On the evaluation of outlier detection and one-class classification: a comparative study of algorithms, model selection, and ensembles.

It has been shown that unsupervised outlier detection methods can be adapted to the one-class classification problem (Janssens and Postma, in: Proceedings of the 18th annual Belgian-Dutch on machine learning, pp 56-64, 2009; Janssens et al. in: Proceedings of the 2009 ICMLA international conference on machine learning and applications, IEEE Computer Society, pp 147-153, 2009. 10.1109/ICMLA.2009.16). In this paper, we focus on the comparison of one-class classification algorithms with such adapted unsupervised outlier detection methods, improving on previous comparison studies in several important aspects. We study a number of one-class classification and unsupervised outlier detection methods in a rigorous experimental setup, comparing them on a large number of datasets with different characteristics, using different performance measures. In contrast to previous comparison studies, where the models (algorithms, parameters) are selected by using examples from both classes (outlier and inlier), here we also study and compare different approaches for model selection in the absence of examples from the outlier class, which is more realistic for practical applications since labeled outliers are rarely available. Our results showed that, overall, SVDD and GMM are top-performers, regardless of whether the ground truth is used for parameter selection or not. However, in specific application scenarios, other methods exhibited better performance. Combining one-class classifiers into ensembles showed better performance than individual methods in terms of accuracy, as long as the ensemble members are properly selected. Supplementary Information: The online version contains supplementary material available at 10.1007/s10618-023-00931-x.

Reconstruction and completion of high-resolution 3D cardiac shapes using anisotropic CMRI segmentations and continuous implicit neural representations.

Since the onset of computer-aided diagnosis in medical imaging, voxel-based segmentation has emerged as the primary methodology for automatic analysis of left ventricle (LV) function and morphology in cardiac magnetic resonance images (CMRI). In standard clinical practice, simultaneous multi-slice 2D cine short-axis MR imaging is performed under multiple breath-holds resulting in highly anisotropic 3D images. Furthermore, sparse-view CMRI often lacks whole heart coverage caused by large slice thickness and often suffers from inter-slice misalignment induced by respiratory motion. Therefore, these volumes only provide limited information about the true 3D cardiac anatomy which may hamper highly accurate assessment of functional and anatomical abnormalities. To address this, we propose a method that learns a continuous implicit function representing 3D LV shapes by training an auto-decoder. For training, high-resolution segmentations from cardiac CT angiography are used. The ability of our approach to reconstruct and complete high-resolution shapes from manually or automatically obtained sparse-view cardiac shape information is evaluated by using paired high- and low-resolution CMRI LV segmentations. The results show that the reconstructed LV shapes have an unconstrained subvoxel resolution and appear smooth and plausible in through-plane direction. Furthermore, Bland-Altman analysis reveals that reconstructed high-resolution ventricle volumes are closer to the corresponding reference volumes than reference low-resolution volumes with bias of [limits of agreement] -3.51 [-18.87, 11.85] mL, and 12.96 [-10.01, 35.92] mL respectively. Finally, the results demonstrate that the proposed approach allows recovering missing shape information and can indirectly correct for limited motion-induced artifacts.

Towards automatic classification of cardiovascular magnetic resonance Task Force Criteria for diagnosis of arrhythmogenic right ventricular cardiomyopathy.

BACKGROUND: Arrhythmogenic right ventricular cardiomyopathy (ARVC) is diagnosed according to the Task Force Criteria (TFC) in which cardiovascular magnetic resonance (CMR) imaging plays an important role. Our study aims to apply an automatic deep learning-based segmentation for right and left ventricular CMR assessment and evaluate this approach for classification of the CMR TFC. METHODS: We included 227 subjects suspected of ARVC who underwent CMR. Subjects were classified into (1) ARVC patients fulfilling TFC; (2) at-risk family members; and (3) controls. To perform automatic segmentation, a Bayesian Dilated Residual Neural Network was trained and tested. Performance of automatic versus manual segmentation was assessed using Dice-coefficient and Hausdorff distance. Since automatic segmentation is most challenging in basal slices, manual correction of the automatic segmentation in the most basal slice was simulated (automatic-basal). CMR TFC calculated using manual and automatic-basal segmentation were compared using Cohen's Kappa (κ). RESULTS: Automatic segmentation was trained on CMRs of 70 subjects (39.6 ± 18.1 years, 47% female) and tested on 157 subjects (36.9 ± 17.6 years, 59% female). Dice-coefficient and Hausdorff distance showed good agreement between manual and automatic segmentations (≥ 0.89 and ≤ 10.6 mm, respectively) which further improved after simulated correction of the most basal slice (≥ 0.92 and ≤ 9.2 mm, p < 0.001). Pearson correlation of volumetric and functional CMR measurements was good to excellent (automatic (r = 0.78-0.99, p < 0.001) and automatic-basal (r = 0.88-0.99, p < 0.001) measurements). CMR TFC classification using automatic-basal segmentations was comparable to manual segmentations (κ 0.98 ± 0.02) with comparable diagnostic performance. CONCLUSIONS: Combining automatic segmentation of CMRs with correction of the most basal slice results in accurate CMR TFC classification of subjects suspected of ARVC.

Autoencoding low-resolution MRI for semantically smooth interpolation of anisotropic MRI.

High-resolution medical images are beneficial for analysis but their acquisition may not always be feasible. Alternatively, high-resolution images can be created from low-resolution acquisitions using conventional upsampling methods, but such methods cannot exploit high-level contextual information contained in the images. Recently, better performing deep-learning based super-resolution methods have been introduced. However, these methods are limited by their supervised character, i.e. they require high-resolution examples for training. Instead, we propose an unsupervised deep learning semantic interpolation approach that synthesizes new intermediate slices from encoded low-resolution examples. To achieve semantically smooth interpolation in through-plane direction, the method exploits the latent space generated by autoencoders. To generate new intermediate slices, latent space encodings of two spatially adjacent slices are combined using their convex combination. Subsequently, the combined encoding is decoded to an intermediate slice. To constrain the model, a notion of semantic similarity is defined for a given dataset. For this, a new loss is introduced that exploits the spatial relationship between slices of the same volume. During training, an existing in-between slice is generated using a convex combination of its neighboring slice encodings. The method was trained and evaluated using publicly available cardiac cine, neonatal brain and adult brain MRI scans. In all evaluations, the new method produces significantly better results in terms of Structural Similarity Index Measure and Peak Signal-to-Noise Ratio (p<0.001 using one-sided Wilcoxon signed-rank test) than a cubic B-spline interpolation approach. Given the unsupervised nature of the method, high-resolution training data is not required and hence, the method can be readily applied in clinical settings.

Biological data annotation via a human-augmenting AI-based labeling system.

Biology has become a prime area for the deployment of deep learning and artificial intelligence (AI), enabled largely by the massive data sets that the field can generate. Key to most AI tasks is the availability of a sufficiently large, labeled data set with which to train AI models. In the context of microscopy, it is easy to generate image data sets containing millions of cells and structures. However, it is challenging to obtain large-scale high-quality annotations for AI models. Here, we present HALS (Human-Augmenting Labeling System), a human-in-the-loop data labeling AI, which begins uninitialized and learns annotations from a human, in real-time. Using a multi-part AI composed of three deep learning models, HALS learns from just a few examples and immediately decreases the workload of the annotator, while increasing the quality of their annotations. Using a highly repetitive use-case-annotating cell types-and running experiments with seven pathologists-experts at the microscopic analysis of biological specimens-we demonstrate a manual work reduction of 90.60%, and an average data-quality boost of 4.34%, measured across four use-cases and two tissue stain types.

Automatic segmentation with detection of local segmentation failures in cardiac MRI.

Segmentation of cardiac anatomical structures in cardiac magnetic resonance images (CMRI) is a prerequisite for automatic diagnosis and prognosis of cardiovascular diseases. To increase robustness and performance of segmentation methods this study combines automatic segmentation and assessment of segmentation uncertainty in CMRI to detect image regions containing local segmentation failures. Three existing state-of-the-art convolutional neural networks (CNN) were trained to automatically segment cardiac anatomical structures and obtain two measures of predictive uncertainty: entropy and a measure derived by MC-dropout. Thereafter, using the uncertainties another CNN was trained to detect local segmentation failures that potentially need correction by an expert. Finally, manual correction of the detected regions was simulated in the complete set of scans of 100 patients and manually performed in a random subset of scans of 50 patients. Using publicly available CMR scans from the MICCAI 2017 ACDC challenge, the impact of CNN architecture and loss function for segmentation, and the uncertainty measure was investigated. Performance was evaluated using the Dice coefficient, 3D Hausdorff distance and clinical metrics between manual and (corrected) automatic segmentation. The experiments reveal that combining automatic segmentation with manual correction of detected segmentation failures results in improved segmentation and to 10-fold reduction of expert time compared to manual expert segmentation.

A unified view of density-based methods for semi-supervised clustering and classification.

Semi-supervised learning is drawing increasing attention in the era of big data, as the gap between the abundance of cheap, automatically collected unlabeled data and the scarcity of labeled data that are laborious and expensive to obtain is dramatically increasing. In this paper, we first introduce a unified view of density-based clustering algorithms. We then build upon this view and bridge the areas of semi-supervised clustering and classification under a common umbrella of density-based techniques. We show that there are close relations between density-based clustering algorithms and the graph-based approach for transductive classification. These relations are then used as a basis for a new framework for semi-supervised classification based on building-blocks from density-based clustering. This framework is not only efficient and effective, but it is also statistically sound. In addition, we generalize the core algorithm in our framework, HDBSCAN*, so that it can also perform semi-supervised clustering by directly taking advantage of any fraction of labeled data that may be available. Experimental results on a large collection of datasets show the advantages of the proposed approach both for semi-supervised classification as well as for semi-supervised clustering.

Chemoenzymatic Synthesis of a Characteristic Glycophosphopeptide from the Transactivation Domain of the Serum Response Factor.

Glycopeptides, phosphopeptides, and glycophosphopeptides can be synthesized efficiently by a strategy based on a combination of suitable enzyme-labile protecting groups. Thus, probes for biological studies can be accessed. An example is the glycosylated and phosphorylated heptapeptide 1 from the transactivation domain of the human serum response factor, which contains an additional biotin label for detection with streptavidin.

A simpler and more accurate AUTO-HDS framework for clustering and visualization of biological data.

In [1], the authors proposed a framework for automated clustering and visualization of biological data sets named AUTO-HDS. This letter is intended to complement that framework by showing that it is possible to get rid of a user-defined parameter in a way that the clustering stage can be implemented more accurately while having reduced computational complexity.

Combining gene expression and interaction network data to improve kidney lesion score prediction.

Current method of diagnosing kidney rejection based on histopathology of renal biopsies in form of lesion scores is error-prone. Researchers use gene expression microarrays in combination of machine learning to build better kidney rejection predictors. However the high dimensionality of data makes this task challenging and compels application of feature selection methods. We present a method for predicting lesions using combination of statistical and biological feature selection methods along with an ensemble learning technique. Results show that combining highly interacting genes (Hub Genes) from protein-protein interaction network with genes selected by squared t-test method brings the most accurate kidney lesion score predictor.

Finding the nearest neighbors in biological databases using less distance computations.

Modern biological applications usually involve the similarity comparison between two objects, which is often computationally very expensive, such as whole genome pairwise alignment and protein 3D structure alignment. Nevertheless, being able to quickly identify the closest neighboring objects from very large databases for a newly obtained sequence or structure can provide timely hints to its functions and more. This paper presents a substantial speedup technique for the well-studied k-nearest neighbor (k-nn) search, based on novel concepts of virtual pivots and partial pivots, such that a significant number of the expensive distance computations can be avoided. The new method is able to dynamically locate virtual pivots, according to the query, with increasing pruning ability. Using the same or less amount of database preprocessing effort, the new method outperformed the second best method by using no more than 40 percent distance computations per query, on a database of 10,000 gene sequences, compared to several best known k-nn search methods including M-Tree, OMNI, SA-Tree, and LAESA. We demonstrated the use of this method on two biological sequence data sets, one of which is for HIV-1 viral strain computational genotyping.

Efficient composite pattern finding from monad patterns.

Automatically identifying frequent composite patterns in DNA sequences is an important task in bioinformatics, especially when all the basic elements (or monad patterns) of a composite pattern are weak. In this paper, we compare one straightforward approach to assemble the monad patterns into composite patterns to two other rather complex approaches. Both our theoretical analysis and empirical results show that this overlooked straightforward method can be several orders of magnitude faster. Furthermore, different from the previous understandings, the empirical results show that the runtime superiority among the three approaches is closely related to the insignificance of the monad patterns.