U2PNet: An Unsupervised Underwater Image-Restoration Network Using Polarization.

This article presents U2PNet, a novel unsupervised underwater image restoration network using polarization for improving signal-to-noise ratio and image quality in underwater imaging environments. Traditional methods for underwater image restoration using polarization require specific cues or pairs of underwater polarization datasets, which limit their practical applications. Our proposed method requires only one mosaicked polarized image of the scene and does not require datasets for pretraining or specific cues. We design two subnetworks (T-net and B ∞ -net) to accurately estimate the transmission map and background light, and unique nonreference loss functions to ensure effective restoration. Our experiments are based on an indoor polarization simulated dataset and a real polarization image dataset constructed from our underwater robotic platform equipped with polarization cameras. Experiment results demonstrate that our proposed method achieves state-of-the-art performance on both simulated and real underwater polarization images. The code and datasets will be available at https://github.com/polwork/U-2Pnet.

Predictive Classification System for Low Back Pain Based on Unsupervised Clustering.

STUDY DESIGN: Retrospective study. OBJECTIVE: Lumbar magnetic resonance imaging (MRI) findings are believed to be associated with low back pain (LBP). This study sought to develop a new predictive classification system for low back pain. METHOD: Normal subjects with repeated lumbar MRI scans were retrospectively enrolled. A new classification system, based on the radiological features on MRI, was developed using an unsupervised clustering method. RESULTS: One hundred and fifty-nine subjects were included. Three distinguishable clusters were identified with unsupervised clustering that were significantly correlated with LBP (P = .017). The incidence of LBP was highest in cluster 3 (57.14%), nearly twice the incidence in cluster 1 (30.11%). There were obvious differences in the sagittal parameters among the 3 clusters. Cluster 3 had the smallest intervertebral height. Based on follow-up findings, 27% of subjects changed clusters. More subjects changed from cluster 1 to clusters 2 or 3 (14.5%) than changed from cluster 2 or cluster 3 to cluster 1 (5%). Participation in sport was more frequent in subjects who changed from cluster 3 to cluster 1. CONCLUSION: Using an unsupervised clustering method, we developed a new classification system comprising 3 clusters, which were significantly correlated with LBP. The prediction of LBP is independent of age and better than that based on individual sagittal parameters derived from MRI. A change in cluster during follow-up may partially predict lumbar degeneration. This study provides a new system for the prediction of LBP that should be useful for its diagnosis and treatment.

Image-matching enhancement using a polarized intensity-hue-saturation fusion method.

Atmospheric absorption and scattering (e.g., haze) cause degradation in the image quality of outdoor scenes, which affects the image-matching process. The scale-invariant feature transform (SIFT) algorithm is not effective in haze. Edge information is required to enhance the matching process. Utilizing the polarization information expressed by the Stokes vector component S1 with its edge information can improve the keypoint localization in the matching process. In this paper, a novel, to the best of our knowledge, fusion method called polarized intensity-hue-saturation is proposed that uses polarization and depth information by fusion of a polarized haze-removed image with the estimated depth and by applying S1. The instant dehazing method uses polarized images to obtain a haze-removed image and its estimated depth map. The fused image has high spatial details required for enhancing the matching process. The experimental results show that the proposed method outperforms the existing image-matching schemes and improves the conventional SIFT matching method.

Deep learning for digitizing highly noisy paper-based ECG records.

Electrocardiography (ECG) is essential in many heart diseases. However, some ECGs are recorded by paper, which can be highly noisy. Digitizing the paper-based ECG records into a high-quality signal is critical for further analysis. We formulated the digitization problem as a segmentation problem and proposed a deep learning method to digitize highly noisy ECG scans. Our method extracts the ECG signal in an end-to-end manner and can handle different paper record layouts. In the experiment, our model clearly extracted the ECG waveform with a Dice coefficient of 0.85 and accurately measured the common ECG parameters with more than 0.90 Pearson's correlation. We showed that the end-to-end approach with deep learning can be powerful in ECG digitization. To the best of our knowledge, we provide the first approach to digitize the least informative noisy binary ECG scans and potentially be generalized to digitize various ECG records.

Detect differentially methylated regions using non-homogeneous hidden Markov model for methylation array data.

MOTIVATION: DNA methylation is an important epigenetic mechanism in gene regulation and the detection of differentially methylated regions (DMRs) is enthralling for many disease studies. There are several aspects that we can improve over existing DMR detection methods: (i) methylation statuses of nearby CpG sites are highly correlated, but this fact has seldom been modelled rigorously due to the uneven spacing; (ii) it is practically important to be able to handle both paired and unpaired samples; and (iii) the capability to detect DMRs from a single pair of samples is demanded. RESULTS: We present DMRMark (DMR detection based on non-homogeneous hidden Markov model), a novel Bayesian framework for detecting DMRs from methylation array data. It combines the constrained Gaussian mixture model that incorporates the biological knowledge with the non-homogeneous hidden Markov model that models spatial correlation. Unlike existing methods, our DMR detection is achieved without predefined boundaries or decision windows. Furthermore, our method can detect DMRs from a single pair of samples and can also incorporate unpaired samples. Both simulation studies and real datasets from The Cancer Genome Atlas showed the significant improvement of DMRMark over other methods. AVAILABILITY AND IMPLEMENTATION: DMRMark is freely available as an R package at the CRAN R package repository. CONTACT: xfan@cuhk.edu.hk. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

A full bayesian approach for boolean genetic network inference.

Boolean networks are a simple but efficient model for describing gene regulatory systems. A number of algorithms have been proposed to infer Boolean networks. However, these methods do not take full consideration of the effects of noise and model uncertainty. In this paper, we propose a full Bayesian approach to infer Boolean genetic networks. Markov chain Monte Carlo algorithms are used to obtain the posterior samples of both the network structure and the related parameters. In addition to regular link addition and removal moves, which can guarantee the irreducibility of the Markov chain for traversing the whole network space, carefully constructed mixture proposals are used to improve the Markov chain Monte Carlo convergence. Both simulations and a real application on cell-cycle data show that our method is more powerful than existing methods for the inference of both the topology and logic relations of the Boolean network from observed data.