and the orientation of the cervical vertebrae in medical images. We propose a method for localization of vertebrae by extracting the anterior-left-faces of vertebra contours. This approach is based on automatic corner points of interest detection. For this task, we use the Harris corner detector. The final goal is to determine vertebral motion induced by their movement between two or several positions. The proposed system proceeds in several phases as follows: (a) image acquisition, (b) corner detection, (c) extracting of the corners belonging to vertebra left sides, (d) global estimation of the spine curvature, and (e) anterior face vertebra detection.

A new decision support model for preanesthetic evaluation.

BACKGROUND AND OBJECTIVE: The principal challenges in the field of anesthesia and intensive care consist of reducing both anesthetic risks and mortality rate. The ASA score plays an important role in patients' preanesthetic evaluation. In this paper, we propose a methodology to derive simple rules which classify patients in a category of the ASA scale on the basis of their medical characteristics. METHODS: This diagnosis system is based on MR-Sort, a multiple criteria decision analysis model. The proposed method intends to support two steps in this process. The first is the assignment of an ASA score to the patient; the second concerns the decision to accept-or not-the patient for surgery. RESULTS: In order to learn the model parameters and assess its effectiveness, we use a database containing the parameters of 898 patients who underwent preanesthesia evaluation. The accuracy of the learned models for predicting the ASA score and the decision of accepting the patient for surgery is assessed and proves to be better than that of other machine learning methods. Furthermore, simple decision rules can be explicitly derived from the learned model. These are easily interpretable by doctors, and their consistency with medical knowledge can be checked. CONCLUSIONS: The proposed model for assessing the ASA score produces accurate predictions on the basis of the (limited) set of patient attributes in the database available for the tests. Moreover, the learned MR-Sort model allows for easy interpretation by providing human-readable classification rules.

Benchmark for Algorithms Segmenting the Left Atrium From 3D CT and MRI Datasets.

Knowledge of left atrial (LA) anatomy is important for atrial fibrillation ablation guidance, fibrosis quantification and biophysical modelling. Segmentation of the LA from Magnetic Resonance Imaging (MRI) and Computed Tomography (CT) images is a complex problem. This manuscript presents a benchmark to evaluate algorithms that address LA segmentation. The datasets, ground truth and evaluation code have been made publicly available through the http://www.cardiacatlas.org website. This manuscript also reports the results of the Left Atrial Segmentation Challenge (LASC) carried out at the STACOM'13 workshop, in conjunction with MICCAI'13. Thirty CT and 30 MRI datasets were provided to participants for segmentation. Each participant segmented the LA including a short part of the LA appendage trunk and proximal sections of the pulmonary veins (PVs). We present results for nine algorithms for CT and eight algorithms for MRI. Results showed that methodologies combining statistical models with region growing approaches were the most appropriate to handle the proposed task. The ground truth and automatic segmentations were standardised to reduce the influence of inconsistently defined regions (e.g., mitral plane, PVs end points, LA appendage). This standardisation framework, which is a contribution of this work, can be used to label and further analyse anatomical regions of the LA. By performing the standardisation directly on the left atrial surface, we can process multiple input data, including meshes exported from different electroanatomical mapping systems.

Vertebra identification using template matching modelmp and K-means clustering.

PURPOSE: Accurate vertebra detection and segmentation are essential steps for automating the diagnosis of spinal disorders. This study is dedicated to vertebra alignment measurement, the first step in a computer-aided diagnosis tool for cervical spine trauma. Automated vertebral segment alignment determination is a challenging task due to low contrast imaging and noise. A software tool for segmenting vertebrae and detecting subluxations has clinical significance. A robust method was developed and tested for cervical vertebra identification and segmentation that extracts parameters used for vertebra alignment measurement. METHODS: Our contribution involves a novel combination of a template matching method and an unsupervised clustering algorithm. In this method, we build a geometric vertebra mean model. To achieve vertebra detection, manual selection of the region of interest is performed initially on the input image. Subsequent preprocessing is done to enhance image contrast and detect edges. Candidate vertebra localization is then carried out by using a modified generalized Hough transform (GHT). Next, an adapted cost function is used to compute local voted centers and filter boundary data. Thereafter, a K-means clustering algorithm is applied to obtain clusters distribution corresponding to the targeted vertebrae. These clusters are combined with the vote parameters to detect vertebra centers. Rigid segmentation is then carried out by using GHT parameters. Finally, cervical spine curves are extracted to measure vertebra alignment. RESULTS: The proposed approach was successfully applied to a set of 66 high-resolution X-ray images. Robust detection was achieved in 97.5 % of the 330 tested cervical vertebrae. CONCLUSIONS: An automated vertebral identification method was developed and demonstrated to be robust to noise and occlusion. This work presents a first step toward an automated computer-aided diagnosis system for cervical spine trauma detection.

Three-dimensional spine model reconstruction using one-class SVM regularization.

Statistical shape models have become essential for medical image registration or segmentation and are used in many biomedical applications. These models are often based on Gaussian distributions learned from a training set. We propose in this paper a shape model which does not rely on the estimation of a Gaussian distribution, but on similarities computed with a kernel function. Our model takes advantage of the one-class support vector machine (OCSVM) to do so. In this context, we propose in this paper a method for reconstructing the spine of scoliotic patients using OCSVM regularization. Current state-of-the-art methods use conventional statistical shape models, and the reconstruction is commonly processed by minimizing a Mahalanobis distance. Nevertheless, when a shape differs significantly from the statistical model, the associated Mahalanobis distance often overstates the need for statistical regularization. We show that OCSVM regularization is more robust and is less sensitive to weak landmarks definition and is hardly influenced by the presence of outliers in the training data. The proposed OCSVM model applied to 3-D spine reconstruction was evaluated on real patient data, and results showed that our approach allows precise reconstruction.

Cervical spine mobility analysis on radiographs: a fully automatic approach.

Conventional X-ray radiography remains nowadays the most common method to analyze spinal mobility in two dimensions. Therefore, the objective of this paper is to develop a framework dedicated to the fully automatic cervical spine mobility analysis on X-ray images. To this aim, we propose an approach based on three main steps: fully automatic vertebra detection, vertebra segmentation and angular measurement. The accuracy of the method was assessed for a total of 245 vertebræ. For the vertebra detection, we proposed an adapted version of two descriptors, namely Scale-invariant Feature Transform (SIFT) and Speeded-up Robust Features (SURF), coupled with a multi-class Support Vector Machine (SVM) classifier. Vertebræ are successfully detected in 89.8% of cases and it is demonstrated that SURF slightly outperforms SIFT. The Active Shape Model approach was considered as a segmentation procedure. We observed that a statistical shape model specific to the vertebral level improves the results. Angular errors of cervical spine mobility are presented. We showed that these errors remain within the inter-operator variability of the reference method.

Fast 3D spine reconstruction of postoperative patients using a multilevel statistical model.

Severe cases of spinal deformities such as scoliosis are usually treated by a surgery where instrumentation (hooks, screws and rods) is installed to the spine to correct deformities. Even if the purpose is to obtain a normal spine curve, the result is often straighter than normal. In this paper, we propose a fast statistical reconstruction algorithm based on a general model which can deal with such instrumented spines. To this end, we present the concept of multilevel statistical model where the data are decomposed into a within-group and a between-group component. The reconstruction procedure is formulated as a second-order cone program which can be solved very fast (few tenths of a second). Reconstruction errors were evaluated on real patient data and results showed that multilevel modeling allows better 3D reconstruction than classical models.

A framework of vertebra segmentation using the active shape model-based approach.

We propose a medical image segmentation approach based on the Active Shape Model theory. We apply this method for cervical vertebra detection. The main advantage of this approach is the application of a statistical model created after a training stage. Thus, the knowledge and interaction of the domain expert intervene in this approach. Our application allows the use of two different models, that is, a global one (with several vertebrae) and a local one (with a single vertebra). Two modes of segmentation are also proposed: manual and semiautomatic. For the manual mode, only two points are selected by the user on a given image. The first point needs to be close to the lower anterior corner of the last vertebra and the second near the upper anterior corner of the first vertebra. These two points are required to initialize the segmentation process. We propose to use the Harris corner detector combined with three successive filters to carry out the semiautomatic process. The results obtained on a large set of X-ray images are very promising.

Heterogeneous computing for vertebra detection and segmentation in x-ray images.

The context of this work is related to the vertebra segmentation. The method we propose is based on the active shape model (ASM). An original approach taking advantage of the edge polygonal approximation was developed to locate the vertebra positions in a X-ray image. Despite the fact that segmentation results show good efficiency, the time is a key variable that has always to be optimized in a medical context. Therefore, we present how vertebra extraction can efficiently be performed in exploiting the full computing power of parallel (GPU) and heterogeneous (multi-CPU/multi-GPU) architectures. We propose a parallel hybrid implementation of the most intensive steps enabling to boost performance. Experimentations have been conducted using a set of high-resolution X-ray medical images, showing a global speedup ranging from 3 to 22, by comparison with the CPU implementation. Data transfer times between CPU and GPU memories were included in the execution times of our proposed implementation.

Spine localization and vertebral mobility analysis using faces contours detection.

This paper describes a new segmentation approach for detecting the location and the orientation of the cervical vertebrae in medical X-ray images. A first pre-processing step consists on determining a global polygonal region for each vertebra. After this, we propose a new approach of vertebrae localization based on the four faces detection of each vertebra contour. A specific goal of the proposed application is to create an efficient semi-automated method of identifying the overall angle of curvature of the spine and the angles between vertebrae. The final goal is to determine the motion of the vertebrae induced by their movement between two or several positions.