Interpretable artificial intelligence-based app assists inexperienced radiologists in diagnosing biliary atresia from sonographic gallbladder images.

BACKGROUND: A previously trained deep learning-based smartphone app provides an artificial intelligence solution to help diagnose biliary atresia from sonographic gallbladder images, but it might be impractical to launch it in real clinical settings. This study aimed to redevelop a new model using original sonographic images and their derived smartphone photos and then test the new model's performance in assisting radiologists with different experiences to detect biliary atresia in real-world mimic settings. METHODS: A new model was first trained retrospectively using 3659 original sonographic gallbladder images and their derived 51,226 smartphone photos and tested on 11,410 external validation smartphone photos. Afterward, the new model was tested in 333 prospectively collected sonographic gallbladder videos from 207 infants by 14 inexperienced radiologists (9 juniors and 5 seniors) and 4 experienced pediatric radiologists in real-world mimic settings. Diagnostic performance was expressed as the area under the receiver operating characteristic curve (AUC). RESULTS: The new model outperformed the previously published model in diagnosing BA on the external validation set (AUC 0.924 vs 0.908, P = 0.004) with higher consistency (kappa value 0.708 vs 0.609). When tested in real-world mimic settings using 333 sonographic gallbladder videos, the new model performed comparable to experienced pediatric radiologists (average AUC 0.860 vs 0.876) and outperformed junior radiologists (average AUC 0.838 vs 0.773) and senior radiologists (average AUC 0.829 vs 0.749). Furthermore, the new model could aid both junior and senior radiologists to improve their diagnostic performances, with the average AUC increasing from 0.773 to 0.835 for junior radiologists and from 0.749 to 0.805 for senior radiologists. CONCLUSIONS: The interpretable app-based model showed robust and satisfactory performance in diagnosing biliary atresia, and it could aid radiologists with limited experiences to improve their diagnostic performances in real-world mimic settings.

Measuring the binary thickness of buccal bone of anterior maxilla in low-resolution cone-beam computed tomography via a bilinear convolutional neural network.

Background: The thickness of the buccal bone of the anterior maxilla is an important aesthetic-determining factor for dental implant, which is divided into the thick (≥1 mm) and thin type (<1 mm). However, as a micro-scale structure that is evaluated through low-resolution cone-beam computed tomography (CBCT), its thickness measurement is error-prone under the circumstance of enormous patients and relatively inexperienced primary dentists. Further, the challenges of deep learning-based analysis of the binary thickness of buccal bone include the substantial real-world variance caused by pixel error, the extraction of fine-grained features, and burdensome annotations. Methods: This study built bilinear convolutional neural network (BCNN) with 2 convolutional neural network (CNN) backbones and a bilinear pooling module to predict the binary thickness of buccal bone (thick or thin) of the anterior maxilla in an end-to-end manner. The methods of 5-fold cross-validation and model ensemble were adopted at the training and testing stages. The visualization methods of Gradient Weighted Class Activation Mapping (Grad-CAM), Guided Grad-CAM, and layer-wise relevance propagation (LRP) were used for revealing the important features on which the model focused. The performance metrics and efficacy were compared between BCNN, dentists of different clinical experience (i.e., dental student, junior dentist, and senior dentist), and the fusion of BCNN and dentists to investigate the clinical feasibility of BCNN. Results: Based on the dataset of 4,000 CBCT images from 1,000 patients (aged 36.15±13.09 years), the BCNN with visual geometry group (VGG)16 backbone achieved an accuracy of 0.870 [95% confidence interval (CI): 0.838-0.902] and an area under the receiver operating characteristic (ROC) curve (AUC) of 0.924 (95% CI: 0.896-0.948). Compared with the conventional CNNs, BCNN precisely located the buccal bone wall over irrelevant regions. The BCNN generally outperformed the expert-level dentists. The clinical diagnostic performance of the dentists was improved with the assistance of BCNN. Conclusions: The application of BCNN to the quantitative analysis of binary buccal bone thickness validated the model's excellent ability of subtle feature extraction and achieved expert-level performance. This work signals the potential of fine-grained image recognition networks to the precise quantitative analysis of micro-scale structures.

Bright white electroluminescence from polycrystalline dysprosium-doped yttrium gallium garnet nanofilms fabricated by atomic layer deposition on silicon.

Bright white emission is obtained under electrical excitation from dysprosium doped Y3Ga5O12 garnet (YGG:Dy) nanofilms fabricated by atomic layer deposition on silicon substrates. Electroluminescence (EL) composed of yellow (580 nm) and blue (482 and 492 nm) emission corresponds to the CIE chromaticity coordinates of (0.3568, 0.3807) and a CCT of ∼4700 K and can be used for lighting and displays. The crystallization and micro-morphology of polycrystalline YGG:Dy nanolaminates are explored by adjusting the annealing temperature, Y/Ga ratio, Ga2O3 interlayer thickness and Dy2O3 dopant cycle. The near-stoichiometric device annealed at 1000 °C presents optimal EL performance with the maximum external quantum efficiency and the optical power density reaching 6.35% and 18.13 mW cm-2, respectively. The EL decay time is estimated to be 273.05 µs, with a large excitation section of 8.33 × 10-15 cm2. The conduction mechanism is confirmed to be the Poole-Frenkel mode under operation electric fields and the impact-excitation of Dy3+ ions by energetic electrons results in emission. Bright white emission from Si-based YGG:Dy devices can provide a new route to developing integrated light sources and display applications.

Enhancement of the Electroluminescence from Amorphous Er-Doped Al2O3 Nanolaminate Films by Y2O3 Cladding Layers Using Atomic Layer Deposition.

Amorphous Al2O3-Y2O3:Er nanolaminate films are fabricated on silicon by atomic layer deposition, and ~1530 nm electroluminescence (EL) is obtained from the metal-oxide-semiconductor light-emitting devices based on these nanofilms. The introduction of Y2O3 into Al2O3 reduces the electric field for Er excitation and the EL performance is significantly enhanced, while the electron injection of devices and the radiative recombination of doped Er3+ ions are not impacted. The 0.2 nm Y2O3 cladding layers for Er3+ ions increase the external quantum efficiency from ~3% to 8.7% and the power efficiency is increased by nearly one order of magnitude to 0.12%. The EL is ascribed to the impact excitation of Er3+ ions by hot electrons, which stem from Poole-Frenkel conduction mechanism under sufficient voltage within the Al2O3-Y2O3 matrix.

Improving the luminescence property of the novel yellow-emitting phosphor SrLa2Sc2O7:Bi3+ with charge compensators (Li+, Na+, K+) and its application in NUV-based white LEDs.

In this work, a novel yellow-emitting phosphor SrLa2Sc2O7(SLSO):Bi3+ was synthesized by a high temperature solid-state method, with a wide coverage from 400 nm to 800 nm under near-ultraviolet (NUV) excitation and the full width at half maximum (FWHM) of up to 180 nm. The phosphor emits bright yellow light, which is observed using optical microscopy. Alkali metal (Li+, Na+, K+) ions were used as charge compensators to enhance the luminescence intensities of SrLa2Sc2O7:Bi3+. Finally, the yellow SrLa2Sc2O7:Bi3+,Li+ and blue Sr5(PO4)3Cl:Eu2+ phosphors were homogeneously mixed with silica gel and coated on a NUV light emitting diode (LED) at 370 nm to obtain white LEDs with a high color rendering index (Ra = 91.3), a low correlated color temperature (CCT = 4879 K) and color coordinates (0.34, 0.30). This work can provide a good candidate for making white LEDs.

A non-rare earth ion doped broadband emission phosphor located in NIR-II for ethanol concentration detection.

In this work, a new NIR phosphor Ca2GeO4:xCr was developed, whose emission range (1000-1700 nm) is well within NIR-II, with a full width at half maximum (FWHM) of 215 nm. Interestingly, by sintering under an air atmosphere, the Cr3+ ions in the raw material are oxidized to Cr4+, and the near-infrared light emitted by the material is not the usual luminescence of Cr3+ ions located in NIR-I, which can be attributed to the dominant substitution of Cr4+ ions doped into the lattice sites. A pc-LED fabricated with this phosphor can detect different concentrations of ethanol, which demonstrates its good potential for non-destructive testing applications.

Structural modulation designed a thermally robust blue-cyan emitting phosphor Y2Mg0.8Sr0.2 Al4SiO12:Eu2+ for the high color rendering index white LEDs.

In this work, the novel, to the best of our knowledge, blue-cyan Y2Mg0.8Sr0.2Al4SiO12:Eu2+ (YM0.8S0.2AS:Eu2+) phosphors were synthesized by the solid-state method. At 150°C, the emission intensity of Y2Mg0.8Sr0.2Al4SiO12:0.005Eu2+ can retain 96.38% of the relative intensity, which means that this phosphor shows high thermal stability. A white light-emitting diode (LED) device is fabricated by combining a 370 nm near ultraviolet LED chip and commercial phosphors (green, (Ba,Sr)2SiO4:Eu; red, CaSiAlN3:Eu). The white LED has an excellent property with the correlated color temperature CCT=5236K and superhigh color rendering index Ra=96.1, which indicates the potential application in white LED fields.

Near-Infrared Emitting Phosphor LaMg0.5(SnGe)0.5O3:Cr3+ for Plant Growth Applications: Crystal Structure, Luminescence, and Thermal Stability.

Corresponding to the absorption curve of plant photosensitive pigment Pfr, near-infrared light has a broad application prospect in plant lighting. In order to explore this plant growth lamp, the novel near-infrared emitting phosphor LaMg0.5Sn0.5O3:Cr3+ was synthesized, which can be efficiently excited by 467 nm blue light. There are two luminescence centers, which were proven by testing the low-temperature spectrum, the excitation spectrum at different wavelengths, and the lifetime decay curve, and the two cell sequence substitution processes were obtained by Rietveld refinements of XRD. By introducing Ge4+, its luminescence intensity has been increased 1.6 times and the intensity at 150 °C remains at almost 80% that at room temperature. Finally, two different kinds of illumination experiments for plant growth were carried out, and the feasibility of the LaMg0.5(SnGe)0.5O3:Cr3+ phosphor for plant growth was confirmed.

A single luminescence center ultra-broadband near-infrared LiScGeO4:Cr phosphor for biological tissue penetration.

In this work, in order to meet the application of near-infrared phosphor-converted light emitting diodes (pc-LEDs), an ultra-broadband emission phosphor, LiScGeO4:Cr, was synthesized. Its FWHM reaches 335 nm, and its emission spectrum ranges from 800 nm to 1650 nm, which almost covers the entire near-infrared second window (NIR-II). The broadband emission is thought to be caused by the 4T2 → 4A2 transition of the Cr3+ ion. This transition occurs due to the olivine structure of the crystal, which causes the Cr3+ ions to inhabit a low-symmetric crystal field, and the crystal field strength is very weak. NIR pc-LEDs were fabricated by combining a 460 nm blue LED with this phosphor, which penetrates 4 cm thick beef. The results indicate that there may be a potential application for this phosphor in the field of biological tissue penetration and non-destructive testing.