Rejection Classifier Research Articles

A 2-fold Approach to Polyoma Virus (BK) Nephropathy in Kidney Transplants: Distinguishing Direct Virus Effects From Cognate T Cell-mediated Inflammation.

BK nephropathy (BKN) in kidney transplants diagnosed by histology is challenging because it involves damage from both virus activity and cognate T cell-mediated inflammation, directed against alloantigens (rejection) or viral antigens. The present study of indication biopsies from the Integrated Diagnostic System in the International Collaborative Microarray Study Extension study measured major capsid viral protein 2 (VP2) mRNA to assess virus activity and a T cell-mediated rejection (TCMR) classifier to assess cognate T cell-mediated inflammation. Biopsies were assessed by local standard-of-care histology and by genome-wide microarrays and Molecular Microscope Diagnostic System (MMDx) algorithms to detect rejection and injury. In a subset of 102 biopsies (50 BKN and 52 BKN-negative biopsies with various abnormalities), we measured VP2 transcripts by real-time polymerase chain reaction. BKN was diagnosed in 55 of 1679 biopsies; 30 had cognate T cell-mediated activity assessed by by MMDx and TCMR lesions, but only 3 of 30 were histologically diagnosed as TCMR. We developed a BKN probability classifier that predicted histologic BKN (area under the curve = 0.82). Virus activity (VP2 expression) was highly selective for BKN (area under the curve = 0.94) and correlated with acute injury, atrophy-fibrosis, macrophage activation, and the BKN classifier, but not with the TCMR classifier. BKN with molecular TCMR had more tubulitis and inflammation than BKN without molecular TCMR. In 5 BKN cases with second biopsies, VP2 mRNA decreased in second biopsies, whereas in 4 of 5 TCMR classifiers, scores increased. Genes and pathways associated with BKN and VP2 mRNA were similar, reflecting injury, inflammation, and macrophage activation but none was selective for BKN. Risk-benefit decisions in BKN may be assisted by quantitative assessment of the 2 major pathologic processes, virus activity and cognate T cell-mediated inflammation.

Performance visualization spaces for classification with rejection option

The classification with reject option consists to train a classifier that rejects the examples when the confidence in its prediction is low. The objective is to improve the accuracy of the non-rejected examples and the reliability of the prediction. The performances of the reject classifiers depend on both the error rate and rejection rate. Since these two values are in opposition, we have to make a trade-off between them. This paper is focused on the visualization spaces the performances of the classifiers with rejection option. We analyze two common spaces, the ROC space and the error-rejection (ER) space, then we propose a new space: the cost-reject (CR) space. We show that the ROC space is the less convenient space to represent the performances of the reject classifier. However, it can be recommended for classification problems where the importance of the two classes is different. For the ER space, we point out that the linear interpolation that is commonly used to draw the error-reject curve is not correct and leads to an overestimation of the classifier performances. From the definition of the condition error and rejection rate, we propose a new interpolation of the error-rejection curve that is unbiased. We introduce a new visualization space called the cost reject space. The CR space plots the normalized classification cost in function on the normalized rejection cost. The performance of a classifier is represented in this space by a line. The three visualization spaces are compared on problems of classification algorithms comparison. The advantages and drawbacks of each spaces are discussed and some recommendations are provided in the conclusion.

Separate neural representations for physical pain and social rejection.

Current theories suggest that physical pain and social rejection share common neural mechanisms, largely by virtue of overlapping functional magnetic resonance imaging (fMRI) activity. Here we challenge this notion by identifying distinct multivariate fMRI patterns unique to pain and rejection. Sixty participants experience painful heat and warmth and view photos of ex-partners and friends on separate trials. FMRI pattern classifiers discriminate pain and rejection from their respective control conditions in out-of-sample individuals with 92% and 80% accuracy. The rejection classifier performs at chance on pain, and vice versa. Pain- and rejection-related representations are uncorrelated within regions thought to encode pain affect (for example, dorsal anterior cingulate) and show distinct functional connectivity with other regions in a separate resting-state data set (N = 91). These findings demonstrate that separate representations underlie pain and rejection despite common fMRI activity at the gross anatomical level. Rather than co-opting pain circuitry, rejection involves distinct affective representations in humans.

Expression robust three-dimensional face recognition based on gaussian filter and dual-tree complex wavelet transform

In this paper, a fully automatic framework is proposed for 3D face recognition and its superiority performance is justified by the FRGC v2 data. For 3D data preprocessing, a new face smoothing method is proposed. Meanwhile, 3D facial representation, which is extracted by the Dual-tree Complex Wavelet Transform DT-CWT, is introduced to reflect the facial geometry properties. Low redundancy makes it more effective and efficient to describe the discriminant feature in 2.5D range data. In this paper, DT-CWT is used into 2.5D range data in conjunction with the Linear Discriminant Analysis LDA to form a rejection classifier, which can quickly eliminate a large number of candidate gallery faces. The remaining faces are then verified using sparse representation based classification. Our method achieves the verification rate of 98.66% on All vs. All experiment at an FAR of 0.1%.

An SVM-AdaBoost facial expression recognition system

This study is focused on improving the recognition rate and processing time of facial recognition systems. First, the skin is detected by pixel based methods to reduce the searching space for maximum rejection classifier (MRC) which detects the face. The detected face is normalized by a discrete cosine transform (DCT) and down-sampled by Bessel transform. Gabor feature extraction techniques were utilized to extract thousands of facial features that represent facial deformation patterns. An AdaBoost-based hypothesis is formulated to select a few hundreds of Gabor features which are potential candidates for expression recognition. The selected features were fed into a saturated vector machine (SVM) classifier to train it. An average recognition rate of 97.57 % and 92.33 % are registered in JAFFE and Yale databases respectively. The execution time of the proposed method is also significantly lower than others. Generally, the proposed method exhibits superior performance than other methods.

Multi-stage classifier design

In many classification systems, sensing modalities have different acquisition costs. It is often unnecessary to use every modality to classify a majority of examples. We study a multi-stage system in a prediction time cost reduction setting, where the full data is available for training, but for a test example, measurements in a new modality can be acquired at each stage for an additional cost. We seek decision rules to reduce the average measurement acquisition cost. We formulate an empirical risk minimization problem (ERM) for a multi-stage reject classifier, wherein the stage k classifier either classifies a sample using only the measurements acquired so far or rejects it to the next stage where more attributes can be acquired for a cost. If we restrict ourselves to binary classification setting then, to solve the ERM problem, we show that the optimal reject classifier at each stage is a combination of two binary classifiers, one biased towards positive examples and the other biased towards negative examples. We use this parameterization to construct stage-by-stage global surrogate risk, develop an iterative algorithm in the boosting framework and present convergence and generalization results. We test our work on synthetic, medical and explosives detection datasets. Our results demonstrate that substantial cost reduction without a significant sacrifice in accuracy is achievable.

Robust Facial Data Recognition using Multimodal Fusion Features in Multi-Variant Face Acquisition

Biometric use physiological traits such as fingerprints, face and behavioral traits such as voice, hand signatures characteristics to verify an individual’s identity. The two process involved in biometrics are verification and identification. Verification process is performed by matching an individual’s biometric with the template of claimed identity only. The identification process performed by matching an individual’s biometric with template of every identity in the database. Existing work presented a multimodal (2D and 3D) face recognition algorithm by way of performing hybrid matching which was based on both feature and holistic metrics. The Pose of 3D face and its texture corrected using single automatically detected point and hotelling transform. 3D Spherical Face Representation (SFR) is used in conjunction with Scale-Invariant Feature Transform (SIFT) descriptor which results in formation of a rejection classifier. It quickly eliminates a large number of candidate faces at an early stage for efficient recognition for known faces as attributes. Our proposed work presented a Feature Based Multimodality Face Recognition System to recognize the human individuals in environment of known faces using features like shape of the eyes, nose and jaw. Case study and preliminary experimental results conducted in Mat lab proves to be a viable approach using multimodality method based on 2D and 3D facial representation for known faces. Facial recognition rate is measured in terms of validation index generated for false acceptance rate, false rejection rate.

Confidence-Based Rejection for Improved Pattern Recognition Myoelectric Control

This study describes a novel myoelectric control scheme that is capable of motion rejection. As an extension of the commonly used linear discriminant analysis (LDA), this system generates a confidence score for each decision, providing the ability to reject those with a score below a selected threshold. The thresholds are class-specific and affect only the rejection characteristics of the associated class. Furthermore, because the rejection stage is implemented using the outputs of the LDA, the active motion classification accuracy of the proposed system is shown to outperform that of the LDA for all values of rejection threshold. The proposed scheme was compared to a baseline LDA-based pattern recognition system using a real-time Fitts' law-based target acquisition task. The use of velocity-based myoelectric control using the rejection classifier is shown to obey Fitts' law, producing linear regression fittings with high coefficients of determination (R(2) > 0.943). Significantly higher (p < 0.001) throughput, path efficiency, and completion rates were observed with the rejection-capable system for both able-bodied and amputee subjects.

Efficient 3D face recognition handling facial expression and hair occlusion

This paper presents an efficient 3D face recognition method to handle facial expression and hair occlusion. The proposed method uses facial curves to form a rejection classifier and produce a facial deformation mapping and then adaptively selects regions for matching. When a new 3D face with an arbitrary pose and expression is queried, the pose is normalized based on the automatically detected nose tip and the principal component analysis (PCA) follows. Then, the facial curve in the nose region is extracted and used to form the rejection classifier which quickly eliminates dissimilar faces in the gallery for efficient recognition. Next, six facial regions which cover the face are segmented and curves in these regions are used to map facial deformation. Regions used for matching are automatically selected based on the deformation mapping. In the end, results of all the matching engines are fused by weighted sum rule. The approach is applied on the FRGC v2.0 dataset and a verification rate of 96.0% for ROC III is achieved as a false acceptance rate (FAR) of 0.1%. In the identification scenario, a rank-one accuracy of 97.8% is achieved.

The Nature of Biopsies with “Borderline Rejection” and Prospects for Eliminating This Category

In kidney transplantation, many inflamed biopsies with changes insufficient to be called T-cell-mediated rejection (TCMR) are labeled "borderline", leaving management uncertain. This study examined the nature of borderline biopsies as a step toward eventual elimination of this category. We compared 40 borderline, 35 TCMR and 116 nonrejection biopsies. TCMR biopsies had more inflammation than borderline but similar degrees of tubulitis and scarring. Surprisingly, recovery of function after biopsy was similar in all categories, indicating that response to treatment is unreliable for defining TCMR. We studied the molecular changes in TCMR, borderline and nonrejection using microarrays, measuring four published features: T-cell burden; a rejection classifier; a canonical TCMR classifier; and risk score. These reassigned borderline biopsies as TCMR-like 13/40 (33%) or nonrejection-like 27/40 (67%). A major reason that histology diagnosed molecularly defined TCMR as borderline was atrophy-scarring, which interfered with assessment of inflammation and tubulitis. Decision tree analysis showed that i-total >27% and tubulitis extent >3% match the molecular diagnosis of TCMR in 85% of cases. In summary, most cases designated borderline by histopathology are found to be nonrejection by molecular phenotyping. Both molecular measurements and histopathology offer opportunities for more precise assignment of these cases after clinical validation.

Efficient Detection and Recognition of 3D Ears

The use of ear shape as a biometric trait is a recent trend in research. However, fast and accurate detection and recognition of the ear are very challenging because of its complex geometry. In this work, a very fast 2D AdaBoost detector is combined with fast 3D local feature matching and fine matching via an Iterative Closest Point (ICP) algorithm to obtain a complete, robust and fully automatic system with a good balance between speed and accuracy. Ear images are detected from 2D profile images using the proposed Cascaded AdaBoost detector. The corresponding 3D ear data is then extracted from the co-registered range image and represented with local 3D features. Unlike previous approaches, local features are used to construct a rejection classifier, to extract a minimal region with feature-rich data points and finally, to compute the initial transformation for matching with the ICP algorithm. The proposed system provides a detection rate of 99.9% and an identification rate of 95.4% on Collection F of the UND database. On a Core 2 Quad 9550, 2.83 GHz machine, it takes around 7.7 ms to detect an ear from a 640×480 image. Extracting features from an ear takes 22.2 sec and matching it with a gallery using only the local features takes 0.06 sec while using the full matching including ICP requires 2.28 sec on average.

Relationship between a validated molecular cardiac transplant rejection classifier and routine organ function parameters

As acute cellular cardiac allograft rejection is a systemic process affecting the entire organism, we hypothesized that scores of a peripheral blood mononuclear cell gene expression profiling (GEP) test developed and validated to rule out International Society of Heart and Lung Transplantation (ISHLT) grade > or = 3A/2R acute cellular cardiac allograft rejection also reflects biologically plausible changes of the routinely assessed clinical parameters. We retrospectively analyzed 76 patients who underwent GEP testing, at the time of their routine clinical follow-up in our Institution between February 1, 2006 and January 31, 2007. Data were analyzed with t-test, nonparametric tests, bivariate Spearman's correlation, and multivariate linear regression modeling. More activated GEP-score correlated with longer corrected QT (QTc)-interval (r = 0.377, p = 0.001, n = 63), longer QRS duration (r = 0.231, p = 0.03, n = 66), higher heart rate (r = 0.221, p = 0.037, n = 66), higher serum creatinine (r = 0.26, p = 0.01, n = 75), higher gamma-glutamyl transferase (GGT) GGT (r = 0.266, p = 0.037, n = 46), lower pulmonary artery oxygen saturation (r = -0.313, p = 0.003, n = 76), lower platelet count (r = -0.372, p = 0.001, n = 74), lower monocyte count (r = -0.208, p = 0.040, n = 72), and lower high-density lipoprotein (HDL) HDL level (r = -0.242, p = 0.041, n = 53). Multivariate analysis showed a significant amount of variance in the GEP score independently explained by the variability of QTc-interval (beta = 1.998, p = 0.001) and platelet count (beta = -1.540, p = 0.017). Post hoc analysis of the 11 individual GEP-classifier genes showed WDRA40 (p = 0.02) and ras homolog gene family, member U (RHOU) RHOU (p = 0.01) independently related to mixed venous O(2)Sat%. A GEP test developed and validated to detect the absence of cardiac rejection correlates with electrocardiographic and hemodynamic cardiac parameters as well as renal, hepatic, bone marrow, and lipid metabolism parameters suggesting a complex relationship between rejection, leukocytes, and organ function within the continuum between alloimmunological quiescence and rejection.

A Rapid Method for 3D Face Recognition Based on Rejection Algorithm

A rapid method for 3D face recognition based on rejection algorithm is proposed.First, the automatically segmented face with a different pose is normalized by means of principal component analysis(PCA), and transformed into the uniform pose coordinate system.Then, the central profile is extracted and matched based on the least trimmed square Hausdorff distance(LTS-HD) to form a rejection classifier which can eliminate a large number of candidate faces.Finally, the remaining faces are verified using a novel region-based iterative closest point(ICP) algorithm, and the result of the recognition is obtained.The simulation experiment on FRGC V2.0 database demonstrates that the proposed method is simple, efficient, and robust.

Relationship between prolonged QTc interval, cardiac allograft dysfunction and elevated molecular gene expression profiling test score after heart transplantation

Acute cellular cardiac allograft rejection after heart transplantation (HTx) is diagnosed histologically using endomyocardial biopsy (EMB), which is associated with limitations. 1 Marboe C.C. Billingham M. Eisen H. et al. Nodular endocardial infiltrates (Quilty lesions) cause significant variability in diagnosis of ISHLT Grade 2 and 3A rejection in cardiac allograft recipients. J Heart Lung Transplant. 2005; 24: S219-S226 Abstract Full Text Full Text PDF PubMed Scopus (116) Google Scholar Using gene expression profiling (GEP) of peripheral blood mononuclear cells (PBMCs), a molecular classifier was developed and validated to rule out International Society of Heart and Lung Transplantation (ISHLT) moderate/severe (Grade ≥3A/2R) rejection, 2 Deng M.C. Eisen H.J. Mehra M.R. et al. Noninvasive discrimination of rejection in cardiac allograft recipients using gene expression profiling. Am J Transplant. 2006; 6: 150-160 Crossref PubMed Scopus (414) Google Scholar with a high negative predictive value of 99.6%. The corrected QT (QTc) interval has been described as a sensitive and specific tool to predict acute cellular cardiac allograft rejection. 3 Richartz B.M. Radovancevic B. Bologna M.T. et al. Usefulness of the QTc interval in predicting acute allograft rejection. Thorac Cardiovasc Surg. 1998; 46: 217-221 Crossref PubMed Scopus (18) Google Scholar We have recently shown that the GEP score correlates with the QTc interval during quiescence. 4 Cadeiras M. Shahzad K. John M.M. et al. Relationship between a validated molecular cardiac transplant rejection classifier and routine organ function parameters. Clin Transplant. 2009; (Aug 27. [Epub ahead of print]) PubMed Google Scholar In the present pilot study we hypothesized that both increased QTc interval and elevated GEP score are associated with allograft dysfunction, as defined by echocardiographic (TTE) criteria.

Gene Expression Profiling Distinguishes a Molecular Signature for Grade 1B Mild Acute Cellular Rejection in Cardiac Allograft Recipients

Gene expression profiling distinguishes the absence or presence of moderate to severe grades of acute cellular rejection in cardiac allograft recipients using a 20-gene classifier. We explored the hypothesis that the rejection classifier also differentiates various forms of mild rejection and we performed sub-analyses based on time post-transplant and confirmatory pathology interpretations. A post hoc analysis of 265 CARGO study patients and 714 clinical encounters focused on the correlation of rejection classifier-derived gene expression (GE) scores for blood samples accompanying endomyocardial biopsies. Biopsy grades assigned by a study center pathologist (center) were re-interpreted by three pathologists (panel) in a blinded manner. Mean GE scores not only differentiated Grades >or=3A from Grade 0 (p < 0.00001, center or panel), but also from Grades 1A or 2 (p < 0.05, center or panel), based on mild rejection sub-groups defined by the ISHLT 1990 grading system. In contrast, mean GE scores for Grades 1B and >or=3A were indistinguishable, using either center or panel interpretation. Sub-group analyses of encounters from 2 to 6 months or >6 months post-transplant showed similar results for the classifier's ability to discriminate moderate to severe rejection from Grades 1A and 2 mild rejection, but indistinguishable mean GE scores for Grades >or=3A and the Grade 1B sub-group. Of the classifier's 11 informative genes, expression of MIR and WDR40 showed statistically significant increases for both Grade 1B and Grade >or=3A rejection, while expression of PDCD1 or SEMA7A showed similar directional patterns without achieving statistical significance. These data demonstrate that GE scores discriminate moderate to severe rejection from Grades 1A and 2 mild rejection. However, a sub-group of mild rejection cases, defined as Grade 1B according to the 1990 grading system, share a molecular signature more consistent with moderate to severe rejection. The clinical relevance of these data remains to be defined.

An Efficient Multimodal 2D-3D Hybrid Approach to Automatic Face Recognition

We present a fully automatic face recognition algorithm and demonstrate its performance on the FRGC v2.0 data. Our algorithm is multimodal (2D and 3D) and performs hybrid (feature-based and holistic) matching in order to achieve efficiency and robustness to facial expressions. The pose of a 3D face along with its texture is automatically corrected using a novel approach based on a single automatically detected point and the Hotelling transform. A novel 3D Spherical Face Representation (SFR) is used in conjunction with the SIFT descriptor to form a rejection classifier which quickly eliminates a large number of candidate faces at an early stage for efficient recognition in case of large galleries. The remaining faces are then verified using a novel region-based matching approach which is robust to facial expressions. This approach automatically segments the eyes-forehead and the nose regions, which are relatively less sensitive to expressions, and matches them separately using a modified ICP algorithm. The results of all the matching engines are fused at the metric level to achieve higher accuracy. We use the FRGC benchmark to compare our results to other algorithms which used the same database. Our multimodal hybrid algorithm performed better than others by achieving 99.74% and 98.31% verification rates at 0.001 FAR and identification rates of 99.02% and 95.37% for probes with neutral and non-neutral expression respectively.

Rejection based classifier for face detection

Pattern detection problems require a separation between two classes, Target and Clutter, where the probability of the former is substantially smaller compared to that of the latter. In this paper we propose a new classifier that exploits this property, yielding a low complexity yet effective target detection algorithm. This algorithm, called the maximal rejection classifier (MRC), is based on linear successive rejection operations. An application of detecting faces in images is demonstrated using the MRC with encouraging results.