PortiK: A computer vision based solution for real-time automatic solid waste characterization - Application to an aluminium stream.

In Material Recovery Facilities (MRFs), recyclable municipal solid waste is turned into a precious commodity. However, effective recycling relies on effective waste sorting, which is still a challenge to sustainable development of our society. To help the operations improve and optimise their process, this paper describes PortiK, a solution for automatic waste analysis. Based on image analysis and object recognition, it allows for continuous, real-time, non-intrusive measurements of mass composition of waste streams. The end-to-end solution is detailed with all the steps necessary for the system to operate, from hardware specifications and data collection to supervisory information obtained by deep learning and statistical analysis. The overall system was tested and validated in an operational environment in a material recovery facility. PortiK monitored an aluminium can stream to estimate its purity. Aluminium cans were detected with 91.2% precision and 90.3% recall, respectively, resulting in an underestimation of the number of cans by less than 1%. Regarding contaminants (i.e. other types of waste), precision and recall were 80.2% and 78.4%, respectively, giving an 2.2% underestimation. Based on five sample analyses where pieces of waste were counted and weighed per batch, the detection results were used to estimate purity and its confidence level. The estimation error was calculated to be within ±7% after 5 minutes of monitoring and ±5% after 8 hours. These results have demonstrated the feasibility and the relevance of the proposed solution for online quality control of aluminium can stream.

Automatic classification of patients with Alzheimer's disease from structural MRI: a comparison of ten methods using the ADNI database.

Recently, several high dimensional classification methods have been proposed to automatically discriminate between patients with Alzheimer's disease (AD) or mild cognitive impairment (MCI) and elderly controls (CN) based on T1-weighted MRI. However, these methods were assessed on different populations, making it difficult to compare their performance. In this paper, we evaluated the performance of ten approaches (five voxel-based methods, three methods based on cortical thickness and two methods based on the hippocampus) using 509 subjects from the ADNI database. Three classification experiments were performed: CN vs AD, CN vs MCIc (MCI who had converted to AD within 18 months, MCI converters - MCIc) and MCIc vs MCInc (MCI who had not converted to AD within 18 months, MCI non-converters - MCInc). Data from 81 CN, 67 MCInc, 39 MCIc and 69 AD were used for training and hyperparameters optimization. The remaining independent samples of 81 CN, 67 MCInc, 37 MCIc and 68 AD were used to obtain an unbiased estimate of the performance of the methods. For AD vs CN, whole-brain methods (voxel-based or cortical thickness-based) achieved high accuracies (up to 81% sensitivity and 95% specificity). For the detection of prodromal AD (CN vs MCIc), the sensitivity was substantially lower. For the prediction of conversion, no classifier obtained significantly better results than chance. We also compared the results obtained using the DARTEL registration to that using SPM5 unified segmentation. DARTEL significantly improved six out of 20 classification experiments and led to lower results in only two cases. Overall, the use of feature selection did not improve the performance but substantially increased the computation times.

Multidimensional classification of hippocampal shape features discriminates Alzheimer's disease and mild cognitive impairment from normal aging.

We describe a new method to automatically discriminate between patients with Alzheimer's disease (AD) or mild cognitive impairment (MCI) and elderly controls, based on multidimensional classification of hippocampal shape features. This approach uses spherical harmonics (SPHARM) coefficients to model the shape of the hippocampi, which are segmented from magnetic resonance images (MRI) using a fully automatic method that we previously developed. SPHARM coefficients are used as features in a classification procedure based on support vector machines (SVM). The most relevant features for classification are selected using a bagging strategy. We evaluate the accuracy of our method in a group of 23 patients with AD (10 males, 13 females, age+/-standard-deviation (SD)=73+/-6 years, mini-mental score (MMS)=24.4+/-2.8), 23 patients with amnestic MCI (10 males, 13 females, age+/-SD=74+/-8 years, MMS=27.3+/-1.4) and 25 elderly healthy controls (13 males, 12 females, age+/-SD=64+/-8 years), using leave-one-out cross-validation. For AD vs controls, we obtain a correct classification rate of 94%, a sensitivity of 96%, and a specificity of 92%. For MCI vs controls, we obtain a classification rate of 83%, a sensitivity of 83%, and a specificity of 84%. This accuracy is superior to that of hippocampal volumetry and is comparable to recently published SVM-based whole-brain classification methods, which relied on a different strategy. This new method may become a useful tool to assist in the diagnosis of Alzheimer's disease.

Fully automatic hippocampus segmentation and classification in Alzheimer's disease and mild cognitive impairment applied on data from ADNI.

The hippocampus is among the first structures affected in Alzheimer's disease (AD). Hippocampal magnetic resonance imaging volumetry is a potential biomarker for AD but is hindered by the limitations of manual segmentation. We proposed a fully automatic method using probabilistic and anatomical priors for hippocampus segmentation. Probabilistic information is derived from 16 young controls and anatomical knowledge is modeled with automatically detected landmarks. The results were previously evaluated by comparison with manual segmentation on data from the 16 young healthy controls, with a leave-one-out strategy, and eight patients with AD. High accuracy was found for both groups (volume error 6 and 7%, overlap 87 and 86%, respectively). In this article, the method was used to segment 145 patients with AD, 294 patients with mild cognitive impairment (MCI), and 166 elderly normal subjects from the Alzheimer's Disease Neuroimaging Initiative database. On the basis of a qualitative rating protocol, the segmentation proved acceptable in 94% of the cases. We used the obtained hippocampal volumes to automatically discriminate between AD patients, MCI patients, and elderly controls. The classification proved accurate: 76% of the patients with AD and 71% of the MCI converting to AD before 18 months were correctly classified with respect to the elderly controls, using only hippocampal volume.

Multi-organ localization with cascaded global-to-local regression and shape prior.

We propose a method for fast, accurate and robust localization of several organs in medical images. We generalize the global-to-local cascade of regression random forest to multiple organs. A first regressor encodes the global relationships between organs, learning simultaneously all organs parameters. Then subsequent regressors refine the localization of each organ locally and independently for improved accuracy. By combining the regression vote distribution and the organ shape prior (through probabilistic atlas representation) we compute confidence maps that are organ-dedicated probability maps. They are used within the cascade itself, to better select the test voxels for the second set of regressors, and to provide richer information than the classical bounding boxes result thanks to the shape prior. We propose an extensive study of the different learning and testing parameters, showing both their robustness to reasonable perturbations and their influence on the final algorithm accuracy. Finally we demonstrate the robustness and accuracy of our approach by evaluating the localization of six abdominal organs (liver, two kidneys, spleen, gallbladder and stomach) on a large and diverse database of 130 CT volumes. Moreover, the comparison of our results with two existing methods shows significant improvements brought by our approach and our deep understanding and optimization of the parameters.

Multi-organ localization combining global-to-local regression and confidence maps.

We propose a method for fast, accurate and robust localization of several organs in medical images. We generalize global-to-local cascades of regression forests [1] to multiple organs. A first regressor encodes global relationships between organs. Subsequent regressors refine the localization of each organ locally and independently for improved accuracy. We introduce confidence maps, which incorporate information about both the regression vote distribution and the organ shape through probabilistic atlases. They are used within the cascade itself, to better select the test voxels for the second set of regressors, and to provide richer information than the classical bounding boxes thanks to the shape prior. We demonstrate the robustness and accuracy of our approach through a quantitative evaluation on a large database of 130 CT volumes.

Tagged template deformation.

Model-based approaches are very popular for medical image segmentation as they carry useful prior information on the target structure. Among them, the implicit template deformation framework recently bridged the gap between the efficiency and flexibility of level-set region competition and the robustness of atlas deformation approaches. This paper generalizes this method by introducing the notion of tagged templates. A tagged template is an implicit model in which different subregions are defined. In each of these subregions, specific image features can be used with various confidence levels. The tags can be either set manually or automatically learnt via a process also hereby described. This generalization therefore greatly widens the scope of potential clinical application of implicit template deformation while maintaining its appealing algorithmic efficiency. We show the great potential of our approach in myocardium segmentation of ultrasound images.

Spatial and Anatomical Regularization of SVM: A General Framework for Neuroimaging Data.

This paper presents a framework to introduce spatial and anatomical priors in SVM for brain image analysis based on regularization operators. A notion of proximity based on prior anatomical knowledge between the image points is defined by a graph (e.g., brain connectivity graph) or a metric (e.g., Fisher metric on statistical manifolds). A regularization operator is then defined from the graph Laplacian, in the discrete case, or from the Laplace-Beltrami operator, in the continuous case. The regularization operator is then introduced into the SVM, which exponentially penalizes high-frequency components with respect to the graph or to the metric and thus constrains the classification function to be smooth with respect to the prior. It yields a new SVM optimization problem whose kernel is a heat kernel on graphs or on manifolds. We then present different types of priors and provide efficient computations of the Gram matrix. The proposed framework is finally applied to the classification of brain Magnetic Resonance (MR) images (based on Gray Matter (GM) concentration maps and cortical thickness measures) from 137 patients with Alzheimer's Disease (AD) and 162 elderly controls. The results demonstrate that the proposed classifier generates less-noisy and consequently more interpretable feature maps with high classification performances.

Joint co-segmentation and registration of 3D ultrasound images.

Contrast-enhanced ultrasound (CEUS) allows a visualization of the vascularization and complements the anatomical information provided by conventional ultrasound (US). However, these images are inherently subject to noise and shadows, which hinders standard segmentation algorithms. In this paper, we propose to use simultaneously the different information coming from 3D US and CEUS images to address the problem of kidney segmentation. To that end, we introduce a generic framework for joint co-segmentation and registration that seeks objects having the same shape in several images. From this framework, we derive both an ellipsoid co-detection and a model-based co-segmentation algorithm. These methods rely on voxel-classification maps that we estimate using random forests in a structured way. This yields a fast and fully automated pipeline, in which an ellipsoid is first estimated to locate the kidney in both US and CEUS volumes and then deformed to segment it accurately. The proposed method outperforms state-of-the-art results (by dividing the kidney volume error by two) on a clinically representative database of 64 images.

Registration of free-breathing 3D+t abdominal perfusion CT images via co-segmentation.

Dynamic contrast-enhanced computed tomography (DCE-CT) is a valuable imaging modality to assess tissues properties, particularly in tumours, by estimating pharmacokinetic parameters from the evolution of pixels intensities in 3D+t acquisitions. However, this requires a registration of the whole sequence of volumes, which is challenging especially when the patient breathes freely. In this paper, we propose a generic, fast and automatic method to address this problem. As standard iconic registration methods are not robust to contrast intake, we rather rely on the segmentation of the organ of interest. This segmentation is performed jointly with the registration of the sequence within a novel co-segmentation framework. Our approach is based on implicit template deformation, that we extend to a co-segmentation algorithm which provides as outputs both a segmentation of the organ of interest in every image and stabilising transformations for the whole sequence. The proposed method is validated on 15 datasets acquired from patients with renal lesions and shows improvement in terms of registration and estimation of pharmacokinetic parameters over the state-of-the-art method.

Automatic detection and segmentation of kidneys in 3D CT images using random forests.

Kidney segmentation in 3D CT images allows extracting useful information for nephrologists. For practical use in clinical routine, such an algorithm should be fast, automatic and robust to contrast-agent enhancement and fields of view. By combining and refining state-of-the-art techniques (random forests and template deformation), we demonstrate the possibility of building an algorithm that meets these requirements. Kidneys are localized with random forests following a coarse-to-fine strategy. Their initial positions detected with global contextual information are refined with a cascade of local regression forests. A classification forest is then used to obtain a probabilistic segmentation of both kidneys. The final segmentation is performed with an implicit template deformation algorithm driven by these kidney probability maps. Our method has been validated on a highly heterogeneous database of 233 CT scans from 89 patients. 80% of the kidneys were accurately detected and segmented (Dice coefficient > 0.90) in a few seconds per volume.

Sulcal span in Azheimer's disease, amnestic mild cognitive impairment, and healthy controls.

Differences of cortical morphology between healthy controls (HC), amnestic mild cognitive impairment (MCI), and Alzheimer's disease (AD) have been repeatedly investigated using voxel-based morphometry (VBM). However, the results obtained using mainly VBM remain difficult to interpret as they can be explained by various mechanisms. The aim of the present study was to evaluate the differences of cortical morphology between HC, MCI, and AD patients using a new post-processing method based on reconstruction and identification of cortical sulci. Thirty HC, 33 MCI, and 30 AD patients were randomly selected from the ADNI database. For each subject, cortical sulci were reconstructed and automatically identified using Brainvisa software. Depth and fold opening of nine large sulci were compared between HC, MCI, and AD patients. Fold opening of parietaloccipital fissure and intraparietal sulcus on both sides strongly differed between the 3 groups, with gradual increase from HC to MCI of about 1 mm and from MCI to AD of about 2 mm (right intraparietal: p = 0.005; left intraparietal: p = 0.004; right parietaloccipital: p = 0.003; left parietaloccipital: p = 0.0009). Results were left unchanged after adjustment for age, gender, and level of education. These variables were also strongly linked to neuropsychological scores, independent of age, gender, and level of education. In the present study, we found important regional differences of cortical morphology with gradual deterioration from HC to MCI to AD. The most important differences were found in parietaloccipital fissure and intraparietal sulcus. Further studies are needed to understand the involved underlying mechanisms.

Contrast-based fully automatic segmentation of white matter hyperintensities: method and validation.

White matter hyperintensities (WMH) on T2 or FLAIR sequences have been commonly observed on MR images of elderly people. They have been associated with various disorders and have been shown to be a strong risk factor for stroke and dementia. WMH studies usually required visual evaluation of WMH load or time-consuming manual delineation. This paper introduced WHASA (White matter Hyperintensities Automated Segmentation Algorithm), a new method for automatically segmenting WMH from FLAIR and T1 images in multicentre studies. Contrary to previous approaches that were based on intensities, this method relied on contrast: non linear diffusion filtering alternated with watershed segmentation to obtain piecewise constant images with increased contrast between WMH and surroundings tissues. WMH were then selected based on subject dependant automatically computed threshold and anatomical information. WHASA was evaluated on 67 patients from two studies, acquired on six different MRI scanners and displaying a wide range of lesion load. Accuracy of the segmentation was assessed through volume and spatial agreement measures with respect to manual segmentation; an intraclass correlation coefficient (ICC) of 0.96 and a mean similarity index (SI) of 0.72 were obtained. WHASA was compared to four other approaches: Freesurfer and a thresholding approach as unsupervised methods; k-nearest neighbours (kNN) and support vector machines (SVM) as supervised ones. For these latter, influence of the training set was also investigated. WHASA clearly outperformed both unsupervised methods, while performing at least as good as supervised approaches (ICC range: 0.87-0.91 for kNN; 0.89-0.94 for SVM. Mean SI: 0.63-0.71 for kNN, 0.67-0.72 for SVM), and did not need any training set.

Spatial regularization of SVM for the detection of diffusion alterations associated with stroke outcome.

In this paper, we propose a new method to detect differences at the group level in brain images based on spatially regularized support vector machines (SVM). We propose to spatially regularize the SVM using a graph Laplacian. This provides a flexible approach to model different types of proximity between voxels. We propose a proximity graph which accounts for tissue types. An efficient computation of the Gram matrix is provided. Then, significant differences between two populations are detected using statistical tests on the outputs of the SVM. The method was first tested on synthetic examples. It was then applied to 72 stroke patients to detect brain areas associated with motor outcome at 90 days, based on diffusion-weighted images acquired at the acute stage (median delay one day). The proposed method showed that poor motor outcome is associated to changes in the corticospinal bundle and white matter tracts originating from the premotor cortex. Standard mass univariate analyses failed to detect any difference on the same population.

Spatially regularized SVM for the detection of brain areas associated with stroke outcome.

This paper introduces a new method to detect group differences in brain images based on spatially regularized support vector machines (SVM). First, we propose to spatially regularize the SVM using a graph encoding the voxels' proximity. Two examples of regularization graphs are provided. Significant differences between two populations are detected using statistical tests on the margins of the SVM. We first tested our method on synthetic examples. We then applied it to 72 stroke patients to detect brain areas associated with motor outcome at 90 days, based on diffusion-weighted images acquired at the acute stage (one day delay). The proposed method showed that poor motor outcome is associated to changes in the corticospinal bundle and white matter tracts originating from the premotor cortex. Standard mass univariate analyses failed to detect any difference.