Transfer learning in deep neural network based under-sampled MR image reconstruction

In Magnetic Resonance Imaging (MRI), the success of deep learning-based under-sampled MR image reconstruction depends on: (i) size of the training dataset, (ii) generalization capabilities of the trained neural network. Whenever there is a mismatch between the training and testing data, there is a need to retrain the neural network from scratch with thousands of MR images obtained using the same protocol. This may not be possible in MRI as it is costly and time consuming to acquire data. In this research, a transfer learning approach i.e. end-to-end fine tuning is proposed for U-Net to address the data scarcity and generalization problems of deep learning-based MR image reconstruction. First the generalization capabilities of a pre-trained U-Net (initially trained on the human brain images of 1.5 T scanner) are assessed for: (a) MR images acquired from MRI scanners of different magnetic field strengths, (b) MR images of different anatomies and (c) MR images under-sampled by different acceleration factors. Later, end-to-end fine tuning of the pre-trained U-Net is proposed for the reconstruction of the above-mentioned MR images (i.e. (a), (b) and (c)). The results show successful reconstructions obtained from the proposed method as reflected by the Structural SIMilarity index, Root Mean Square Error, Peak Signal-to-Noise Ratio and central line profile of the reconstructed images.     

Circular regular dessins

Journal of Algebraic Combinatorics - A dessin is called circular if the boundary of each face is a circuit (a simple cycle). In this paper, we address the problem of characterizing edge-transitive...     

Anti-Staphylococcus aureus evaluation of gallic acid by isothermal microcalorimetry and principle component analysis

Journal of Thermal Analysis and Calorimetry - The emergence of antibiotic-resistant Staphylococcus aureus requires new antibacterial drugs. In this study, using a thermal activity monitor air...     

A New Method of Testing and Evaluating the Quality of Refined Montan Wax: Digital Color and GC Fingerprint

To establish a new method of testing and evaluating the quality of refined montan wax (RMW), digital color and GC fingerprint technology were introduced and applied. CIE Lab color mode was used to digitize the exterior colors of RMW, and the score obtained through a fitting function was also used to reflect its quality. It is shown that they were in complete accord with the human visual perception trend. The GC fingerprint was used to characterize the internal chemical information of RMW, and the composition of its internal features was reflected through the relative retention time (RRT) and relative peak area (RPA) values. It is shown that there was a high degree of similarity between the fingerprints, while certain differences also existed. This can be used to implement effective application of RMW to aspects such as quality control, adulteration identification, and origin attributions.

     

Fabrication and characterization of cellulose nanoparticles from maize stalk pith via ultrasonic-mediated cationic etherification

In this study, parenchyma cellulose, which was extracted from maize stalk pith as an abundant source of agricultural residues, was applied for preparing cellulose nanoparticles (CNPs) via an ultrasound-assisted etherification and a subsequent sonication process. The ultrasonic-assisted treatment greatly improved the modification of the pith cellulose with glycidyltrimethylammonium chloride, leading to a partial increase in the dissolubility of the as-obtained product and thus disintegration of sheet-like cellulose into nanoparticles. While the formation of CNPs by ultrasonication was largely dependent on the cellulose consistency in the cationic-modified system. Under the condition of 25% cellulose consistency, the longer sono-treated duration yielded a more stable and dispersible suspension of CNP due to its higher zeta potential. Degree of substitution and FT-IR analyses indicated that quaternary ammonium salts were grafted onto hydroxyl groups of cellulose chain. SEM and TEM images exhibited the CNP to have spherical morphology with an average dimeter from 15 to 55 nm. XRD investigation revealed that CNPs consisted mainly of a crystalline cellulose Ι structure, and they had a lower crystallinity than the starting cellulose. Moreover, thermogravimetric results illustrated the thermal resistance of the CNPs was lower than the pith cellulose. The optimal CNP with highly cationic charges, good stability and acceptable thermostability might be considered as one of the alternatively renewable reinforcement additives for nanocomposite production.     

Salen–Mg-doped NH2–MIL-101(Cr) for effective CO2 adsorption under ambient conditions

A novel metal-doped metal–organic framework (MOF) was developed by incorporating salen–Mg into NH₂–MIL-101(Cr) structure under ambient conditions. The Schiff base complex was successfully prepared by condensing salicylaldehyde with a free amino group and then coordinating metal ions. Such a structure can endow the sample with higher CO₂ adsorption performance. At 0°C and 1 bar, the salen–Mg-modified sample achieves the maximum adsorption capacity of 2.18 mmol g⁻¹ for CO₂, which was 5.8% higher than the pristine salen–MOF under the same conditions. Notably, the Freundlich model indicates that the CO₂ adsorption process of all samples conforms to reversible adsorption. However, the correlation coefficients (R²) of the Mg-doped sample are lower than that of the pristine sample. Besides, the CO₂/N₂ adsorption selectivity and isosteric heat also show a similar trend. These results indicate that the salen–Mg can enhance the interaction between the material and CO₂ molecules.     

Cooperative effect of Fe and Ti species over Fe–Ti–Ox catalysts on the catalytic hydrolysis performance of hydrogen cyanide

FeO_x, TiO₂, and Fe–Ti–O_x catalysts were synthesized and used in the catalytic hydrolysis of hydrogen cyanide (HCN). Nearly 100% HCN conversion was achieved at 250 °C over the Fe–Ti–O_x catalyst. TiO₂ rutile was detected over TiO₂, but not over Fe–Ti–O_x, which suggested that the interaction between Fe and Ti species could inhibit the TiO₂ phase transition. Furthermore, the interaction between Fe and Ti species over Fe–Ti–O_x could promote the selectivity of NH₃ and CO. The mechanism of hydrolysis of HCN over FeO_x, TiO₂, and Fe–Ti–O_x can be given as follows: HCN + H₂O → methanamide → ammonium formate → formic acid → H₂O + CO.     

Hölder metric subregularity for constraint systems in Asplund spaces

Positivity - This paper deals with the Hölder metric subregularity property of a certain constraint system in Asplund space. Using the techniques of variational analysis, its main part is...     

Well-posedness and generalized metric subregularity with respect to an admissible function

Science China Mathematics - In the framework of complete metric spaces, this paper provides several sufficient conditions for the well-posedness with respect to an admissible function, which...     

柚皮素/β-环糊精超分子体系的包合行为

本文采用超声法制备了柚皮素（NAR）与β-环糊精（βCD）的包合物.粉末-X射线衍射（XRD）和红外吸收光谱（IR）测定均表明形成的包合物具有不同于主客体的新的结构性质.¹H NMR与ROESY核磁共振（NMR）实验表明NAR以苯环端从βCD的宽口端进入，并形成稳定的超分子包合物.量子化学计算分析NAR/βCD包合物的形成过程表明，驱动力源于焓驱动与氢键弱相互作用力；能隙和结合能分析得到的最优包合模式与NMR研究结果一致；ONIOM分层计算验证了上述结果.分子对接模拟出的最优包合模式也与量子化学计算、NMR的分析结果吻合.本文获取了清晰的NAR/βCD包合物构型及其形成机理，为该超分子药物的定量构效关系研究提供了理论参考.