Clean mucosal area detection of gastroenterologists versus artificial intelligence in small bowel capsule endoscopy.

Studies comparing the detection of clean mucosal areas in capsule endoscopy (CE) using human judgment versus artificial intelligence (AI) are rare. This study statistically analyzed gastroenterologist judgments and AI results. Three hundred CE video clips (100 patients) were prepared. Five gastroenterologists classified the video clips into 3 groups (≥75% [high], 50%-75% [middle], and < 50% [low]) according to their subjective judgment of cleanliness. Visualization scores were calculated using an AI algorithm based on the predicted visible area, and the 5 gastroenterologists' judgments and AI results were compared. The 5 gastroenterologists evaluated CE clip video quality as "high" in 10.7% to 36.7% and as "low" in 28.7% to 60.3% and 29.7% of cases, respectively. The AI evaluated CE clip video quality as "high" in 27.7% and as "low" in 29.7% of cases. Repeated-measures analysis of variance (ANOVA) revealed significant differences in the 6 evaluation indicators (5 gastroenterologists and 1 AI) (P < .001). Among the 300 judgments, 90 (30%) were consistent with 5 gastroenterologists' judgments, and 82 (91.1%) agreed with the AI judgments. The "high" and "low" judgments of the gastroenterologists and AI agreed in 95.0% and 94.9% of cases, respectively. Bonferroni's multiple comparison test showed no significant difference between 3 gastroenterologists and AI (P = .0961, P = 1.0000, and P = .0676, respectively) but a significant difference between the other 2 with AI (P < .0001). When evaluating CE images for cleanliness, the judgments of 5 gastroenterologists were relatively diverse. The AI produced a relatively universal judgment that was consistent with the gastroenterologists' judgements.

N-Step Pre-Training and Décalcomanie Data Augmentation for Micro-Expression Recognition.

Facial expressions are divided into micro- and macro-expressions. Micro-expressions are low-intensity emotions presented for a short moment of about 0.25 s, whereas macro-expressions last up to 4 s. To derive micro-expressions, participants are asked to suppress their emotions as much as possible while watching emotion-inducing videos. However, it is a challenging process, and the number of samples collected tends to be less than those of macro-expressions. Because training models with insufficient data may lead to decreased performance, this study proposes two ways to solve the problem of insufficient data for micro-expression training. The first method involves N-step pre-training, which performs multiple transfer learning from action recognition datasets to those in the facial domain. Second, we propose Décalcomanie data augmentation, which is based on facial symmetry, to create a composite image by cutting and pasting both faces around their center lines. The results show that the proposed methods can successfully overcome the data shortage problem and achieve high performance.

Semantic Segmentation Dataset for AI-Based Quantification of Clean Mucosa in Capsule Endoscopy.

Background and Objectives: Capsule endoscopy (CE) for bowel cleanliness evaluation primarily depends on subjective methods. To objectively evaluate bowel cleanliness, we focused on artificial intelligence (AI)-based assessments. We aimed to generate a large segmentation dataset from CE images and verify its quality using a convolutional neural network (CNN)-based algorithm. Materials and Methods: Images were extracted and divided into 10 stages according to the clean regions in a CE video. Each image was classified into three classes (clean, dark, and floats/bubbles) or two classes (clean and non-clean). Using this semantic segmentation dataset, a CNN training was performed with 169 videos, and a clean region (visualization scale (VS)) formula was developed. Then, measuring mean intersection over union (mIoU), Dice index, and clean mucosal predictions were performed. The VS performance was tested using 10 videos. Results: A total of 10,033 frames of the semantic segmentation dataset were constructed from 179 patients. The 3-class and 2-class semantic segmentation's testing performance was 0.7716 mIoU (range: 0.7031-0.8071), 0.8627 Dice index (range: 0.7846-0.8891), and 0.8927 mIoU (range: 0.8562-0.9330), 0.9457 Dice index (range: 0.9225-0.9654), respectively. In addition, the 3-class and 2-class clean mucosal prediction accuracy was 94.4% and 95.7%, respectively. The VS prediction performance for both 3-class and 2-class segmentation was almost identical to the ground truth. Conclusions: We established a semantic segmentation dataset spanning 10 stages uniformly from 179 patients. The prediction accuracy for clean mucosa was significantly high (above 94%). Our VS equation can approximately measure the region of clean mucosa. These results confirmed our dataset to be ideal for an accurate and quantitative assessment of AI-based bowel cleanliness.

Concentration Gradient Induced Delithiation Failure of MoO3 for Li-Ion Batteries.

Electric vehicle manufacturers worldwide are demanding superior lithium-ion batteries, with high energy and power densities, compared to gasoline engines. Although conversion-type metal oxides are promising candidates for high-capacity anodes, low initial Coulombic efficiency (ICE) and poor capacity retention have hindered research on their applications. In this study, the ICE of conversion-type MoO3 is investigated, with a particular focus on the delithiation failure. A computational modeling predicts the concentration gradient of Li+ in MoO3 particles. The highly delithiated outer region of the particle forms a layer with low electronic conductivity, which impedes further delithiation. A comparative study using various sizes of MoO3 particles demonstrated that the electrode failure during delithiation is governed by the concentration gradient and the subsequent formation of a resistive shell. The proposed failure mechanism provides critical guidance for the development of conversion-type anode materials with improved electrochemical reversibility.

SalfMix: A Novel Single Image-Based Data Augmentation Technique Using a Saliency Map.

Modern data augmentation strategies such as Cutout, Mixup, and CutMix, have achieved good performance in image recognition tasks. Particularly, the data augmentation approaches, such as Mixup and CutMix, that mix two images to generate a mixed training image, could generalize convolutional neural networks better than single image-based data augmentation approaches such as Cutout. We focus on the fact that the mixed image can improve generalization ability, and we wondered if it would be effective to apply it to a single image. Consequently, we propose a new data augmentation method to produce a self-mixed image based on a saliency map, called SalfMix. Furthermore, we combined SalfMix with state-of-the-art two images-based approaches, such as Mixup, SaliencyMix, and CutMix, to increase the performance, called HybridMix. The proposed SalfMix achieved better accuracies than Cutout, and HybridMix achieved state-of-the-art performance on three classification datasets: CIFAR-10, CIFAR-100, and TinyImageNet-200. Furthermore, HybridMix achieved the best accuracy in object detection tasks on the VOC dataset, in terms of mean average precision.

Interplay between electrochemical reactions and mechanical responses in silicon-graphite anodes and its impact on degradation.

Durability of high-energy throughput batteries is a prerequisite for electric vehicles to penetrate the market. Despite remarkable progresses in silicon anodes with high energy densities, rapid capacity fading of full cells with silicon-graphite anodes limits their use. In this work, we unveil degradation mechanisms such as Li+ crosstalk between silicon and graphite, consequent Li+ accumulation in silicon, and capacity depression of graphite due to silicon expansion. The active material properties, i.e. silicon particle size and graphite hardness, are then modified based on these results to reduce Li+ accumulation in silicon and the subsequent degradation of the active materials in the anode. Finally, the cycling performance is tailored by designing electrodes to regulate Li+ crosstalk. The resultant full cell with an areal capacity of 6 mAh cm-2 has a cycle life of >750 cycles the volumetric energy density of 800 Wh L-1 in a commercial cell format.

Metabolic and Lipidomic Profiling of Vegetable Juices Fermented with Various Probiotics.

Fermented vegetable juices have gained attention due to their various beneficial effects on human health. In this study, we employed gas chromatography-mass spectrometry, direct infusion-mass spectrometry, and liquid chromatography-mass spectrometry to identify useful metabolites, lipids, and carotenoids in vegetable juice (VJ) fermented with Lactobacillus plantarum HY7712, Lactobacillus plantarum HY7715, Lactobacillus helveticus HY7801, and Bifidobacterium animalis ssp. lactis HY8002. A total of 41 metabolites, 24 lipids, and 4 carotenoids were detected in the fermented and non-fermented VJ (control). The lycopene, α-carotene, and ß-carotene levels were higher in VJ fermented with L. plantarum strains (HY7712 and HY7715) than in the control. Proline content was also elevated in VJ fermented with HY7715. Uracil, succinic acid, and α-carotene concentration was increased in VJ fermented with HY7801, while glycine and lycopene levels were raised in VJ fermented with HY8002. This study confirmed that each probiotic strain has distinctive characteristics and produces unique changes to metabolic profiles of VJ during fermentation. Our results suggest that probiotic-fermented VJ is a promising functional beverage that contains more beneficial metabolites and carotenoids than commercial non-fermented VJ.

Nanoscale Electrical Degradation of Silicon-Carbon Composite Anode Materials for Lithium-Ion Batteries.

High-performance lithium-ion batteries (LIBs) are in increasing demand for a variety of applications in rapidly growing energy-related fields including electric vehicles. To develop high-performance LIBs, it is necessary to comprehensively understand the degradation mechanism of the LIB electrodes. From this viewpoint, it is crucial to investigate how the electrical properties of LIB electrodes change under charging and discharging. Here, we probe the local electrical properties of LIB electrodes with nanoscale resolution by scanning spreading resistance microscopy (SSRM). Via quantitative and comparative SSRM measurements on pristine and degraded LIB anodes of Si-C composites blended with graphite (Gr) particles, the electrical degradation of the LIB anodes is visualized. The electrical conductivity of the Si-C composite particles considerably degraded over 300 cycles of charging and discharging, whereas the Gr particles maintained their conductivity.

Sn-Based Nanocomposite for Li-Ion Battery Anode with High Energy Density, Rate Capability, and Reversibility.

To design an easily manufactured, large energy density, highly reversible, and fast rate-capable Li-ion battery (LIB) anode, Co-Sn intermetallics (CoSn2, CoSn, and Co3Sn2) were synthesized, and their potential as anode materials for LIBs was investigated. Based on their electrochemical performances, CoSn2 was selected, and its C-modified nanocomposite (CoSn2/C) as well as Ti- and C-modified nanocomposite (CoSn2/ a-TiC/C) was straightforwardly prepared. Interestingly, the CoSn2, CoSn2/C, and CoSn2/ a-TiC/C showed conversion/nonrecombination, conversion/partial recombination, and conversion/full recombination during Li insertion/extraction, respectively, which were thoroughly investigated using ex situ X-ray diffraction and extended X-ray absorption fine structure analyses. As a result of the interesting conversion/full recombination mechanism, the easily manufactured CoSn2/ a-TiC/C nanocomposite for the Sn-based Li-ion battery anode showed large energy density (first reversible capacity of 1399 mAh cm-3), high reversibility (first Coulombic efficiency of 83.2%), long cycling behavior (100% capacity retention after 180 cycles), and fast rate capability (appoximately 1110 mAh cm-3 at 3 C rate). In addition, degradation/enhancement mechanisms for high-capacity and high-performance Li-alloy-based anode materials for next-generation LIBs were also suggested.

Protective Oxide Coating for Ionic Conductive Solid Electrolyte Interphase.

To employ Li-based batteries to their full potential in a wide range of energy-storage applications, their capacity and performance stability must be improved. Si is a viable anode material for Li-based batteries in electric vehicles due to its high theoretical capacity and good economic feasibility. However, it suffers from physical and chemical degradation, leading to unstable electrochemical performance and preventing its incorporation in new Li-based battery systems. Herein, we applied a poly(vinyl alcohol)-PO4 protective coating for Si-graphite anodes and confirmed an improvement in the electrochemical performance. The experimental results revealed that the polymer acts as a binder to alleviate the pulverization of the electrode. Furthermore, the oxide coating reduces the loss of Li2O, which has high ionic conductivity, during operation, resulting in the formation of a stable solid electrolyte interphase. Our findings suggest that a stable and ion-conducting anode/interphase can be developed by applying an oxide and polymer coating in combined approach. Therefore, this study is expected to provide a basis for the further development and design of high-performance Li-based battery systems.

Silicon/Carbon Nanotube/BaTiO3 Nanocomposite Anode: Evidence for Enhanced Lithium-Ion Mobility Induced by the Local Piezoelectric Potential.

We report on the synergetic effects of silicon (Si) and BaTiO3 (BTO) for applications as the anode of Li-ion batteries. The large expansion of Si during lithiation was exploited as an energy source via piezoelectric BTO nanoparticles. Si and BTO nanoparticles were dispersed in a matrix consisting of multiwalled carbon nanotubes (CNTs) using a high-energy ball-milling process. The mechanical stress resulting from the expansion of Si was transferred via the CNT matrix to the BTO, which can be poled, so that a piezoelectric potential is generated. We found that this local piezoelectric potential can improve the electrochemical performance of the Si/CNT/BTO nanocomposite anodes. Experimental measurements and simulation results support the increased mobility of Li-ions due to the local piezoelectric potential.

Novel multi-layered 1-D nanostructure exhibiting the theoretical capacity of silicon for a super-enhanced lithium-ion battery.

Silicon/carbon (Si/C) nanocomposites have recently received much attention as Li-ion battery negative electrodes due to their mutual synergetic effects in capacity and mechanical integrity. The contribution of Si to the total capacity of the Si/C nanocomposites determines their structural efficiency. Herein, we report on a multi-layered, one-dimensional nanostructure that exhibits the theoretical specific capacity of Si in the nanocomposite. Concentrically tri-layered, compartmentalized, C-core/Si-medium/C-shell nanofibers were fabricated by triple coaxial electrospinning. The pulverization of Si was accommodated inside the C-shell, whereas the conductive pathway of the Li-ions and electrons was provided by the C-core, which was proven by ex situ Raman spectroscopy. The compartmentalized Si in between the C-core and C-shell led to excellent specific capacity at a high current rate (>820 mA h g(-1) at 12000 mA g(-1)) and the realization of the theoretical specific capacity of the Li15Si4 phase of Si nanoparticles (3627 mA h g(-1)). The electrochemical characterization and inductively coupled plasma-atomic emission spectrometry provided direct evidence of full participation of Si in the electrochemical reactions.