No abstract

No abstract

Background and Aim The majority of patients with eosinophilic esophagitis (EoE) are likely to have observable features under narrow‐band imaging, namely beige mucosa. However, the histological features and clinical implications of beige mucosa have not been investigated. The aim of this study was to determine whether beige mucosa could serve as an endoscopic marker for predicting active inflammatory sites of EoE. Methods We retrospectively analyzed both the narrow‐band images and biopsied specimens of 77 esophageal lesions from 35 consecutive patients with EoE. We divided these specimens into two groups: target biopsied specimens from beige mucosa (beige group) and specimens biopsied from non‐beige mucosa (non‐beige group). The number of eosinophils per high‐powered field, thickness of the superficial differentiated cell layer, and depth of the hemoglobin component from the surface layer were compared between the two groups. Results Forty‐four out of the 45 specimens were diagnosed as histological active lesions in the beige group. The sensitivity, specificity, and overall accuracy of beige mucosa in predicting EoE activity were 97.8%, 96.9%, and 97.8%, respectively. Compared with the non‐beige group, specimens in the beige group had a significantly thinner superficial differentiated cell layer. Conclusions Beige mucosa is associated with thinning of the normal superficial differentiated cell layer, and these histological changes in the active inflammatory sites of EoE could be recognized endoscopically as color differences. Beige mucosa may serve as an endoscopic indicator for predicting the histological activity of EoE.

Ant Colony Optimization (ACO) encompasses a family of metaheuristics inspired by the foraging behaviour of ants. Since the introduction of the first ACO algorithm, called Ant System (AS), several ACO variants have been proposed in the literature. Owing to their superior performance over other alternatives, the most popular ACO algorithms are Rank-based Ant System (ASRank), Max-Min Ant System (MMAS) and Ant Colony System (ACS). While ASRank shows a fast convergence to high-quality solutions, its performance is improved by other more widely used ACO variants such as MMAS and ACS, which are currently considered the state-of-the-art ACO algorithms for static combinatorial optimization problems. With the purpose of diversifying the search process and avoiding early convergence to a local optimal, the proposed approach extends ASRank with an originality reinforcement strategy of the top-ranked solutions and a pheromone smoothing mechanism that is triggered before the algorithm reaches stagnation. The approach is tested on several symmetric and asymmetric Traveling Salesman Problem and Sequential Ordering Problem instances from TSPLIB benchmark. Our experimental results show that the proposed method achieves fast convergence to high-quality solutions and outperforms the current state-of-the-art ACO algorithms ASRank, MMAS and ACS, for most instances of the benchmark.

This work addresses the IFS-based image reconstruction problem for binary images. Given a binary image as the input, the goal is to obtain all the parameters of an iterated function system whose attractor approximates the input image accurately; the quality of this approximation is measured according to a similarity function between the original and the reconstructed images. This paper introduces a new method to tackle this issue. The method is based on functional networks, a powerful extension of neural networks that uses functions instead of the scalar weights typically found in standard neural networks. The method relies on an artificial network comprised of several functional networks, one for each of the contractive affine maps forming the IFS. The method is applied to an illustrative and challenging example of a fractal binary image exhibiting a complicated shape. The graphical and numerical results show that the method performs very well and is able to reconstruct the input image using IFS with high accuracy. The results also show that the method is not yet optimal and offers room for further improvement.

No abstract

No abstract

The rapid development of computer science and telecommunications has brought new ways and practices to sport training. The artificial sport trainer, founded on computational intelligence algorithms, has gained momentum in the last years. However, artificial sport trainer usually suffers from a lack of automatisation in realization and control phases of the training. In this study, the Digital Twin is proposed as a framework for helping athletes, during realization of training sessions, to make the proper decisions in situations they encounter. The digital twin for artificial sport trainer is based on the cognitive model of humans. This concept has been applied to cycling, where a version of the system on a Raspberry Pi already exists. The results of porting the digital twin on the mentioned platform shows promising potential for its extension to other sport disciplines.