Localized fine-tuning and clinical evaluation of deep-learning based auto-segmentation (DLAS) model for clinical target volume (CTV) and organs-at-risk (OAR) in rectal cancer radiotherapy.

BACKGROUND AND PURPOSE: Various deep learning auto-segmentation (DLAS) models have been proposed, some of which have been commercialized. However, the issue of performance degradation is notable when pretrained models are deployed in the clinic. This study aims to enhance precision of a popular commercial DLAS product in rectal cancer radiotherapy by localized fine-tuning, addressing challenges in practicality and generalizability in real-world clinical settings. MATERIALS AND METHODS: A total of 120 Stage II/III mid-low rectal cancer patients were retrospectively enrolled and divided into three datasets: training (n = 60), external validation (ExVal, n = 30), and generalizability evaluation (GenEva, n = 30) datasets respectively. The patients in the training and ExVal dataset were acquired on the same CT simulator, while those in GenEva were on a different CT simulator. The commercial DLAS software was first localized fine-tuned (LFT) for clinical target volume (CTV) and organs-at-risk (OAR) using the training data, and then validated on ExVal and GenEva respectively. Performance evaluation involved comparing the LFT model with the vendor-provided pretrained model (VPM) against ground truth contours, using metrics like Dice similarity coefficient (DSC), 95th Hausdorff distance (95HD), sensitivity and specificity. RESULTS: LFT significantly improved CTV delineation accuracy (p < 0.05) with LFT outperforming VPM in target volume, DSC, 95HD and specificity. Both models exhibited adequate accuracy for bladder and femoral heads, and LFT demonstrated significant enhancement in segmenting the more complex small intestine. We did not identify performance degradation when LFT and VPM models were applied in the GenEva dataset. CONCLUSIONS: The necessity and potential benefits of LFT DLAS towards institution-specific model adaption is underscored. The commercial DLAS software exhibits superior accuracy once localized fine-tuned, and is highly robust to imaging equipment changes.

The warming-up effects of quantum-dot light emitting diodes: A reversible stability issue related to shell traps.

The aging phenomenon is commonly observed in quantum-dot light emitting diodes (QLEDs), involving complex chemical or physical processes. Resolving the underlying mechanism of these aging issues is crucial to deliver reliable electroluminescent devices in future display applications. Here, we report a reversible positive aging phenomenon that the device brightness and efficiency significantly improve after device operation, but recover to initial states after long-time storage or mild heat treatment, which can be termed as warming-up effects. Steady and transient equivalent circuit analysis suggest that the radiative recombination current dramatically increases but electron leakage from the quantum dots (QDs) to hole transport layer becomes more accessible during the warming-up process. Further analysis discloses that the notable enhancement of device efficiency can be ascribed to the filling of shell traps in gradient alloyed QDs. This work reveals a distinct positive aging phenomenon featured with reversibility, and further guidelines would be provided to achieve stable QLED devices in real display applications.

Anticipation dynamics of pedestrians based on the elliptical social force model.

Humans have excellent predictive capabilities, and this anticipation would reflect in the interactions between people. In this work, we utilize the elliptical specification of the social force model (SFM) for pedestrian movements to study how anticipation affects motion dynamics. An elliptical potential determines the interaction between pedestrians not in contact. Anticipation is introduced by shaping the ellipse according to the relative velocity. By adjusting the time to extrapolate, we can control the strength of anticipation. Simulations are conducted in four typical scenarios, i.e., circular motion, crowd gathering, escape through a bottleneck, and free wander. In each case, the qualitative observations from visual animations are followed by quantitative analyses involving different indicators. Simulation results demonstrate that anticipation plays an important role in pedestrian dynamics in several aspects. Briefly, it helps stabilize the movement by reducing perturbations, facilitates a more ordered crowd configuration, and promotes spontaneous collective motion. The findings may set avenues for further research in anticipation dynamics.

Machine Learning Assisted Stability Analysis of Blue Quantum Dot Light-Emitting Diodes.

The operational stability of the blue quantum dot light-emitting diode (QLED) has been one of the most important obstacles to initialize its industrialization. In this work, we demonstrate a machine learning assisted methodology to illustrate the operational stability of blue QLEDs by analyzing the measurements of over 200 samples (824 QLED devices) including current density-voltage-luminance (J-V-L), impedance spectra (IS), and operational lifetime (T95@1000 cd/m2). The methodology is able to predict the operational lifetime of the QLED with a Pearson correlation coefficient of 0.70 with a convolutional neural network (CNN) model. By applying a classification decision tree analysis of 26 extracted features of J-V-L and IS curves, we illustrate the key features in determining the operational stability. Furthermore, we simulated the device operation using an equivalent circuit model to discuss the device degradation related operational mechanisms.

Effects of the direction and speed strategies on pedestrian dynamics.

In this work, we propose a velocity-based model for pedestrian movements containing two sub-models to study how the direction and speed selection strategies affect pedestrian dynamics. Affected by others in the view, pedestrians deviate their moving direction from the desired one to resolve space and velocity conflicts, the sensitivity to which is adjusted by two proportional parameters. After determining the moving direction, they choose a reasonable speed to avoid immediate collisions. The tolerance of personal-space violations is considered given that people accept physical contacts in some contexts and may squeeze despite the lack of space instead of matching the speed of the one in front. Simulation results demonstrate that the direction and speed selection strategies affect pedestrian dynamics in several aspects, including the time and metabolic energy cost, density-velocity relations, and crowd stability. Intriguingly, the discrepant empirical fundamental diagrams of pedestrians can be reproduced and explained in a natural way.

Experimental Study on Panic during Simulated Fire Evacuation Using Psycho- and Physiological Metrics.

Under circumstances of fire, panic usually brings uncertainty and unpredictability to evacuation. Therefore, a deep understanding of panic is desired. This study aims to dig into the underlying mechanism of fire evacuation panic by measuring and analysing psycho- and physiological indicators. In the experiment, participants watched a simulated train station within which three sets of stimuli were triggered separately. Eye movement and brain haemodynamic responses were collected during the watch, while questionnaires and interviews of emotions were conducted after. The analysed physiological indicators include the amplitude of pupil dilation, the time ratios of fixation and saccade, the binned entropy of gaze location, and the brain activation coefficients. The results of this research indicate that fire evacuation panic can be broken down into two elements. (1) Unawareness of situation: less knowledge of the situation leads to a higher level of panic; (2) Intensity of visual stimulation: the panic level is escalated with increased severity of fire that is perceived.

Inspection of Line Defects in Transition Metal Dichalcogenides Using a Microscopic Hyperspectral Imaging Technique.

The line defects of two-dimensional (2D) transition metal dichalcogenides (TMDs) play a vital role in determining their device performance. In this work, a microscopic hyperspectral imaging technique based on differential reflectance was introduced for the online inspection of line defects in TMDs. Upon comparison of the measurement results of imaging and spectra, the relationship between optical contrast and differential reflectance spectra was established. A light selection method was proposed to optimize the optical contrast of line defects. Via application of an image processing algorithm, an automatic detection of the line defects with a classification accuracy of 95% was achieved for WS2, MoS2, and MoSe2. This work not only provides a microscopic hyperspectral imaging technique for detecting 2D material defects but also introduces a versatile design strategy for developing an advanced machine vision spectroscopic system.

Imaging of Defect-Accelerated Energy Transfer in MoS2/hBN/WS2 Heterostructures.

Engineering energy transfer (ET) plays an important role in the exploration of novel optoelectronic devices. The efficient ET has been reasonably regulated using different strategies, such as dielectric properties, distance, and stacking angle. However, these strategies show limited degrees of freedom in regulation. Defects can provide more degrees of freedom, such as the type and density of defects. Herein, atomic-scale defect-accelerated ET is directly observed in MoS2/hBN/WS2 heterostructures by fluorescence lifetime imaging microscopy. Sulfur vacancies with different densities are introduced by controlling the oxygen plasma irradiation time. Our study shows that the ET rate can be increased from 1.25 to 6.58 ns-1 by accurately controlling the defect density. Also, the corresponding ET time is shortened from 0.80 to 0.15 ns, attributing to the participation of more neutral excitons in the ET process. These neutral excitons are transformed from trion excitons in MoS2, assisted by oxygen substitution at sulfur vacancies. Our insights not only help us better understand the role of defects in the ET process but also provide a new approach to engineer ET for further exploration of novel optoelectronic devices in van der Waals heterostructures.

Influence of elastic property on the friction between atomic force microscope tips and 2D materials.

The relationship between the elastic property of solid materials and friction has been discussed and studied by theoretical calculation and analysis. In the present work, we perform an experimental study concerning this relationship. Atomic force microscope (AFM) scanning of four different transition metal dichalcogenides is conducted under different experimental conditions. It is found that materials with smaller vertical interlayer force constant, which also means smaller elasticity modulus, have larger friction. We attribute this phenomenon to larger elastic deformation in softer materials, which results in a larger obstacle to the motion of AFM tips.

Reduced Binding Energy and Layer-Dependent Exciton Dynamics in Monolayer and Multilayer WS2.

The exciton dynamics in WS2 from monolayer to four-layer was investigated by using fluorescence lifetime imaging measurement (FLIM). The transition process of negatively charged trions is measured and detected using a fluorescence detection method. Compared with neutral excitons, negatively charged trions have a longer fluorescence lifetime. Further exploration illustrated that the fluorescence lifetime of both neutral excitons and trions get longer when the thickness increased. When WS2 was added from monolayer to four-layer, lifetimes of direct transition excitons and trions tended to increase over 10 and 2.5 times, separately, whereas the lifetime of indirect transition excitons tended to be reduced by nearly 2.5 times. This layer-dependent signature is ascribed to the reduced binding energy in thicker WS2 at room temperature, which is verified by density theory functional calculation. Although the direct transition exciton dominates the whole fluorescence decay process, it is influenced by trions and dark excitons. Based on the FLIM results, we proposed four main exciton transition channels during the fluorescence luminescence process. Such layer-dependent transition channel conception helps to control the fluorescence lifetime, which determines the efficiency of the carriers' separation.

Ray-trace simulation of CuInS(Se)2 quantum dot based luminescent solar concentrators.

To enhance the performance of luminescent solar concentrator (LSC), there is an increased need to search novel emissive materials with broad absorption and large Stokes shifts. I-III-VI colloidal CuInS2 and CuInSe2 based nanocrystals, which exhibit strong photoluminescence emissions in the visible to near infrared region with large Stokes shifts, are expected to improve performance in luminescent solar concentrator applications. In this work, the performance of CuInS(Se)2 quantum dots in simple planar LSC is evaluated by applying Monte-Carlo ray-trace simulation. A systematic parameters study was conducted to optimize the performance. An optimized photon concentration ratio of 0.34 for CuInS2 nanocrystals and 1.25 for CuInSe2 nanocrystals doping LSC are obtained from the simulation. The results demonstrated that CuInSe2 based nanocrystals are particularly interesting for luminescent solar concentrator applications, especially to combine with low price Si solar cells.

Brightly Luminescent and Color-Tunable Colloidal CH3NH3PbX3 (X = Br, I, Cl) Quantum Dots: Potential Alternatives for Display Technology.

Organometal halide perovskites are inexpensive materials with desirable characteristics of color-tunable and narrow-band emissions for lighting and display technology, but they suffer from low photoluminescence quantum yields at low excitation fluencies. Here we developed a ligand-assisted reprecipitation strategy to fabricate brightly luminescent and color-tunable colloidal CH3NH3PbX3 (X = Br, I, Cl) quantum dots with absolute quantum yield up to 70% at room temperature and low excitation fluencies. To illustrate the photoluminescence enhancements in these quantum dots, we conducted comprehensive composition and surface characterizations and determined the time- and temperature-dependent photoluminescence spectra. Comparisons between small-sized CH3NH3PbBr3 quantum dots (average diameter 3.3 nm) and corresponding micrometer-sized bulk particles (2-8 µm) suggest that the intense increased photoluminescence quantum yield originates from the increase of exciton binding energy due to size reduction as well as proper chemical passivations of the Br-rich surface. We further demonstrated wide-color gamut white-light-emitting diodes using green emissive CH3NH3PbBr3 quantum dots and red emissive K2SiF6:Mn(4+) as color converters, providing enhanced color quality for display technology. Moreover, colloidal CH3NH3PbX3 quantum dots are expected to exhibit interesting nanoscale excitonic properties and also have other potential applications in lasers, electroluminescence devices, and optical sensors.