RESUMEN
BACKGROUND: Spine MR image segmentation is important foundation for computer-aided diagnostic (CAD) algorithms of spine disorders. Convolutional neural networks segment effectively, but require high computational costs. PURPOSE: To design a lightweight model based on dynamic level-set loss function for high segmentation performance. STUDY TYPE: Retrospective. POPULATION: Four hundred forty-eight subjects (3163 images) from two separate datasets. Dataset-1: 276 subjects/994 images (53.26% female, mean age 49.02 ± 14.09), all for disc degeneration screening, 188 had disc degeneration, 67 had herniated disc. Dataset-2: public dataset with 172 subjects/2169 images, 142 patients with vertebral degeneration, 163 patients with disc degeneration. FIELD STRENGTH/SEQUENCE: T2 weighted turbo spin echo sequences at 3T. ASSESSMENT: Dynamic Level-set Net (DLS-Net) was compared with four mainstream (including U-net++) and four lightweight models, and manual label made by five radiologists (vertebrae, discs, spinal fluid) used as segmentation evaluation standard. Five-fold cross-validation are used for all experiments. Based on segmentation, a CAD algorithm of lumbar disc was designed for assessing DLS-Net's practicality, and the text annotation (normal, bulging, or herniated) from medical history data were used as evaluation standard. STATISTICAL TESTS: All segmentation models were evaluated with DSC, accuracy, precision, and AUC. The pixel numbers of segmented results were compared with manual label using paired t-tests, with P < 0.05 indicating significance. The CAD algorithm was evaluated with accuracy of lumbar disc diagnosis. RESULTS: With only 1.48% parameters of U-net++, DLS-Net achieved similar accuracy in both datasets (Dataset-1: DSC 0.88 vs. 0.89, AUC 0.94 vs. 0.94; Dataset-2: DSC 0.86 vs. 0.86, AUC 0.93 vs. 0.93). The segmentation results of DLS-Net showed no significant differences with manual labels in pixel numbers for discs (Dataset-1: 1603.30 vs. 1588.77, P = 0.22; Dataset-2: 863.61 vs. 886.4, P = 0.14) and vertebrae (Dataset-1: 3984.28 vs. 3961.94, P = 0.38; Dataset-2: 4806.91 vs. 4732.85, P = 0.21). Based on DLS-Net's segmentation results, the CAD algorithm achieved higher accuracy than using non-cropped MR images (87.47% vs. 61.82%). DATA CONCLUSION: The proposed DLS-Net has fewer parameters but achieves similar accuracy to U-net++, helps CAD algorithm achieve higher accuracy, which facilitates wider application. EVIDENCE LEVEL: 2 TECHNICAL EFFICACY: Stage 1.
Asunto(s)
Procesamiento de Imagen Asistido por Computador , Degeneración del Disco Intervertebral , Humanos , Femenino , Adulto , Persona de Mediana Edad , Masculino , Procesamiento de Imagen Asistido por Computador/métodos , Estudios Retrospectivos , Degeneración del Disco Intervertebral/diagnóstico por imagen , Redes Neurales de la Computación , Columna Vertebral/diagnóstico por imagenRESUMEN
BACKGROUND: Segmenting spinal tissues from MR images is important for automatic image analysis. Deep neural network-based segmentation methods are efficient, yet have high computational costs. PURPOSE: To design a lightweight model based on small-world properties (LSW-Net) to segment spinal MR images, suitable for low-computing-power embedded devices. STUDY TYPE: Retrospective. POPULATION: A total of 386 subjects (2948 images) from two independent sources. Dataset I: 214 subjects/779 images, all for disk degeneration screening, 147 had disk degeneration, 52 had herniated disc. Dataset II: 172 subjects/2169 images, 142 patients with vertebral degeneration, 163 patients with disc degeneration. 70% images in each dataset for training, 20% for validation, and 10% for testing. FIELD STRENGTH/SEQUENCE: T1- and T2-weighted turbo spin echo sequences at 3 T. ASSESSMENT: Segmentation performance of LSW-Net was compared with four mainstream (including U-net and U-net++) and five lightweight models using five radiologists' manual segmentations (vertebrae, disks, spinal fluid) as reference standard. LSW-Net was also deployed on NVIDIA Jetson nano to compare the pixels number in segmented vertebrae and disks. STATISTICAL TESTS: All models were evaluated with accuracy, precision, Dice similarity coefficient (DSC), and area under the receiver operating characteristic (AUC). Pixel numbers segmented by LSW-Net on the embedded device were compared with manual segmentation using paired t-tests, with P < 0.05 indicating significance. RESULTS: LSW-Net had 98.5% fewer parameters than U-net but achieved similar accuracy in both datasets (dataset I: DSC 0.84 vs. 0.87, AUC 0.92 vs. 0.94; dataset II: DSC 0.82 vs. 0.82, AUC 0.88 vs. 0.88). LSW-Net showed no significant differences in pixel numbers for vertebrae (dataset I: 5893.49 vs. 5752.61, P = 0.21; dataset II: 5073.42 vs. 5137.12, P = 0.56) and disks (dataset I: 1513.07 vs. 1535.69, P = 0.42; dataset II: 1049.74 vs. 1087.88, P = 0.24) segmentation on an embedded device compared to manual segmentation. DATA CONCLUSION: Proposed LSW-Net achieves high accuracy with fewer parameters than U-net and can be deployed on embedded device, facilitating wider application. EVIDENCE LEVEL: 2. TECHNICAL EFFICACY: 1.
Asunto(s)
Degeneración del Disco Intervertebral , Imagen por Resonancia Magnética , Humanos , Estudios Retrospectivos , Imagen por Resonancia Magnética/métodos , Degeneración del Disco Intervertebral/diagnóstico por imagen , Redes Neurales de la Computación , Procesamiento de Imagen Asistido por Computador/métodos , Columna Vertebral/diagnóstico por imagenRESUMEN
Molecular structural modifications are utilized to improve the short-circuit current (JSC ) of high-voltage organic photovoltaics (OPVs). Herein, the classic non-fullerene acceptor (NFA), BTA3, is chosen as a benchmark, with BTA3b containing the linear alkyl chains on the middle core and JC14 fusing thiophene on the benzotriazole (BTA) unit as a contrast. The photovoltaic devices based on J52-F: BTA3b and J52-F: JC14 achieve wider external quantum efficiency responses with band edges of 730 and 800 nm, respectively than that of the device based on J52-F: BTA3 (715 nm). The corresponding JSC increases to 14.08 and 15.78 mA cm-2 , respectively, compared to BTA3 (11.56 mA cm-2 ). The smaller Urbach energy and higher electroluminescence efficiency guarantee J52-F: JC14 a decreased energy loss (0.528 eV) and a high open-circuit voltage (VOC ) of 1.07 V. Finally, J52-F: JC14 combination achieves an increased power conversion efficiency (PCE) of 10.33% than that of J52-F: BTA3b (PCE = 9.81%) and J52-F: BTA3 (PCE = 9.04%). Overall, the research results indicate that subtle structure modification of NFAs, especially introducing fused rings, is a simple and effective strategy to extend the photoelectric response, boosting the JSC and ensuring a high VOC beyond 1.0 V.
RESUMEN
Simulations can mimic the environment that refers to the surgery operation to improve the technical skills of the trainees. In this paper, we designed a new cardiac surgery simulative training system. The isolated pig heart was selected as the heart model. A mechanical device was designed to achieve the beating of heart model. At the same time, adjusting frequencies of mechanical movement could change the rating of heartbeat. In order to validate the rationality of the system, 12 non-medical specialty students and 12 medical specialty students were divided into two groups, which consecutively accepted seven-days of training for off-pump coronary artery bypass grafting using the cardiac surgery simulative training system. The time for completing bypass grafting before and after training were recorded. And the bridging outcomes of each trainee were assessed by 3 surgery cardiac surgeons using the object structured assessments of technical skill (OSATS) criteria. After training, each trainee could finish the bypass suturing in a shorter time than before training, and the scores of each trainee assessed by OSATS criteria were also improved. The results showed that the cardiac surgery simulative training system had better training effect in improving the surgical techniques, operation skills and proficiency of surgical instruments of trainees.
RESUMEN
Segmentation of intervertebral discs and vertebrae from spine magnetic resonance (MR) images is essential to aid diagnosis algorithms for lumbar disc herniation. Convolutional neural networks (CNN) are effective methods, but often require high computational costs. Designing a lightweight CNN is more suitable for medical sites lacking high-computing power devices, yet due to the unbalanced pixel distribution in spine MR images, the segmentation is often sub-optimal. To address this issue, a lightweight spine segmentation CNN based on a self-adjusting loss function, which is named SALW-Net, is proposed in this study. For SALW-Net, the self-adjusting loss function could dynamically adjust the loss weights of the two branches according to the differences in segmentation results and labels during the training; thus, the ability for learning unbalanced pixels is enhanced. Two separate datasets are used to evaluate the proposed SALW-Net. Specifically, the proposed SALW-Net has fewer parameter numbers than U-net (only 2%) but achieves higher evaluation scores than that of U-net (the average DSC score of SALW-Net is 0.8781, and that of U-net is 0.8482). In addition, the practicality validation for SALW-Net is also proceeding, including deploying the model on a lightweight device and producing an aid diagnosis algorithm based on segmentation results. This means our SALW-Net has clinical application potential for assisted diagnosis in low computational power scenarios.
Asunto(s)
Procesamiento de Imagen Asistido por Computador , Redes Neurales de la Computación , Procesamiento de Imagen Asistido por Computador/métodos , Imagen por Resonancia Magnética/métodos , Algoritmos , Columna Vertebral/diagnóstico por imagenRESUMEN
Topological electronic transition is the very promising strategy for achieving high band degeneracy (NV) and for optimizing thermoelectric performance. Herein, this work verifies in p-type Mg3Sb2- xBix that topological electronic transition could be the key mechanism responsible for elevating the NV of valence band edge from 1 to 6, leading to much improved thermoelectric performance. Through comprehensive spectroscopy characterizations and theoretical calculations of electronic structures, the topological electronic transition from trivial semiconductor is unambiguously demonstrated to topological semimetal of Mg3Sb2- xBix with increasing the Bi content, due to the strong spin-orbit coupling of Bi and the band inversion. The distinct evolution of Fermi surface configuration and the multivalley valence band edge with NV of 6 are discovered in the Bi-rich compositions, while a peculiar two-step band inversion is revealed for the first time in the end compound Mg3Bi2. As a result, the optimal p-type Mg3Sb0.5Bi1.5 simultaneously obtains a positive bandgap and high NV of 6, and thus acquires the largest thermoelectric power factor of 3.54 and 6.93 µW cm-1 K-2 at 300 and 575 K, respectively, outperforming the values in other compositions. This work provides important guidance on improving thermoelectric performance of p-type Mg3Sb2- xBix utilizing the topological electronic transition.
RESUMEN
3D reconstruction for lumbar spine based on segmentation of Magnetic Resonance (MR) images is meaningful for diagnosis of degenerative lumbar spine diseases. However, spine MR images with unbalanced pixel distribution often cause the segmentation performance of Convolutional Neural Network (CNN) reduced. Designing a composite loss function for CNN is an effective way to enhance the segmentation capacity, yet composition loss values with fixed weight may still cause underfitting in CNN training. In this study, we designed a composite loss function with a dynamic weight, called Dynamic Energy Loss, for spine MR images segmentation. In our loss function, the weight percentage of different loss values could be dynamically adjusted during training, thus CNN could fast converge in earlier training stage and focus on detail learning in the later stage. Two datasets were used in control experiments, and the U-net CNN model with our proposed loss function achieved superior performance with Dice similarity coefficient values of 0.9484 and 0.8284 respectively, which were also verified by the Pearson correlation, Bland-Altman, and intra-class correlation coefficient analysis. Furthermore, to improve the 3D reconstruction based on the segmentation results, we proposed a filling algorithm to generate contextually related slices by computing the pixel-level difference between adjacent slices of segmented images, which could enhance the structural information of tissues between slices, and improve the performance of 3D lumbar spine model rendering. Our methods could help radiologists to build a 3D lumbar spine graphical model accurately for diagnosis while reducing burden of manual image reading.
Asunto(s)
Procesamiento de Imagen Asistido por Computador , Imagenología Tridimensional , Algoritmos , Procesamiento de Imagen Asistido por Computador/métodos , Imagenología Tridimensional/métodos , Imagen por Resonancia Magnética/métodos , Redes Neurales de la Computación , HumanosRESUMEN
A large number of epidemics, including COVID-19 and SARS, quickly swept the world and claimed the precious lives of large numbers of people. Due to the concealment and rapid spread of the virus, it is difficult to track down individuals with mild or asymptomatic symptoms with limited human resources. Building a low-cost and real-time epidemic early warning system to identify individuals who have been in contact with infected individuals and determine whether they need to be quarantined is an effective means to mitigate the spread of the epidemic. In this paper, we propose a smartphone-based zero-effort epidemic warning method for mitigating epidemic propagation. Firstly, we recognize epidemic-related voice activity relevant to epidemics spread by hierarchical attention mechanism and temporal convolutional network. Subsequently, we estimate the social distance between users through sensors built-in smartphone. Furthermore, we combine Wi-Fi network logs and social distance to comprehensively judge whether there is spatiotemporal contact between users and determine the duration of contact. Finally, we estimate infection risk based on epidemic-related vocal activity, social distance, and contact time. We conduct a large number of well-designed experiments in typical scenarios to fully verify the proposed method. The proposed method does not rely on any additional infrastructure and historical training data, which is conducive to integration with epidemic prevention and control systems and large-scale applications.
RESUMEN
Herein, we synthesized an asymmetric A2-A1-D-A1-A2 type small molecule nonfullerene acceptor (NFA), HCl-BTA3, by chlorination on one side of A1. The synergistic effect of the asymmetric structure and chlorination endows HCl-BTA3 with a large dipole moment, close molecular packing, and high-efficiency charge transfer and transport. After being blended with a carboxylate-based polymer donor, TTC-Cl, HCl-BTA3 achieved a high open-circuit voltage (VOC) of 1.20 V and a satisfactory power conversion efficiency (PCE) of 11.2%, which are among the highest values for high-voltage carboxylate-based polymers.
RESUMEN
As one of the polymer modification strategies, carboxylate functionalization has proved effective in downshifting the energy levels and enhancing polymer crystallinity and aggregation. However, high-performance carboxylate-containing polymers are still limited for organic solar cells (OSCs), especially with open-circuit voltage (VOC) above 1.0 V. Herein, we utilize two carboxylate-functionalized wide-band gap (WBG) donor polymers (TTC-F and TTC-Cl) to pair with two WBG electron acceptors (BTA5 and F-BTA5) for high-voltage OSCs. Due to the deeper molecular energy levels, chlorinated polymer TTC-Cl shows higher VOC than fluorinated polymer TTC-F. Furthermore, because of the stronger aggregation in the film, the TTC-Cl-based devices attain suppressed energetic disorders and trap-assisted recombination, decreasing voltage loss and JSC loss. Finally, the TTC-Cl: F-BTA5 blend achieves a higher VOC of 1.17 V and an excellent PCE of 10.98%, one of the best results for high-voltage carboxylate-containing polymers. In addition, the TTC-Cl: BTA5 combination demonstrates the highest VOC of 1.25 V with an ultralow nonradiative energy loss of 0.17 eV. Our results indicate that the carboxylate-containing polymer donors have significant application potential for high-voltage OSCs due to reduced energy loss and improved charge transport and dissociation. Furthermore, the matched absorption spectra with the indoor light sources and low voltage loss promote these material combinations to construct high-performance indoor photovoltaics.
RESUMEN
Bacteria are a crucial component of forest soil biodiversity and play an important role in numerous ecosystem processes. Here, we studied the patterns of soil bacterial community diversity and structure in a climax forest (Larix gmelinii; LG) compared with those in degraded forest ecosystems of four forest vegetation types (BD, Betula dahurica; BP, Betula platyphylla; QM, Quercus mongolica; and LGQM, a mixed coniferous-broadleaved forest composed of Larix gmelinii and Quercus mongolica) in the Heilongjiang Zhongyangzhan Black-billed Capercaillie Nature Reserve in northern China, using Illumina MiSeq sequencing of 16 S rRNA genes. Soil physicochemical properties (pH, soil organic carbon = SOC, total nitrogen = TN, carbon/nitrogen = C/N, total phosphorous = TP, available nitrogen = AN, available phosphorous = AP) differed significantly (p < .05) among the five forests. SOC, C/N, TP, AN, and AP were highest in QM, whereas SOC was lowest in LGQM. Soil pH was lowest in BD and highest in LGQM. α diversity was highest in LG and lowest in QM. The soil bacterial community composition in the climax forest was significantly different from that in the four degraded forests (p < .05). The dominant bacterial phyla were Acidobacteria, Proteobacteria, Verrucomicrobia, Bacteroidetes, Actinobacteria, Gemmatimonadetes, Firmicutes, Chloroflexi, and Rokubacteria. The highest relative abundances of these phyla were: 30.7% for Acidobacteria in LGQM, 42.6% for Proteobacteria in LG, 17.6% for Verrucomicrobia in BD, 5.5% for Firmicutes in BP, and 6.9% for Actinobacteria in QM. The dominant bacterial genera across the five forest vegetation types were Bryobacter and some poorly characterized taxa (e.g., Candidatus_Udaeobacter and Candidatus_Solibacter). Redundancy analysis indicated that SOC, C/N, TP, AN, and AP were the main soil physicochemical properties that shaped the bacterial communities. Our study revealed distinct bacterial diversity and composition in the climax forest compared with values in degraded forests, suggesting that the biotic and abiotic factors associated with climax ecosystems play an important role in shaping soil bacterial community structure and thus biogeochemical functions. The results of this study contribute to a deeper understanding and better predictions of the network among belowground systems and of the functions and services of degraded forests compared with climax ecosystems.
RESUMEN
Mg3Sb2-based compounds are one type of important room-temperature thermoelectric materials and the appropriate candidate of type-II nodal line semimetals. In Mg3Sb2-based films, compelling research topics such as dimensionality reduction and topological states rely on the controllable preparation of films with high crystallinity, which remains a big challenge. In this work, high quality Mg3Sb2 films are successfully grown on mismatched substrates of sapphire (000l), while the temperature-driven twin structure evolution and characteristics of the electronic structure are revealed in the as-grown Mg3Sb2 films by in situ and ex situ measurements. The transition of layer-to-island growth of Mg3Sb2 films is kinetically controlled by increasing the substrate temperature (Tsub), which is accompanied with the rational manipulation of twin structure and epitaxial strains. Twin-free structure could be acquired in the Mg3Sb2 film grown at a low Tsub of 573 K, while the formation of twin structure is significantly promoted by elevating the Tsub and annealing, in close relation to the processes of strain relaxation and enhanced mass transfer. Measurements of scanning tunneling spectroscopy (STS) and angle-resolved photoemission spectroscopy (ARPES) elucidate the intrinsic p-type conduction of Mg3Sb2 films and a bulk band gap of ~0.89 eV, and a prominent Fermi level downshift of ~0.2 eV could be achieved by controlling the film growth parameters. As elucidated in this work, the effective manipulation of the epitaxial strains, twin structure and Fermi level is instructive and beneficial for the further exploration and optimization of thermoelectric and topological properties of Mg3Sb2-based films.