DM-CNN: Dynamic Multi-scale Convolutional Neural Network with uncertainty quantification for medical image classification.

Convolutional neural network (CNN) has promoted the development of diagnosis technology of medical images. However, the performance of CNN is limited by insufficient feature information and inaccurate attention weight. Previous works have improved the accuracy and speed of CNN but ignored the uncertainty of the prediction, that is to say, uncertainty of CNN has not received enough attention. Therefore, it is still a great challenge for extracting effective features and uncertainty quantification of medical deep learning models In order to solve the above problems, this paper proposes a novel convolutional neural network model named DM-CNN, which mainly contains the four proposed sub-modules : dynamic multi-scale feature fusion module (DMFF), hierarchical dynamic uncertainty quantifies attention (HDUQ-Attention) and multi-scale fusion pooling method (MF Pooling) and multi-objective loss (MO loss). DMFF select different convolution kernels according to the feature maps at different levels, extract different-scale feature information, and make the feature information of each layer have stronger representation ability for information fusion HDUQ-Attention includes a tuning block that adjust the attention weight according to the different information of each layer, and a Monte-Carlo (MC) dropout structure for quantifying uncertainty MF Pooling is a pooling method designed for multi-scale models, which can speed up the calculation and prevent overfitting while retaining the main important information Because the number of parameters in the backbone part of DM-CNN is different from other modules, MO loss is proposed, which has a fast optimization speed and good classification effect DM-CNN conducts experiments on publicly available datasets in four areas of medicine (Dermatology, Histopathology, Respirology, Ophthalmology), achieving state-of-the-art classification performance on all datasets. DM-CNN can not only maintain excellent performance, but also solve the problem of quantification of uncertainty, which is a very important task for the medical field. The code is available: https://github.com/QIANXIN22/DM-CNN.

HWA-SegNet: Multi-channel skin lesion image segmentation network with hierarchical analysis and weight adjustment.

Convolutional neural networks (CNNs) show excellent performance in accurate medical image segmentation. However, the characteristics of sample with small size and insufficient feature expression, irregular shape of the segmented target and inaccurate judgment of edge texture have always been problems to be faced in the field of skin lesion image segmentation. Therefore, in order to solve these problems, discrete Fourier transform (DFT) is introduced to enrich the input data and a CNN architecture (HWA-SegNet) is proposed in this paper. Firstly, DFT is improved to analyze the features of the skin lesions image, and multi-channel data is extended for each image. Secondly, a hierarchical dilated analysis module is constructed to understand the semantic features under multi-channel. Finally, the pre-prediction results are fine-tuned using a weight adjustment structure with fully connected layers to obtain higher accuracy prediction results. Then, 520 skin lesion images are tested on the ISIC 2018 dataset. Extensive experimental results show that our HWA-SegNet improve the average segmentation Dice Similarity Coefficient from 88.30% to 91.88%, Sensitivity from 89.29% to 92.99%, and Jaccard similarity index from 81.15% to 85.90% compared with U-Net. Compared with the State-of-the-Art method, the Jaccard similarity index and Specificity are close, but the Dice Similarity Coefficient is higher. The experimental data show that the data augmentation strategy based on improved DFT and HWA-SegNet are effective for skin lesion image segmentation.

ILCS: An Improved Lightweight Convolution Structure and Mixed Interactive Attention for Steel Surface Defect Classification.

The classification method of steel surface defects based on deep learning provides a basis for quality control of industrial steel manufacturing. Due to a large number of interference in the steel production area and the limited computing resources of the edge equipment deployed in the production area, it is a challenge to develop a lightweight model to achieve rapid and accurate classification in the case of limited computing resources. In this article, an improved lightweight convolution structure (LCS) is proposed, which combines the separable structure of convolution and introduces depth convolution and point direction convolution instead of the traditional convolutional module, so as to realize the lightweight of the model. In order to ensure the classification accuracy, spatial attention and channel attention are combined to compensate for the accuracy loss after deep convolution and point direction convolution respectively. Further, in order to improve the classification accuracy, a mixed interactive attention module (MIAM) is proposed to enhance the extracted feature information after LCS. The experimental results show that the recognition accuracy of our method exceeds that of the traditional model, and the number of parameters and the amount of calculation are greatly reduced, which realizes the lightweight of the steel surface defect classification model.

Convolutional Neural Network with Multiscale Fusion and Attention Mechanism for Skin Diseases Assisted Diagnosis.

Melanoma segmentation based on a convolutional neural network (CNN) has recently attracted extensive attention. However, the features captured by CNN are always local that result in discontinuous feature extraction. To solve this problem, we propose a novel multiscale feature fusion network (MSFA-Net). MSFA-Net can extract feature information at different scales through a multiscale feature fusion structure (MSF) in the network and then calibrate and restore the extracted information to achieve the purpose of melanoma segmentation. Specifically, based on the popular encoder-decoder structure, we designed three functional modules, namely MSF, asymmetric skip connection structure (ASCS), and calibration decoder (Decoder). In addition, a weighted cross-entropy loss and two-stage learning rate optimization strategy are designed to train the network more effectively. Compared qualitatively and quantitatively with the representative neural network methods with encoder-decoder structure, such as U-Net, the proposed method can achieve advanced performance.

Enhancement of oral bioavailability of cyclosporine A: comparison of various nanoscale drug-delivery systems.

A variety of nanoscale delivery systems have been shown to enhance the oral absorption of poorly water-soluble and poorly permeable drugs. However, the performance of these systems has seldom been evaluated simultaneously. The aim of this study was to compare the bioavailability enhancement effect of lipid-based nanocarriers with poly(lactic-co-glycolic acid) (PLGA) nanoparticles (NPs) to highlight the importance of the lipid composition, with cyclosporine A (CyA) as a model drug. CyA-loaded PLGA NPs, nanostructured lipid carriers (NLCs), and self-microemulsifying drug-delivery systems (SMEDDS) were prepared. The particle size of PLGA NPs (182.2 ± 12.8 nm) was larger than that of NLCs (89.7 ± 9.0 nm) and SMEDDS (26.9 ± 1.9 nm). All vehicles are charged negatively. The entrapment efficiency of PLGA NPs and NLCs was 87.6%± 1.6% and 80.3%± 0.6%, respectively. In vitro release tests indicated that the cumulative release of CyA was lower than 4% from all vehicles, including Sandimmun Neoral(®), according to the dialysis method. Both NLCs and SMEDDS showed high relative oral bioavailability, 111.8% and 73.6%, respectively, after oral gavage administration to beagle dogs, which was not statistically different from commercial Sandimmun Neoral(®). However, PLGA NPs failed to achieve efficient absorption, with relative bioavailability of about 22.7%. It is concluded that lipid-based nanoscale drug-delivery systems are superior to polymeric NPs in enhancing oral bioavailability of poorly water-soluble and poorly permeable drugs.

The role of lipid-based nano delivery systems on oral bioavailability enhancement of fenofibrate, a BCS II drug: comparison with fast-release formulations.

The aim of this study was to compare various formulations solid dispersion pellets (SDP), nanostructured lipid carriers (NLCs) and a self-microemulsifying drug delivery system (SMEDDS) generally accepted to be the most efficient drug delivery systems for BCS II drugs using fenofibrate (FNB) as a model drug. The size and morphology of NLCs and SMEDDS was characterized by dynamic light scattering (DLS) and transmission electron microscopy (TEM). Their release behaviors were investigated in medium with or without pancreatic lipase. The oral bioavailability of the various formulations was compared in beagle dogs using commercial Lipanthyl® capsules (micronized formulation) as a reference. The release of FNB from SDP was much faster than that from NLCs and SMEDDS in medium without lipase, whereas the release rate from NLCs and SMEDDS was increased after adding pancreatic lipase into the release medium. However, NLCs and SMEDDS increased the bioavailability of FNB to 705.11% and 809.10%, respectively, in comparison with Lipanthyl® capsules, although the relative bioavailability of FNB was only 366.05% after administration of SDPs. Thus, lipid-based drug delivery systems (such as NLCs and SMEDDS) may have more advantages than immediate release systems (such as SDPs and Lipanthyl® capsules).