Multi-source composite kernels for urban image classification

This letter presents advanced classification methods for very high resolution images. Efficient multisource information, both spectral and spatial, is exploited through the use of composite kernels in support vector machines. Weighted summations of kernels accounting for separate sources of spectral and spatial information are analyzed and compared to classical approaches such as pure spectral classification or stacked approaches using all the features in a single vector. Model selection problems are addressed, as well as the importance of the different kernels in the weighted summation.

Cluster-based active learning for compact image classification

In this paper, we consider active sampling to label pixels grouped with hierarchical clustering. The objective of the method is to match the data relationships discovered by the clustering algorithm with the user's desired class semantics. The first is represented as a complete tree to be pruned and the second is iteratively provided by the user. The active learning algorithm proposed searches the pruning of the tree that best matches the labels of the sampled points. By choosing the part of the tree to sample from according to current pruning's uncertainty, sampling is focused on most uncertain clusters. This way, large clusters for which the class membership is already fixed are no longer queried and sampling is focused on division of clusters showing mixed labels. The model is tested on a VHR image in a multiclass classification setting. The method clearly outperforms random sampling in a transductive setting, but cannot generalize to unseen data, since it aims at optimizing the classification of a given cluster structure.

Automatic Mallampati Classification Using Active Appearance Models

Difficult tracheal intubation assessment is an important research topic in anesthesia as failed intubations are important causes of mortality in anesthetic practice. The modified Mallampati score is widely used, alone or in conjunction with other criteria, to predict the difficulty of intubation. This work presents an automatic method to assess the modified Mallampati score from an image of a patient with the mouth wide open. For this purpose we propose an active appearance models (AAM) based method and use linear support vector machines (SVM) to select a subset of relevant features obtained using the AAM. This feature selection step proves to be essential as it improves drastically the performance of classification, which is obtained using SVM with RBF kernel and majority voting. We test our method on images of 100 patients undergoing elective surgery and achieve 97.9% accuracy in the leave-one-out crossvalidation test and provide a key element to an automatic difficult intubation assessment system.

Modified 'dumbbell' technique: a simple and intuitive method to position balloon-expandable stent valves.

Intraoperative cardiac imaging plays a key role during transcatheter aortic valve replacement. In recent years, new techniques and new tools for improved image quality and virtual navigation have been proposed, in order to simplify and standardize stent valve positioning and implantation. But routine performance of the new techniques may require major economic investments or specific knowledge and skills and, for this reason, they may not be accessible to the majority of cardiac centres involved in transcatheter valve replacement projects. Additionally, they still require injections of contrast medium to obtain computed images. Therefore, we have developed and describe here a very simple and intuitive method of positioning balloon-expandable stent valves, which represents the evolution of the 'dumbbell' technique for echocardiography-guided transcatheter valve replacement without angiography. This method, based on the partial inflation of the balloon catheter during positioning, traps the crimped valve in the aortic valve orifice and, consequently, very near to the ideal landing zone. It does not require specific echocardiographic knowledge; it does not require angiographies that increase the risk of postoperative kidney failure in elderly patients, and it can be also performed in centres not equipped with a hybrid operating room.

Inexperienced sonographers can successfully visualize and assess a three-dimensional image of the fetal face using a standardized ultrasound protocol

Introduction: A standardized three-dimensional ultrasonographic (3DUS) protocol is described that allows fetal face reconstruction. Ability to identify cleft lip with 3DUS using this protocol was assessed by operators with minimal 3DUS experience. Material and Methods: 260 stored volumes of fetal face were analyzed using a standardized protocol by operators with different levels of competence in 3DUS. The outcomes studied were: (1) the performance of post-processing 3D face volumes for the detection of facial clefts; (2) the ability of a resident with minimal 3DUS experience to reconstruct the acquired facial volumes, and (3) the time needed to reconstruct each plane to allow proper diagnosis of a cleft. Results: The three orthogonal planes of the fetal face (axial, sagittal and coronal) were adequately reconstructed with similar performance when acquired by a maternal-fetal medicine specialist or by residents with minimal experience (72 vs. 76%, p = 0.629). The learning curve for manipulation of 3DUS volumes of the fetal face corresponds to 30 cases and is independent of the operator's level of experience. Discussion: The learning curve for the standardized protocol we describe is short, even for inexperienced sonographers. This technique might decrease the length of anatomy ultrasounds and improve the ability to visualize fetal face anomalies.

High-resolution selective three-dimensional magnetic resonance coronary angiography with navigator-echo technique: segment-by-segment evaluation of coronary artery stenosis.

PURPOSE: To investigate the feasibility of high-resolution selective three-dimensional (3D) magnetic resonance coronary angiography (MRCA) in the evaluation of coronary artery stenoses. MATERIALS AND METHODS: In 12 patients with coronary artery stenoses, MRCA of the coronary artery groups, including the coronary segments with stenoses of 50% or greater based on conventional x-ray coronary angiography (CAG), was performed with double-oblique imaging planes by orienting the 3D slab along the major axis of each right coronary artery-left circumflex artery (RCA-LCX) group and each left main trunk-left anterior descending artery (LMT-LAD) group. Ten RCA-LCX and five LMT-LAD MR angiograms were obtained, and the results were compared with those of conventional x-ray angiography. RESULTS: Among 70 coronary artery segments expected to be covered, a total of 49 (70%) segments were fully demonstrated in diagnostic quality. The identification of segmental location of stenoses showed as high an accuracy as 96%. The retrospective analysis for stenosis of 50% or greater yielded the sensitivity, specificity, and accuracy of 80%, 85%, and 84%, respectively. CONCLUSION: Selective 3D MRCA has the potential for segment-by-segment evaluation of major portions of the right and left coronary arteries with high accuracy.

IEF pattern classification-derived criteria for the identification of epoetin-delta in urine.

Epoetin-delta (Dynepo Shire Pharmaceuticals, Basing stoke, UK) is a synthetic form of erythropoietin (EPO) whose resemblance with endogenous EPO makes it hard to identify using the classical identification criteria. Urine samples collected from six healthy volunteers treated with epoetin-delta injections and from a control population were immuno-purified and analyzed with the usual IEF method. On the basis of the EPO profiles integration, a linear multivariate model was computed for discriminant analysis. For each sample, a pattern classification algorithm returned a bands distribution and intensity score (bands intensity score) saying how representative this sample is of one of the two classes, positive or negative. Effort profiles were also integrated in the model. The method yielded a good sensitivity versus specificity relation and was used to determine the detection window of the molecule following multiple injections. The bands intensity score, which can be generalized to epoetin-alpha and epoetin-beta, is proposed as an alternative criterion and a supplementary evidence for the identification of EPO abuse.

Cluster kernels for semisupervised classification of VHR urban images

In this paper, we present and apply a semisupervised support vector machine based on cluster kernels for the problem of very high resolution image classification. In the proposed setting, a base kernel working with labeled samples only is deformed by a likelihood kernel encoding similarities between unlabeled examples. The resulting kernel is used to train a standard support vector machine (SVM) classifier. Experiments carried out on very high resolution (VHR) multispectral and hyperspectral images using very few labeled examples show the relevancy of the method in the context of urban image classification. Its simplicity and the small number of parameters involved make it versatile and workable by unexperimented users.

MDCT Arthrography of the Hip: Value of the Adaptive Statistical Iterative Reconstruction Technique and Potential for Radiation Dose Reduction

Au cours des deux dernières décennies, la technique d'imagerie arthro-scanner a bénéficié de nombreux progrès technologiques et représente aujourd'hui une excellente alternative à l'imagerie par résonance magnétique (IRM) et / ou arthro-IRM dans l'évaluation des pathologies de la hanche. Cependant, elle reste limitée par l'exposition aux rayonnements ionisants importante. Les techniques de reconstruction itérative (IR) ont récemment été mis en oeuvre avec succès en imagerie ; la littérature montre que l'utilisation ces dernières contribue à réduire la dose d'environ 40 à 55%, comparativement aux protocoles courants utilisant la rétroprojection filtrée (FBP), en scanner de rachis. A notre connaissance, l'utilisation de techniques IR en arthro-scanner de hanche n'a pas été évaluée jusqu'à présent. Le but de notre étude était d'évaluer l'impact de la technique ASIR (GE Healthcare) sur la qualité de l'image objective et subjective en arthro-scanner de hanche, et d'évaluer son potentiel en terme de réduction de dose. Pour cela, trente sept patients examinés par arthro-scanner de hanche ont été randomisés en trois groupes : dose standard (CTDIvol = 38,4 mGy) et deux groupes de dose réduite (CTDIvol = 24,6 ou 15,4 mGy). Les images ont été reconstruites en rétroprojection filtrée (FBP) puis en appliquant différents pourcentages croissants d'ASIR (30, 50, 70 et 90%). Le bruit et le rapport contraste sur bruit (CNR) ont été mesurés. Deux radiologues spécialisés en imagerie musculo-squelettique ont évalué de manière indépendante la qualité de l'image au niveau de plusieurs structures anatomiques en utilisant une échelle de quatre grades. Ils ont également évalué les lésions labrales et du cartilage articulaire. Les résultats révèlent que le bruit augmente (p = 0,0009) et le CNR diminue (p = 0,001) de manière significative lorsque la dose diminue. A l'inverse, le bruit diminue (p = 0,0001) et le contraste sur bruit augmente (p < 0,003) de manière significative lorsque le pourcentage d'ASIR augmente ; on trouve également une augmentation significative des scores de la qualité de l'image pour le labrum, le cartilage, l'os sous-chondral, la qualité de l'image globale (au delà de ASIR 50%), ainsi que le bruit (p < 0,04), et une réduction significative pour l'os trabuculaire et les muscles (p < 0,03). Indépendamment du niveau de dose, il n'y a pas de différence significative pour la détection et la caractérisation des lésions labrales (n=24, p = 1) et des lésions cartilagineuses (n=40, p > 0,89) en fonction du pourcentage d'ASIR. Notre travail a permis de montrer que l'utilisation de plus de 50% d'ASIR permet de reduire de manière significative la dose d'irradiation reçue par le patient lors d'un arthro-scanner de hanche tout en maintenant une qualité d'image diagnostique comparable par rapport à un protocole de dose standard utilisant la rétroprojection filtrée.

Learning spatial filters for multispectral image segmentation

We present a novel filtering method for multispectral satellite image classification. The proposed method learns a set of spatial filters that maximize class separability of binary support vector machine (SVM) through a gradient descent approach. Regularization issues are discussed in detail and a Frobenius-norm regularization is proposed to efficiently exclude uninformative filters coefficients. Experiments carried out on multiclass one-against-all classification and target detection show the capabilities of the learned spatial filters.

A novel technique using echocardiography to evaluate venous cannula performance perioperatively in CPB cardiac surgery.

OBJECTIVE: Transthoracic echocardiography (TTE) has been used clinically to disobstruct venous drainage cannula and to optimise placement of venous cannulae in the vena cava but it has never been used to evaluate performance capabilities. Also, little progress has been made in venous cannula design in order to optimise venous return to the heart lung machine. We designed a self-expandable Smartcanula (SC) and analysed its performance capability using echocardiography. METHODS: An epicardial echocardiography probe was placed over the SC or control cannula (CTRL) and a Doppler image was obtained. Mean (V(m)) and maximum (V(max)) velocities, flow and diameter were obtained. Also, pressure drop (DeltaP(CPB)) was obtained between the central venous pressure and inlet to venous reservoir. LDH and Free Hb were also compared in 30 patients. Comparison was made between the two groups using the student's t-test with statistical significance established when p<0.05. RESULTS: Age for the SC and CC groups were 61.6+/-17.6 years and 64.6+/-13.1 years, respectively. Weight was 70.3+/-11.6 kg and 72.8+/-14.4 kg, respectively. BSA was 1.80+/-0.2 m(2) and 1.82+/-0.2 m(2), respectively. CPB times were 114+/-53 min and 108+/-44 min, respectively. Cross-clamp time was 59+/-15 min and 76+/-29 min, respectively (p=NS). Free-Hb was 568+/-142 U/l versus 549+/-271 U/l post-CPB for the SC and CC, respectively (p=NS). LDH was 335+/-73 mg/l versus 354+/-116 mg/l for the SC and CC, respectively (p=NS). V(m) was 89+/-10 cm/s (SC) versus 63+/-3 cm/s (CC), V(max) was 139+/-23 cm/s (SC) versus 93+/-11 cm/s (CC) (both p<0.01). DeltaP(CPB) was 30+/-10 mmHg (SC) versus 43+/-13 mmHg (CC) (p<0.05). A Bland-Altman test showed good agreement between the two devices used concerning flow rate calculations between CPB and TTE (bias 300 ml+/-700 ml standard deviation). CONCLUSIONS: This novel Smartcanula design, due to its self-expanding principle, provides superior flow characteristics compared to classic two stage venous cannula used for adult CPB surgery. No detrimental effects were observed concerning blood damage. Echocardiography was effective in analysing venous cannula performance and velocity patterns.

Paediatric cardiac CT examinations: impact of the iterative reconstruction method ASIR on image quality--preliminary findings.

BACKGROUND: Radiation dose exposure is of particular concern in children due to the possible harmful effects of ionizing radiation. The adaptive statistical iterative reconstruction (ASIR) method is a promising new technique that reduces image noise and produces better overall image quality compared with routine-dose contrast-enhanced methods. OBJECTIVE: To assess the benefits of ASIR on the diagnostic image quality in paediatric cardiac CT examinations. MATERIALS AND METHODS: Four paediatric radiologists based at two major hospitals evaluated ten low-dose paediatric cardiac examinations (80 kVp, CTDI(vol) 4.8-7.9 mGy, DLP 37.1-178.9 mGy·cm). The average age of the cohort studied was 2.6 years (range 1 day to 7 years). Acquisitions were performed on a 64-MDCT scanner. All images were reconstructed at various ASIR percentages (0-100%). For each examination, radiologists scored 19 anatomical structures using the relative visual grading analysis method. To estimate the potential for dose reduction, acquisitions were also performed on a Catphan phantom and a paediatric phantom. RESULTS: The best image quality for all clinical images was obtained with 20% and 40% ASIR (p < 0.001) whereas with ASIR above 50%, image quality significantly decreased (p < 0.001). With 100% ASIR, a strong noise-free appearance of the structures reduced image conspicuity. A potential for dose reduction of about 36% is predicted for a 2- to 3-year-old child when using 40% ASIR rather than the standard filtered back-projection method. CONCLUSION: Reconstruction including 20% to 40% ASIR slightly improved the conspicuity of various paediatric cardiac structures in newborns and children with respect to conventional reconstruction (filtered back-projection) alone.

Comparison of Energy Minimization Methods for 3-D Brain Tissue Classification

This paper presents 3-D brain tissue classificationschemes using three recent promising energy minimizationmethods for Markov random fields: graph cuts, loopybelief propagation and tree-reweighted message passing.The classification is performed using the well knownfinite Gaussian mixture Markov Random Field model.Results from the above methods are compared with widelyused iterative conditional modes algorithm. Theevaluation is performed on a dataset containing simulatedT1-weighted MR brain volumes with varying noise andintensity non-uniformities. The comparisons are performedin terms of energies as well as based on ground truthsegmentations, using various quantitative metrics.

Assessment of a new MRI classification of spinal stenosis based on cross-sectional morphology of the dural sac : FM2

Introduction: Quantitative measures of degree of lumbar spinal stenosis (LSS) such as antero-posterior diameter of the canal or dural sac cross sectional area vary widely and do not correlate with clinical symptoms or results of surgical decompression. In an effort to improve quantification of stenosis we have developed a grading system based on the morphology of the dural sac and its contents as seen on T2 axial images. The grading comprises seven categories ranging form normal to the most severe stenosis and takes into account the ratio of rootlet/CSF content. Material and methods: Fifty T2 axial MRI images taken at disc level from twenty seven symptomatic lumbar spinal stenosis patients who underwent decompressive surgery were classified into seven categories by five observers and reclassified 2 weeks later by the same investigators. Intra- and inter-observer reliability of the classification were assessed using Cohen's and Fleiss' kappa statistics, respectively. Results: Generally, the morphology grading system itself was well adopted by the observers. Its success in application is strongly influenced by the identification of the dural sac. The average intraobserver Cohen's kappa was 0.53 ± 0.2. The inter-observer Fleiss' kappa was 0.38 ± 0.02 in the first rating and 0.3 ± 0.03 in the second rating repeated after two weeks. Discussion: In this attempt, the teaching of the observers was limited to an introduction to the general idea of the morphology grading system and one example MRI image per category. The identification of the dimension of the dural sac may be a difficult issue in absence of complete T1 T2 MRI image series as it was the case here. The similarity of the CSF to possibly present fat on T2 images was the main reason of mismatch in the assignment of the cases to a category. The Fleiss correlation factors of the five observers are fair and the proposed morphology grading system is promising.

Automatic classification of MR scans in Alzheimer's disease.

To be diagnostically useful, structural MRI must reliably distinguish Alzheimer's disease (AD) from normal aging in individual scans. Recent advances in statistical learning theory have led to the application of support vector machines to MRI for detection of a variety of disease states. The aims of this study were to assess how successfully support vector machines assigned individual diagnoses and to determine whether data-sets combined from multiple scanners and different centres could be used to obtain effective classification of scans. We used linear support vector machines to classify the grey matter segment of T1-weighted MR scans from pathologically proven AD patients and cognitively normal elderly individuals obtained from two centres with different scanning equipment. Because the clinical diagnosis of mild AD is difficult we also tested the ability of support vector machines to differentiate control scans from patients without post-mortem confirmation. Finally we sought to use these methods to differentiate scans between patients suffering from AD from those with frontotemporal lobar degeneration. Up to 96% of pathologically verified AD patients were correctly classified using whole brain images. Data from different centres were successfully combined achieving comparable results from the separate analyses. Importantly, data from one centre could be used to train a support vector machine to accurately differentiate AD and normal ageing scans obtained from another centre with different subjects and different scanner equipment. Patients with mild, clinically probable AD and age/sex matched controls were correctly separated in 89% of cases which is compatible with published diagnosis rates in the best clinical centres. This method correctly assigned 89% of patients with post-mortem confirmed diagnosis of either AD or frontotemporal lobar degeneration to their respective group. Our study leads to three conclusions: Firstly, support vector machines successfully separate patients with AD from healthy aging subjects. Secondly, they perform well in the differential diagnosis of two different forms of dementia. Thirdly, the method is robust and can be generalized across different centres. This suggests an important role for computer based diagnostic image analysis for clinical practice.