Antiepileptic drug concentration detection based on Raman spectroscopy and an improved snake optimization-convolutional neural network algorithm.

A method for measurement of antiepileptic drug concentrations based on Raman spectroscopy and an optimization algorithm for mathematical models are proposed and investigated. This study uses Raman spectroscopy to measure mixed antiepileptic drugs, and an Improved Snake Optimization (ISO)-Convolutional Neural Network (CNN) algorithm is proposed. Raman spectroscopy is widely used in the identification of pharmaceutical ingredients due to its sharp peaks, no pre-treatment of samples and non-destructive detection. To analyze the spectral data precisely, a machine learning method is used in this paper. The ISO algorithm is an improved intelligent swarm algorithm in which the method of generating random solutions is improved, which can ensure that a comprehensive local search of the model is performed, the global search capability is maintained at a later stage, and the convergence speed is accelerated. In this study, 360 groups of oxcarbazepine, carbamazepine, and lamotrigine drug mixtures are measured using Raman spectroscopy, and the raw spectral data after pre-processing are trained and evaluated using ISO-CNN algorithms, and the results are compared and analyzed with those obtained from other algorithms such as the Northern Goshawk Optimization algorithm, Chameleon Swarm Algorithm, and White Shark Optimizer algorithm. The results show that the best ISO-CNN algorithm training is achieved for oxcarbazepine, with a determination coefficient and root mean square error of 0.99378 and 0.0295 for the validation set, and 0.99627 and 0.0278 for the test set. The overall results suggest that Raman spectroscopy combined with machine learning algorithms can be a potential tool for drug concentration prediction.

Active non-uniform illumination-based underwater polarization imaging method for objects with complex polarization properties.

Active polarization imaging is one of the most effective underwater optical imaging methods that can eliminate the degradation of image contrast and clarity caused by macro-molecule scattering. However, the non-uniformity of active illumination and the diversity of object polarization properties may decrease the quality of underwater imaging. This paper proposes a non-uniform illumination-based active polarization imaging method for underwater objects with complex optical properties. Firstly, illumination homogenization in the frequency domain is proposed to extract and homogenize the natural incident light from the total receiving light. Then, the weight values of the polarized and non-polarized images are computed according to each pixel's degree of linear polarization (DoLP) in the original underwater image. By this means, the two images can be fused to overcome the problem of reflected light loss generated by the complex polarization properties of underwater objects. Finally, the fusion image is normalized as the final result of the proposed underwater polarization imaging method. Both qualitative and quantitative experimental results show that the presented method can effectively eliminate the uneven brightness of the whole image and obtain the underwater fusion image with significantly improved contrast and clarity. In addition, the ablation experiment of different operation combinations shows that each component of the proposed method has noticeable enhancement effects on underwater polarization imaging. Our codes are available in Code 1.

Unsupervised end-to-end infrared and visible image fusion network using learnable fusion strategy.

Infrared and visible image fusion aims to reconstruct fused images with comprehensive visual information by merging the complementary features of source images captured by different imaging sensors. This technology has been widely used in civil and military fields, such as urban security monitoring, remote sensing measurement, and battlefield reconnaissance. However, the existing methods still suffer from the preset fusion strategies that cannot be adjustable to different fusion demands and the loss of information during the feature propagation process, thereby leading to the poor generalization ability and limited fusion performance. Therefore, we propose an unsupervised end-to-end network with learnable fusion strategy for infrared and visible image fusion in this paper. The presented network mainly consists of three parts, including the feature extraction module, the fusion strategy module, and the image reconstruction module. First, in order to preserve more information during the process of feature propagation, dense connections and residual connections are applied to the feature extraction module and the image reconstruction module, respectively. Second, a new convolutional neural network is designed to adaptively learn the fusion strategy, which is able to enhance the generalization ability of our algorithm. Third, due to the lack of ground truth in fusion tasks, a loss function that consists of saliency loss and detail loss is exploited to guide the training direction and balance the retention of different types of information. Finally, the experimental results verify that the proposed algorithm delivers competitive performance when compared with several state-of-the-art algorithms in terms of both subjective and objective evaluations. Our codes are available at https://github.com/MinjieWan/Unsupervised-end-to-end-infrared-and-visible-image-fusion-network-using-learnable-fusion-strategy.