Search

Intracerebral hemorrhage (ICH) is life-threatening. The intraventricular extension of ICH (IVH) also frequently occurs, thus increasing the risk of disability or death. The site and amount of bleeding are important predictors of prognosis. This study aimed to predict Glasgow Outcome Scores (GOSs) by automatically segmenting hemorrhage sites from brain computed tomography data combined with clinical information. Data of 192 patients from Hanyang University Hospital with IVH and ICH from March 2016 to September 2020 were analyzed. To perform automatic segmentation through deep learning, two neurosurgeons manually generated correct answer values. Preprocessing was performed to capture more pathological tissue. Hemorrhagic sites were robustly predicted by feeding high-quality generated data to the HTransUNet that hierarchically combined a Convolution Neural Network and transformer; the predicted mask was combined with clinical information to predict GOSs. This model showed high performance and competitive performance in segmenting IVH and ICH compared with other segmentation models. Additionally, when predicting GOS, it performed better than the ICH score system. The predicted mask with clinical information performed better than clinical information alone. Using the cerebral hemorrhage segmentation and GOS prediction models in clinical settings, as auxiliary indicators for rapid decision-making, can contribute significantly to patient management.

Here, we develop a dry eye syndrome (DES) incidence rate prediction model using air pollutants (PM10, NO2, SO2, O3, and CO), meteorological factors (temperature, humidity, and wind speed), population rate, and clinical data for South Korea. The prediction model is well fitted to the incidence rate (R2 = 0.9443 and 0.9388, p &lt, 2.2 &times, 10&minus, 16). To analyze regional deviations, we classify outpatient data, air pollutant, and meteorological factors in 16 administrative districts (seven metropolitan areas and nine states). Our results confirm NO2 and relative humidity are the factors impacting regional deviations in the prediction model.

Depression is a mental illness that more than 3 million people suffer worldwide. Depression can lead to suicide in severe cases, but only less than half of them get proper treatment. To prevent the worst decision that patients make, it is necessary that let the patients know that he or she has depression and need to go to the hospital. Previous studies have shown that patients with depression have different from healthy people about frontal brain wave asymmetry. Therefore, we propose prefrontal brain wave asymmetry-based image with a deep learning-based model to prescreen the patient’s depression.

Metal nanoparticle has been a promising option for fillers in thermal interface materials due to its low cost and ease of fabrication. However, nanoparticle aggregation effect is not well understood because of its complexity. Theoretical models, like effective medium approximation model, barely cover aggregation effect. In this work, we have fabricated nickel–epoxy nanocomposites and observed higher thermal conductivity than effective medium theory predicts. Smaller particles are also found to show higher thermal conductivity, contrary to classical models indicate. A two-level effective medium approximation (EMA) model is developed to account for aggregation effect and to explain the size-dependent enhancement of thermal conductivity by introducing local concentration in aggregation structures.

In this work, we have observed 60% reduction in total interfacial resistance by adding an intermediate metal layer nickel between gold and aluminum oxide. Two temperature model is applied to explain the change of interfacial resistance, including both lattice mismatch with diffuse mismatch model and electron-phonon coupling effect. Simulation result agrees reasonably well with experimental data. Even though interfacial resistance due to electron-phonon coupling effect for Au-aluminum oxide is much larger than that of Ni-aluminum oxide interface, lattice mismatch is still the dominant factor for interfacial resistance.

Metal nanoparticle has been a promising option for fillers in thermal interface materials due to its low cost and ease of fabrication. However, nanoparticle aggregation effect is not well understood because of its complexity. Theoretical models, like effective medium approximation model, barely cover aggregation effect. In this work, we have fabricated nickel-epoxy nanocomposites and observed higher thermal conductivity than effective medium theory predicts. Smaller particles are also found to show higher thermal conductivity, contrary to classical models indicate. A two-level EMA model is developed to account for aggregation effect and to explain the size-dependent enhancement of thermal conductivity by introducing local concentration in aggregation structures.

Interfacial thermal resistance between metal and dielectric materials is a bottleneck of the thermal management for modern integrated circuits as interface density increases with thinner films. In this work, we have observed that the interfacial resistance across gold and aluminum oxide can be reduced from 4.8 × 10 − 8 m 2 K / W to 1.4 × 10 − 8 m 2 K / W after adding a nickel layer in between, which represents a 70% reduction. The two temperature model is applied to explain the reduction of interfacial resistance, and the results show that the nickel layer functions as a bridge that reduces the phonon mismatch between gold and aluminum oxide. Moreover, nickel has strong electron-phonon coupling, which reduces the thermal resistance caused by the weak electron-phonon coupling in gold.

We developed high-performance thermal interface materials (TIMs) based on few-layer graphene (FLG) composite, where FLG was prepared by the interlayer catalytic exfoliation (ICE) method. We experimentally demonstrated feasibility of FLG composites as TIMs by investigating their thermal and mechanical properties, and reliability. We measured the thermal interface resistance ($R_{int}$) between FLG composite TIMs (FLGTs) and copper and to be 3.2$\pm$1.7 and 4.3$\pm$1.4 $mm^2$K/W for 5 vol.% and 10 vol.% FLGTs at 330 K, respectively, comparable to or even lower than that of many commercial TIMs. In addition, the thermal conductivity ($\kappa_{TIM}$) of FLGTs is increased by an enhancement factor ($\beta$) of ~17 as the FLG concentration increases from 0 to 10 vol.%. We also characterized Vickers hardness and glass transition temperature ($T_g$) of our FLGTs. We find that our FLGTs are thermally and mechanically reliable within practical operating temperature and pressure ranges.

Sandwich structures of aluminum oxide, nickel, and aluminum oxide films are fabricated by atomic layer deposition to study thermal interfacial resistance between a metal and a dielectric material and the interfacial coupling effect across a thin metal layer. Thermal resistance of a thin nickel layer as well as two interfaces is measured using the 3ω method. Experimental results show interfacial resistance between nickel and aluminum oxide to be 6.8×10−3mm2 K/W at 300 K, with a weak dependence on the metal thickness and temperature. A two-temperature model and a detailed diffuse mismatch model have been used to estimate interfacial resistance theoretically, and the results agree reasonably well with experiments. Estimations from the two temperature model indicate that in the overall thermal interfacial resistance, the phonon-phonon interfacial resistance dominates over the resistance due to the electron-phonon coupling effect and inside the metal layer. Also, the phonon-phonon interfacial resistance does n...

We report compressive mechanical response of graphene foams (GFs) and the thermal resistance ($R_{TIM}$) between copper (Cu) and GFs, where GFs were prepared by the chemical vapor deposition (CVD) method. We observe that Young's modulus ($E_{GF}$) and compressive strength ($\sigma_{GF}$) of GFs have a power law dependence on increasing density ($\rho_{GF}$) of GFs. The maximum efficiency of absorbed energy ($\eta_{max}$) for all GFs during the compression is larger than ~0.39. We also find that a GF with a higher $\rho_{GF}$ shows a larger $\eta_{max}$. In addition, we observe that the measured $R_{TIM}$ of Cu/GFs at room temperature with a contact pressure of 0.25 MP applied increases from ~50 to ~90 $mm^2K/W$ when $\rho_{GF}$ increases from 4.7 to 31.9 $mg/cm^3$.

The control and enhancement of the combustion wave propagation velocities of solid monopropellants are very important for the development of low cost and efficient micro power systems such as microthrusters and thermal-to-electrical energy conversion devices. In this work, the flame speed enhancement of a nitrocellulose (NC) solid monopropellant using highly conductive graphene structures was demonstrated. Two different graphene structures, namely, graphene foam (GF) and graphene nano-pellets (GNPs), were studied. For the GNP-doped NC films, fuel layers 500 ± 30 μm thick were deposited and the doping concentrations were varied from 1% to 5% by mass. For the GF, the fuel loading ratio (%) and the foam density were varied to study their effect on the flame speed propagation behavior. Self-propagating combustion waves were observed, with average flame speed enhancements up to 8 times the bulk value. The flame speed enhancement, for both the GNPs and the GF, showed a parabolic trend as a function of their con...