A diffusion model multi-scale feature fusion network for imbalanced medical image classification research.

BACKGROUND AND OBJECTIVE: Medicine image classification are important methods of traditional medical image analysis, but the trainable data in medical image classification is highly imbalanced and the accuracy of medical image classification models is low. In view of the above two common problems in medical image classification. This study aims to: (i) effectively solve the problem of poor training effect caused by the imbalance of class imbalanced data sets. (ii) propose a network framework suitable for improving medical image classification results, which needs to be superior to existing methods. METHODS: In this paper, we put in the diffusion model multi-scale feature fusion network (DMSFF), which mainly uses the diffusion generation model to overcome imbalanced classes (DMOIC) on highly imbalanced medical image datasets. At the same time, it is processed according to the cropped image augmentation strategy through cropping (IASTC). Based on this, we use the new dataset to design a multi-scale feature fusion network (MSFF) that can fully utilize multiple hierarchical features. The DMSFF network can effectively solve the problems of small and imbalanced samples and low accuracy in medical image classification. RESULTS: We evaluated the performance of the DMSFF network on highly imbalanced medical image classification datasets APTOS2019 and ISIC2018. Compared with other classification models, our proposed DMSFF network achieved significant improvements in classification accuracy and F1 score on two datasets, reaching 0.872, 0.731, and 0.906, 0.836, respectively. CONCLUSIONS: Our newly proposed DMSFF architecture outperforms existing methods on two datasets, and verifies the effectiveness of generative model inverse balance for imbalance class datasets and feature enhancement by multi-scale feature fusion. Further, the method can be applied to other class imbalanced data sets where the results will be improved.

Performance of the highly sensitive humidity sensor constructed with nanofibrillated cellulose/graphene oxide/polydimethylsiloxane aerogel via freeze drying.

A kind of capacitive humidity sensor with high sensitivity constructed with nanofibrillated cellulose (NFC), graphene oxide (GO) and polydimethylsiloxane (PDMS) is presented in this work, via a simple ultrasonic dispersion and freeze drying technology. The NFC and GO with a strong adsorption for water molecules were used as a substrate for the promotion of capacitive response of the humidity sensor. Moreover, anhydrous ethanol was added to inhibit the generation of big cracks in the humidity sensor in the freeze drying process, so as to obtain a regular network porous structure, then providing a great deal of conduction channels and active sites for molecular water. Also, the addition of PDMS can effectively enhance the flexibility and stability of its porous structure. The results confirmed that the humidity sensor with 30 wt% GO showed an excellent humidity sensitivity (6576.41 pF/% RH), remarkable reproducibility, low humidity hysteresis characteristic in 11-97% relative humidity (RH) at 25 °C, and short response/recovery times (57 s/2 s). In addition, the presented sensor exhibited small relative deviation of the measured relative humidity value compared with the commercial hygrometer. The realization of the high sensitivity can be attributed to the theories about interaction of the hydrophilic group, proton transfer of water molecules and the three-dimensional network transport structure model. Therefore, the NFC/GO/PDMS humidity sensor finally realizes stable, reproducible and fast humidity sensing via an eco-friendly process, exhibiting promising potential for wide practical application.