No abstract

Several pieces of research during the last decade in intelligent perception are focused on the development of algorithms allowing vehicles to move efficiently in complex environments. Most of existing approaches suffer from either processing time which do not meet real-time requirements, or inefficient in real complex environment, which also doesn't meet the full availability constraint of such a critical function. To improve the existing solutions, an algorithm based on curved lane detection by using a Bayesian framework for the estimation of multi-hyperbola parameters is proposed to detect curved lane under challenging conditions. The general idea is to divide a captured image into several parts. The trajectory is modeled by a hyperbola over each part, whose parameters are estimated using the proposed hierarchical Bayesian model. Compared to the existing works in the state of the art, experimental results prove that our approach is more efficient and more precise in road marking detection. Keywords Autonomous driving • embedded camera • road marking • multi-hyperbola • Bayesian framework

Aims: Prognosis of lung mathology severity after Covid-19 infection using chest X-ray time series Background: We have been inspired by methods analysing time series of images in remote sensing for change detection. During the current Covid-19 pandemic, our motivation is to provide an automatic tool to predict severity of lung pathologies due to Covid-19. This can be done by analysing images of the same patient acquired at different dates. Since no analytical model is available, and also no accurate quantification tools can be used due to many unknowns about the pathology, feature-free methods are good candidates to analyse such temporal images. Objective: This contribution helps improving performances of medical structures facing the Covid-19 pandemic. The first impact is medical and social since more lives could be saved with a 92% rate of good prognosis. In addition to that, patients in intensive care units (up to 15%) could a posteriori suffer from less sequels due to an early and accurate prognosis of their PP. Moreover, accurate prognosis can lead to a better planning of patient’s transfer between units and hospitals, which is linked to the second claimed economical impact. Indeed, prognosis is linked to lower treatment costs due to an optimized predictive protocol using ragiological prognosis. Methods: Using Convolutional Neural Networks (CNN) in combination with Recurrent Neural Networks (RNN). Spatial and temporal features are combines to analyse image time series. A prognosis score is delivered indicating the severity of the pathology. Learning is made on a publicly available database. Results: When applied to radiological time-series, promising results are obtained with an accuracy rates higher than 92%. Sensitivity and specificity rates are also very interesting. Conclusion: Our method is segmentation-free, which makes it competitive with respect to other assessment methods relying on time-consuming lung segmentation algorithms. When applied on radiographic data, the proposed ProgNet architecture showed promising results with good classification performances, especially for ambiguous cases. Specifically, the reported low false positive rates are interesting for an accurate and personalised care workflow.

Humanity is facing nowadays a dramatic pandemic episode with the Coronavirus propagation over all continents. The Covid-19 disease is still not well characterized, and many research teams all over the world are working on either therapeutic or vaccination issues. Massive testing is one of the main recommendations. In addition to laboratory tests, imagerybased tools are being widely investigated. Artificial intelligence is therefore contributing to the efforts made to face this pandemic phase. Regarding patients in hospitals, it is important to monitor the evolution of lung pathologies due to the virus. A prognosis is therefore of great interest for doctors to adapt their care strategy. In this paper, we propose a method for Covid-19 prognosis based on deep learning architectures. The proposed method is based on the combination of a convolutional and recurrent neural networks to classify multi-temporal chest X-ray images and predict the evolution of the observed lung pathology. When applied to radiological time-series, promising results are obtained with an accuracy rates higher than 92%.All rights reserved. No reuse allowed without permission.was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

No abstract