The rapid development of machine learning (ML) techniques has opened up the data-dense field of microbiome research for novel therapeutic, diagnostic, and prognostic applications targeting a wide range of disorders, which could substantially improve healthcare practices in the era of precision medicine. However, several challenges must be addressed to exploit the benefits of ML in this field fully. In particular, there is a need to establish “gold standard” protocols for conducting ML analysis experiments and improve interactions between microbiome researchers and ML experts. The Machine Learning Techniques in Human Microbiome Studies (ML4Microbiome) COST Action CA18131 is a European network established in 2019 to promote collaboration between discovery-oriented microbiome researchers and data-driven ML experts to optimize and standardize ML approaches for microbiome analysis. This perspective paper presents the key achievements of ML4Microbiome, which include identifying predictive and discriminatory ‘omics’ features, improving repeatability and comparability, developing automation procedures, and defining priority areas for the novel development of ML methods targeting the microbiome. The insights gained from ML4Microbiome will help to maximize the potential of ML in microbiome research and pave the way for new and improved healthcare practices.

To avoid motion artefacts when merging multiple exposures into an HDR image, a number of deghosting algorithms have been proposed. These algorithms, however, do not work equally well on all types of scenes, and some may even introduce additional artefacts. Even though subjective methods of evaluation provide reliable means of testing, they need to be repeated for each new proposed method or even its slight modification and are cumbersome to perform. In this work, we evaluate several computational approaches of quantitative evaluation of multi-exposure HDR deghosting algorithms and demonstrate their results on five state- of-the-art algorithms. The quality of HDR images produced by deghosting methods is measured in a subjective experiment, and then evaluated using five objective metrics. The most reliable metric is then selected by testing correlation between subjective and objective metric scores.

Various High Dynamic Range (HDR) deghosting algorithms have been developed to solve the problem of merging dynamic content in multi-exposure HDR imaging. Even though these algorithms may be successful in `ghost' removal, they may fail to reduce noise in the resultant HDR image. As a result, the presence of noise in the generated HDR image degrades the overall image quality. HDR deghosting algorithms should also aim to reconstruct values that are approximately proportional to the luminance of the real scene. In this work we evaluate noise and luminance reconstruction in HDR images generated by five state-of-the-art HDR deghosting algorithms. The observations based on the obtained results are instrumental to guide the development of new HDR deghosting algorithms that will also aim to reduce noise and reconstruct original scene luminance to produce a good quality deghosted HDR image.

Abstract In this work we analyze the results produced by high dynamic range (HDR) multi-exposure merging algorithms that handle motion artefacts. Within this analysis, we introduce the criteria for the evaluation of these algorithms. In order to perform a comprehensive evaluation we propose a dataset categorized into different types of scenes,each posing a challenge for the evaluated algorithm. For the analysis, we select and customize a tone-mapping operator (TMO), which minimizes a chance for TMO artefacts and allows us to inspect resulting HDR images on a conventional sRGB display. We show the results of five state-of-the-art algorithms evaluated based on the proposed criteria by performing expert evaluation analysis and draw a conclusion for each of these algorithms.