Lauge Sørensen

Standardized evaluation of algorithms for computer-aided diagnosis of dementia based on structural MRI: The CADDementia challenge

Algorithms for computer-aided diagnosis of dementia based on structural MRI have demonstrated high performance in the literature, but are difficult to compare as different data sets and methodology were used for evaluation. In addition, it is unclear how the algorithms would perform on previously unseen data, and thus, how they would perform in clinical practice when there is no real opportunity to adapt the algorithm to the data at hand. To address these comparability, generalizability and clinical applicability issues, we organized a grand challenge that aimed to objectively compare algorithms based on a clinically representative multi-center data set. Using clinical practice as the starting point, the goal was to reproduce the clinical diagnosis. Therefore, we evaluated algorithms for multi-class classification of three diagnostic groups: patients with probable Alzheimers disease, patients with mild cognitive impairment and healthy controls. The diagnosis based on clinical criteria was used as reference standard, as it was the best available reference despite its known limitations. For evaluation, a previously unseen test set was used consisting of 354 T1-weighted MRI scans with the diagnoses blinded. Fifteen research teams participated with a total of 29 algorithms. The algorithms were trained on a small training set (n=30) and optionally on data from other sources (e.g., the Alzheimers Disease Neuroimaging Initiative, the Australian Imaging Biomarkers and Lifestyle flagship study of aging). The best performing algorithm yielded an accuracy of 63.0% and an area under the receiver-operating-characteristic curve (AUC) of 78.8%. In general, the best performances were achieved using feature extraction based on voxel-based morphometry or a combination of features that included volume, cortical thickness, shape and intensity. The challenge is open for new submissions via the web-based framework: http://caddementia.grand-challenge.org.

A texton-based approach for the classification of lung parenchyma in CT images

In this paper, a texton-based classification system based on raw pixel representation along with a support vector machine with radial basis function kernel is proposed for the classification of emphysema in computed tomography images of the lung. The proposed approach is tested on 168 annotated regions of interest consisting of normal tissue, centrilobular emphysema, and paraseptal emphysema. The results show the superiority of the proposed approach to common techniques in the literature including moments of the histogram of filter responses based on Gaussian derivatives. The performance of the proposed system, with an accuracy of 96.43%, also slightly improves over a recently proposed approach based on local binary patterns.

Texture-Based Analysis of COPD: A Data-Driven Approach

This study presents a fully automatic, data-driven approach for texture-based quantitative analysis of chronic obstructive pulmonary disease (COPD) in pulmonary computed tomography (CT) images. The approach uses supervised learning where the class labels are, in contrast to previous work, based on measured lung function instead of on manually annotated regions of interest (ROIs). A quantitative measure of COPD is obtained by fusing COPD probabilities computed in ROIs within the lung fields where the individual ROI probabilities are computed using a k nearest neighbor (kNN ) classifier. The distance between two ROIs in the kNN classifier is computed as the textural dissimilarity between the ROIs, where the ROI texture is described by histograms of filter responses from a multi-scale, rotation invariant Gaussian filter bank. The method was trained on 400 images from a lung cancer screening trial and subsequently applied to classify 200 independent images from the same screening trial. The texture-based measure was significantly better at discriminating between subjects with and without COPD than were the two most common quantitative measures of COPD in the literature, which are based on density. The proposed measure achieved an area under the receiver operating characteristic curve (AUC) of 0.713 whereas the best performing density measure achieved an AUC of 0.598. Further, the proposed measure is as reproducible as the density measures, and there were indications that it correlates better with lung function and is less influenced by inspiration level.

Texture Classification in Lung CT Using Local Binary Patterns

In this paper we propose to use local binary patterns (LBP) as features in a classification framework for classifying different texture patterns in lung computed tomography. Image intensity is included by means of the joint LBP and intensity histogram, and classification is performed using the k nearest neighbor classifier with histogram similarity as distance measure. The proposed method is evaluated on a set of 168 regions of interest comprising normal tissue and different emphysema patterns, and compared to a filter bank based on Gaussian derivatives. The joint LBP and intensity histogram, achieving a classification accuracy of 95.2%, shows superior performance to using the common approach of taking moments of the filter response histograms as features, and slightly better performance than using the full filter response histograms instead. Classification results are better than some of those previously reported in the literature.

Early detection of Alzheimer's disease using MRI hippocampal texture

Cognitive impairment in patients with Alzheimers disease (AD) is associated with reduction in hippocampal volume in magnetic resonance imaging (MRI). However, it is unknown whether hippocampal texture changes in persons with mild cognitive impairment (MCI) that does not have a change in hippocampal volume. We tested the hypothesis that hippocampal texture has association to early cognitive loss beyond that of volumetric changes. The texture marker was trained and evaluated using T1‐weighted MRI scans from the Alzheimers Disease Neuroimaging Initiative (ADNI) database, and subsequently applied to score independent data sets from the Australian Imaging, Biomarker & Lifestyle Flagship Study of Ageing (AIBL) and the Metropolit 1953 Danish Male Birth Cohort (Metropolit). Hippocampal texture was superior to volume reduction as predictor of MCI‐to‐AD conversion in ADNI (area under the receiver operating characteristic curve [AUC] 0.74 vs 0.67; DeLong test, p = 0.005), and provided even better prognostic results in AIBL (AUC 0.83). Hippocampal texture, but not volume, correlated with Addenbrookes cognitive examination score (Pearson correlation, r = −0.25, p < 0.001) in the Metropolit cohort. The hippocampal texture marker correlated with hippocampal glucose metabolism as indicated by fluorodeoxyglucose‐positron emission tomography (Pearson correlation, r = −0.57, p < 0.001). Texture statistics remained significant after adjustment for volume in all cases, and the combination of texture and volume did not improve diagnostic or prognostic AUCs significantly. Our study highlights the presence of hippocampal texture abnormalities in MCI, and the possibility that texture may serve as a prognostic neuroimaging biomarker of early cognitive impairment. Hum Brain Mapp 37:1148–1161, 2016.

Differential diagnosis of mild cognitive impairment and Alzheimer's disease using structural MRI cortical thickness, hippocampal shape, hippocampal texture, and volumetry

This paper presents a brain T1-weighted structural magnetic resonance imaging (MRI) biomarker that combines several individual MRI biomarkers (cortical thickness measurements, volumetric measurements, hippocampal shape, and hippocampal texture). The method was developed, trained, and evaluated using two publicly available reference datasets: a standardized dataset from the Alzheimers Disease Neuroimaging Initiative (ADNI) and the imaging arm of the Australian Imaging Biomarkers and Lifestyle flagship study of ageing (AIBL). In addition, the method was evaluated by participation in the Computer-Aided Diagnosis of Dementia (CADDementia) challenge. Cross-validation using ADNI and AIBL data resulted in a multi-class classification accuracy of 62.7% for the discrimination of healthy normal controls (NC), subjects with mild cognitive impairment (MCI), and patients with Alzheimers disease (AD). This performance generalized to the CADDementia challenge where the method, trained using the ADNI and AIBL data, achieved a classification accuracy 63.0%. The obtained classification accuracy resulted in a first place in the challenge, and the method was significantly better (McNemars test) than the bottom 24 methods out of the total of 29 methods contributed by 15 different teams in the challenge. The method was further investigated with learning curve and feature selection experiments using ADNI and AIBL data. The learning curve experiments suggested that neither more training data nor a more complex classifier would have improved the obtained results. The feature selection experiment showed that both common and uncommon individual MRI biomarkers contributed to the performance; hippocampal volume, ventricular volume, hippocampal texture, and parietal lobe thickness were the most important. This study highlights the need for both subtle, localized measurements and global measurements in order to discriminate NC, MCI, and AD simultaneously based on a single structural MRI scan. It is likely that additional non-structural MRI features are needed to further improve the obtained performance, especially to improve the discrimination between NC and MCI.

Brain region’s relative proximity as marker for Alzheimer’s disease based on structural MRI

BackgroundAlzheimer’s disease (AD) is a progressive, incurable neurodegenerative disease and the most common type of dementia. It cannot be prevented, cured or drastically slowed, even though AD research has increased in the past 5-10 years. Instead of focusing on the brain volume or on the single brain structures like hippocampus, this paper investigates the relationship and proximity between regions in the brain and uses this information as a novel way of classifying normal control (NC), mild cognitive impaired (MCI), and AD subjects.MethodsA longitudinal cohort of 528 subjects (170 NC, 240 MCI, and 114 AD) from ADNI at baseline and month 12 was studied. We investigated a marker based on Procrustes aligned center of masses and the percentile surface connectivity between regions. These markers were classified using a linear discriminant analysis in a cross validation setting and compared to whole brain and hippocampus volume.ResultsWe found that both our markers was able to significantly classify the subjects. The surface connectivity marker showed the best results with an area under the curve (AUC) at 0.877 (p<0.001), 0.784 (p<0.001), 0,766 (p<0.001) for NC-AD, NC-MCI, and MCI-AD, respectively, for the functional regions in the brain. The surface connectivity marker was able to classify MCI-converters with an AUC of 0.599 (p<0.05) for the 1-year period.ConclusionOur results show that our relative proximity markers include more information than whole brain and hippocampus volume. Our results demonstrate that our proximity markers have the potential to assist in early diagnosis of AD.

Dissimilarity-based multiple instance learning

In this paper, we propose to solve multiple instance learning problems using a dissimilarity representation of the objects. Once the dissimilarity space has been constructed, the problem is turned into a standard supervised learning problem that can be solved with a general purpose supervised classifier. This approach is less restrictive than kernel-based approaches and therefore allows for the usage of a wider range of proximity measures. Two conceptually different types of dissimilarity measures are considered: one based on point set distance measures and one based on the earth movers distance between distributions of within- and between set point distances, thereby taking relations within and between sets into account. Experiments on five publicly available data sets show competitive performance in terms of classification accuracy compared to previously published results.

Learning COPD Sensitive Filters in Pulmonary CT

The standard approaches to analyzing emphysema in computed tomography (CT) images are visual inspection and the relative area of voxels below a threshold (RA). The former approach is subjective and impractical in a large data set and the latter relies on a single threshold and independent voxel information, ignoring any spatial correlation in intensities. In recent years, supervised learning on texture features has been investigated as an alternative to these approaches, showing good results. However, supervised learning requires labeled samples, and these samples are often obtained via subjective and time consuming visual scoring done by human experts. In this work, we investigate the possibility of applying supervised learning using texture measures on random CT samples where the labels are based on external, non-CT measures. We are not targeting emphysema directly, instead we focus on learning textural differences that discriminate subjects with chronic obstructive pulmonary disease (COPD) from healthy smokers, and it is expected that emphysema plays a major part in this. The proposed texture based approach achieves an 69% classification accuracy which is significantly better than RAs 55% accuracy.

Multi-object tracking of human spermatozoa

We propose a system for tracking of human spermatozoa in phase-contrast microscopy image sequences. One of the main aims of a computer-aided sperm analysis (CASA) system is to automatically assess sperm quality based on spermatozoa motility variables. In our case, the problem of assessing sperm quality is cast as a multi-object tracking problem, where the objects being tracked are the spermatozoa. The system combines a particle filter and Kalman filters for robust motion estimation of the spermatozoa tracks. Further, the combinatorial aspect of assigning observations to labels in the particle filter is formulated as a linear assignment problem solved using the Hungarian algorithm on a rectangular cost matrix, making the algorithm capable of handling missing or spurious observations. The costs are calculated using hidden Markov models that express the plausibility of an observation being the next position in the track history of the particle labels. Observations are extracted using a scale-space blob detector utilizing the fact that the spermatozoa appear as bright blobs in a phase-contrast microscope. The output of the system is the complete motion track of each of the spermatozoa. Based on these tracks, different CASA motility variables can be computed, for example curvilinear velocity or straight-line velocity. The performance of the system is tested on three different phase-contrast image sequences of varying complexity, both by visual inspection of the estimated spermatozoa tracks and by measuring the mean squared error (MSE) between the estimated spermatozoa tracks and manually annotated tracks, showing good agreement.