RESUMEN
Single-cell analyses parse the brain's billions of neurons into thousands of 'cell-type' clusters residing in different brain structures1. Many cell types mediate their functions through targeted long-distance projections allowing interactions between specific cell types. Here we used epi-retro-seq2 to link single-cell epigenomes and cell types to long-distance projections for 33,034 neurons dissected from 32 different regions projecting to 24 different targets (225 source-to-target combinations) across the whole mouse brain. We highlight uses of these data for interrogating principles relating projection types to transcriptomics and epigenomics, and for addressing hypotheses about cell types and connections related to genetics. We provide an overall synthesis with 926 statistical comparisons of discriminability of neurons projecting to each target for every source. We integrate this dataset into the larger BRAIN Initiative Cell Census Network atlas, composed of millions of neurons, to link projection cell types to consensus clusters. Integration with spatial transcriptomics further assigns projection-enriched clusters to smaller source regions than the original dissections. We exemplify this by presenting in-depth analyses of projection neurons from the hypothalamus, thalamus, hindbrain, amygdala and midbrain to provide insights into properties of those cell types, including differentially expressed genes, their associated cis-regulatory elements and transcription-factor-binding motifs, and neurotransmitter use.
Asunto(s)
Encéfalo , Epigenómica , Vías Nerviosas , Neuronas , Animales , Ratones , Amígdala del Cerebelo , Encéfalo/citología , Encéfalo/metabolismo , Secuencia de Consenso , Conjuntos de Datos como Asunto , Perfilación de la Expresión Génica , Hipotálamo/citología , Mesencéfalo/citología , Vías Nerviosas/citología , Neuronas/metabolismo , Neurotransmisores/metabolismo , Secuencias Reguladoras de Ácidos Nucleicos , Rombencéfalo/citología , Análisis de la Célula Individual , Tálamo/citología , Factores de Transcripción/metabolismoRESUMEN
MOTIVATION: In drug discovery, it is crucial to assess the drug-target binding affinity (DTA). Although molecular docking is widely used, computational efficiency limits its application in large-scale virtual screening. Deep learning-based methods learn virtual scoring functions from labeled datasets and can quickly predict affinity. However, there are three limitations. First, existing methods only consider the atom-bond graph or one-dimensional sequence representations of compounds, ignoring the information about functional groups (pharmacophores) with specific biological activities. Second, relying on limited labeled datasets fails to learn comprehensive embedding representations of compounds and proteins, resulting in poor generalization performance in complex scenarios. Third, existing feature fusion methods cannot adequately capture contextual interaction information. RESULTS: Therefore, we propose a novel DTA prediction method named HeteroDTA. Specifically, a multi-view compound feature extraction module is constructed to model the atom-bond graph and pharmacophore graph. The residue concat graph and protein sequence are also utilized to model protein structure and function. Moreover, to enhance the generalization capability and reduce the dependence on task-specific labeled data, pre-trained models are utilized to initialize the atomic features of the compounds and the embedding representations of the protein sequence. A context-aware nonlinear feature fusion method is also proposed to learn interaction patterns between compounds and proteins. Experimental results on public benchmark datasets show that HeteroDTA significantly outperforms existing methods. In addition, HeteroDTA shows excellent generalization performance in cold-start experiments and superiority in the representation learning ability of drug-target pairs. Finally, the effectiveness of HeteroDTA is demonstrated in a real-world drug discovery study. AVAILABILITY AND IMPLEMENTATION: The source code and data are available at https://github.com/daydayupzzl/HeteroDTA.
Asunto(s)
Descubrimiento de Drogas , Descubrimiento de Drogas/métodos , Simulación del Acoplamiento Molecular , Proteínas/química , Proteínas/metabolismo , Aprendizaje Profundo , FarmacóforoRESUMEN
Selecting training samples is crucial in remote sensing image classification. In this paper, we selected three images-Sentinel-2, GF-1, and Landsat 8-and employed three methods for selecting training samples: grouping selection, entropy-based selection, and direct selection. We then used the selected training samples to train three supervised classification models-random forest (RF), support-vector machine (SVM), and k-nearest neighbor (KNN)-and evaluated the classification results of the three images. According to the experimental results, the three classification models performed similarly. Compared with the entropy-based method, the grouping selection method achieved higher classification accuracy using fewer samples. In addition, the grouping selection method outperformed the direct selection method with the same number of samples. Therefore, the grouping selection method performed the best. When using the grouping selection method, the image classification accuracy increased with the increase in the number of samples within a certain sample size range.
RESUMEN
Recurrent Neural Networks (RNNs) are applied in safety-critical fields such as autonomous driving, aircraft collision detection, and smart credit. They are highly susceptible to input perturbations, but little research on RNN-oriented testing techniques has been conducted, leaving a threat to a large number of sequential application domains. To address these gaps, improve the test adequacy of RNNs, find more defects, and improve the performance of RNNs models and their robustness to input perturbations. We aim to propose a test coverage metric for the underlying structure of RNNs, which is used to guide the generation of test inputs to test RNNs. Although coverage metrics have been proposed for RNNs, such as the hidden state coverage in RNN-Test, they ignore the fact that the underlying structure of RNNs is still a fully connected neural network but with an additional "delayer" that records the network state at the time of data input. We use the contributions, i.e., the combination of the outputs of neurons and the weights they emit, as the minimum computational unit of RNNs to explore the finer-grained logical structure inside the recurrent cells. Compared to existing coverage metrics, our research covers the decision mechanism of RNNs in more detail and is more likely to generate more adversarial samples and discover more flaws in the model. In this paper, we redefine the contribution coverage metric applicable to Stacked LSTMs and Stacked GRUs by considering the joint effect of neurons and weights in the underlying structure of the neural network. We propose a new coverage metric, RNNCon, which can be used to guide the generation of adversarial test inputs. And we design and implement a test framework prototype RNNCon-Test. 2 datasets, 4 LSTM models, and 4 GRU models are used to verify the effectiveness of RNNCon-Test. Compared to the current state-of-the-art study RNN-Test, RNNCon can cover a deeper decision logic of RNNs. RNNCon-Test is not only effective in identifying defects in Deep Learning (DL) systems but also in improving the performance of the model if the adversarial inputs generated by RNNCon-Test are filtered and added to the training set to retrain the model. In the case where the accuracy of the model is already high, RNNCon-Test is still able to improve the accuracy of the model by up to 0.45%.
RESUMEN
With the development of the Internet of things (IoT), federated learning (FL) has received increasing attention as a distributed machine learning (ML) framework that does not require data exchange. However, current FL frameworks follow an idealized setup in which the task size is fixed and the storage space is unlimited, which is impossible in the real world. In fact, new classes of these participating clients always emerge over time, and some samples are overwritten or discarded due to storage limitations. We urgently need a new framework to adapt to the dynamic task sequences and strict storage constraints in the real world. Continuous learning or incremental learning is the ultimate goal of deep learning, and we introduce incremental learning into FL to describe a new federated learning framework. New generation federated learning (NGFL) is probably the most desirable framework for FL, in which, in addition to the basic task of training the server, each client needs to learn its private tasks, which arrive continuously independent of communication with the server. We give a rigorous mathematical representation of this framework, detail several major challenges faced under this framework, and address the main challenges of combining incremental learning with federated learning (aggregation of heterogeneous output layers and the task transformation mutual knowledge problem), and show the lower and upper baselines of the framework.
Asunto(s)
Algoritmos , Aprendizaje Automático , Humanos , ComputadoresRESUMEN
Machine learning combined with satellite image time series can quickly, and reliably be implemented to map crop distribution and growth monitoring necessary for food security. However, obtaining a large number of field survey samples for classifier training is often time-consuming and costly, which results in the very slow production of crop distribution maps. To overcome this challenge, we propose an ensemble learning approach from the existing historical crop data layer (CDL) to automatically create multitudes of samples according to the rules of spatiotemporal sample selection. Sentinel-2 monthly composite images from 2017 to 2019 for crop distribution mapping in Jilin Province were mosaicked and classified. Classification accuracies of four machine learning algorithms for a single-month and multi-month time series were compared. The results show that deep neural network (DNN) performed the best, followed by random forest (RF), then decision tree (DT), and support vector machine (SVM) the least. Compared with other months, July and August have higher classification accuracy, and the kappa coefficients of 0.78 and 0.79, respectively. Compared with a single phase, the kappa coefficient gradually increases with the growth of the time series, reaching 0.94 in August at the earliest, and then the increase is not obvious, and the highest in the whole growth cycle is 0.95. During the mapping process, time series of different lengths produced different classification results. Wetland types were misclassified as rice. In such cases, authors combined time series of two lengths to correct the misclassified rice types. By comparing with existing products and field points, rice has the highest consistency, followed by corn, whereas soybeans have the least consistency. This shows that the generated sample data set and trained model in this research can meet the crop mapping accuracy and simultaneously reduce the cost of field surveys. For further research, more years and types of crops should be considered for mapping and validation.
Asunto(s)
Redes Neurales de la Computación , Máquina de Vectores de Soporte , Algoritmos , Productos Agrícolas , Aprendizaje AutomáticoRESUMEN
In this paper, aiming to solve the problem of vital information security as well as neural network application in optical encryption system, we propose an optical image encryption method by using the Hopfield neural network. The algorithm uses a fuzzy single neuronal dynamic system and a chaotic Hopfield neural network for chaotic sequence generation and then obtains chaotic random phase masks. Initially, the original images are decomposed into sub-signals through wavelet packet transform, and the sub-signals are divided into two layers by adaptive classification after scrambling. The double random-phase encoding in 4f system and Fresnel domain is implemented on two layers, respectively. The sub-signals are performed with different conversions according to their standard deviation to assure that the local information's security is guaranteed. Meanwhile, the parameters such as wavelength and diffraction distance are considered as additional keys, which can enhance the overall security. Then, inverse wavelet packet transform is applied to reconstruct the image, and a second scrambling is implemented. In order to handle and manage the parameters used in the scheme, the public key cryptosystem is applied. Finally, experiments and security analysis are presented to demonstrate the feasibility and robustness of the proposed scheme.
RESUMEN
Saturation effects limit the application of vegetation indices (VIs) in dense vegetation areas. The possibility to mitigate them by adopting a negative soil adjustment factor X is addressed. Two leaf area index (LAI) data sets are analyzed using the Google Earth Engine (GEE) for validation. The first one is derived from observations of MODerate resolution Imaging Spectroradiometer (MODIS) from 16 April 2013, to 21 October 2020, in the Apiacás area. Its corresponding VIs are calculated from a combination of Sentinel-2 and Landsat-8 surface reflectance products. The second one is a global LAI dataset with VIs calculated from Landsat-5 surface reflectance products. A linear regression model is applied to both datasets to evaluate four VIs that are commonly used to estimate LAI: normalized difference vegetation index (NDVI), soil adjusted vegetation index (SAVI), transformed SAVI (TSAVI), and enhanced vegetation index (EVI). The optimal soil adjustment factor of SAVI for LAI estimation is determined using an exhaustive search. The Dickey-Fuller test indicates that the time series of LAI data are stable with a confidence level of 99%. The linear regression results stress significant saturation effects in all VIs. Finally, the exhaustive searching results show that a negative soil adjustment factor of SAVI can mitigate the SAVIs' saturation in the Apiacás area (i.e., X = -0.148 for mean LAI = 5.35), and more generally in areas with large LAI values (e.g., X = -0.183 for mean LAI = 6.72). Our study further confirms that the lower boundary of the soil adjustment factor can be negative and that using a negative soil adjustment factor improves the computation of time series of LAI.
Asunto(s)
Hojas de la Planta , Suelo , Modelos Lineales , Imágenes SatelitalesRESUMEN
Image encryption is a confidential strategy to keep the information in digital images from being leaked. Due to excellent chaotic dynamic behavior, self-feedbacked Hopfield networks have been used to design image ciphers. However, Self-feedbacked Hopfield networks have complex structures, large computational amount and fixed parameters; these properties limit the application of them. In this paper, a single neuronal dynamical system in self-feedbacked Hopfield network is unveiled. The discrete form of single neuronal dynamical system is derived from a self-feedbacked Hopfield network. Chaotic performance evaluation indicates that the system has good complexity, high sensitivity, and a large chaotic parameter range. The system is also incorporated into a framework to improve its chaotic performance. The result shows the system is well adapted to this type of framework, which means that there is a lot of room for improvement in the system. To investigate its applications in image encryption, an image encryption scheme is then designed. Simulation results and security analysis indicate that the proposed scheme is highly resistant to various attacks and competitive with some exiting schemes.
RESUMEN
The roles of tumor necrosis factor alpha (TNF-alpha) and its mediators in cellular processes related to intestinal diseases remain elusive. In this study, we aimed to determine the biological role of activated Cdc42-associated kinase 1 (ACK1) in TNF-alpha-mediated apoptosis and proliferation in Caco-2 cells. ACK1 expression was knocked down using ACK1-specific siRNAs, and ACK1 activity was disrupted using a small molecule ACK1 inhibitor. The Terminal deoxynucleotidyl transferase biotin-dUTP Nick End Labeling (TUNEL) and the BrdU incorporation assays were used to measure apoptosis and cell proliferation, respectively. ACK1-specific siRNA and the pharmacological ACK1 inhibitor significantly abrogated the TNF-alpha-mediated anti-apoptotic effects and proliferation of Caco-2 cells. Interestingly, TNF-alpha activated ACK1 at tyrosine 284 (Tyr284), and the ErbB family of proteins was implicated in ACK1 activation in Caco-2 cells. ACK1-Tyr284 was required for protein kinase B (AKT) activation, and ACK1 signaling was mediated through recruiting and phosphorylating the down-stream adaptor protein AKT, which likely promoted cell proliferation in response to TNF-alpha. Moreover, ACK1 activated AKT and Src enhanced nuclear factor-кB (NF-кB) activity, suggesting a correlation between NF-кB signaling and TNF-alpha-mediated apoptosis in Caco-2 cells. Our results demonstrate that ACK1 plays an important role in modulating TNF-alpha-induced aberrant cell proliferation and apoptosis, mediated in part by ACK1 activation. ACK1 and its down-stream effectors may hold promise as therapeutic targets in the prevention and treatment of gastrointestinal cancers, in particular, those induced by chronic intestinal inflammation.
Asunto(s)
Proteínas Tirosina Quinasas/metabolismo , Factor de Necrosis Tumoral alfa/farmacología , Apoptosis/efectos de los fármacos , Apoptosis/fisiología , Células CACO-2 , Proliferación Celular/fisiología , Receptores ErbB/metabolismo , Técnicas de Silenciamiento del Gen , Humanos , Mucosa Intestinal/metabolismo , Intestinos/citología , Intestinos/enzimología , FN-kappa B/metabolismo , Fosforilación , Inhibidores de Proteínas Quinasas/farmacología , Proteínas Tirosina Quinasas/antagonistas & inhibidores , Proteínas Tirosina Quinasas/genética , Proteínas Proto-Oncogénicas c-akt/metabolismo , ARN Interferente Pequeño/administración & dosificación , ARN Interferente Pequeño/genética , Transducción de Señal/efectos de los fármacos , Factor de Necrosis Tumoral alfa/metabolismo , Familia-src Quinasas/metabolismoRESUMEN
The research on the distribution and component of olivine is one of great significance to evaluate the geologic evolution of igneous planetary bodies such as the Moon. In this paper, the Sinus Iidium, as the survey region, was explored by Chang's serial satellite. Here we present an olivine survey of the Sinus Iridium by using Spectral Feature Fitting (SFF) method based on the M3 data. The exposures of olivine were located in the northern crater wall and at the foot of Montes Jura, which were associated with plagioclase and little anorthosite. The stratigraphic units of the located formation were the interior crater slopes and debris ejected from the impact-formed Iridium crater, and the geological age was relatively older. The Mg number of the lunar olivine samples was dependent variables, and the band center of the lunar olivine spectrums were independent, which derived from the fitting analysis using Modified Gaussian Model (MGM). The quantitative inversion models of Mg number (Fo#) of the lunar olivine is established with multiple linear regression analysis. On this basis, the Mg number of the olivine rich point in the Sinus Iridium are calculated with quantitative inversion models of Mg number (Fo#). The result shows that, the Fo# of olivine in the Sinus Iridium are relatively high. The mean value of Fo# is Fo~80.84. As mantle olivine would be expected to be quite Mg-rich, it is suggested that at the vast majority of the olivine detected in the Sinus Iridium come from upper mantle of the Moon.
RESUMEN
Phyllosilicate belongs to hydrated silica, which is a principal form of hydrous minerals on the martian surface. It's also an indicator in comparing different sediments and degree of aqueous alteration. Therefore, it's essential to establish its recognition model for studying the geologic evolution of the Mars. Short-wave infrared (SWIR) spectral bands and thermal infrared (TIR) spectral bands have distinct spectral response to the mineral groups and ions, so they have distinctive advantages in detecting minerals. However the method of combining SWIR and TIR to recognize phyllosilicate is rarely studied. Based on the USGS spectral library, facing Compact Reconnaissance Imaging Spectrometer for Mars(CRISM) and Thermal Emission Imaging System(THEMIS),we conducted the research on the mechanism of the spectral response of phyllosilicate, and established the SWIR and TIR identification model respectively, then combined the SWIR and TIR spectral features to build the combined recognition model of phyllosilicate with Fisher discriminant analysis. The results of cross validation show that the identification accuracy of combined model is the highest, which can correctly classify 90.6% of the mineral samples and improve the identification precision of phyllosilicate effectively.
RESUMEN
In this research, 97 pieces of rock in Xingcheng, Liaoning Province, China were collected to measure the spectral reflectance in 350~2 500 nm, chemical content, and complex dielectric constant of some samples. The absorption depths were calculated by using continuum- removal method. With correlation analysis method, two kinds of correlation curves were obtained based on the theory of spectral characteristics of chemical contents and the principle of dielectric constant. One described the relationship between chemical content and spectral absorption depth, and the other one represented the correlation of complex dielectric constant and reflectance. By summarizing curves morphological characteristics, several conclusions were drawn as follows: (1)There was a strong correlation between the chemical content (SiO2, Al2O3, CaO, K2O, MgO, burnt-loss) and spectral absorption depth in 1 900~2 500 nm, furthermore, at around 1 900, 2 200, 2 300 nm and other identifying characteristic bands, local extreme maximum / minimum values appeared. At Fe3+ characteristic band (400~550 nm), correlation coefficient reached -0.406 between Fe2O3 content and absorption in igneous rock samples collection. Exploring the relationship between rock spectral absorption features and its chemical contents had a positive effect on metallogenic prediction and lithology identification with remote sensing image. (2) Reflectance and complex dielectric constant were negatively correlated totally, compared with the imaginary part; the real part had a better relation reached -0.753 at around 1 900 nm. Curves showed that there were great correlations around 1 900 and 2 200 nm, so, our study adopted different models to simulate response relationships. Dielectric constant of media is one of the basic physical properties, and now most analyses of existing research between electromagnetic characteristics and dielectric constant are studied in microwave band, however, our research is conducted in visible and near infrared range. The conclusions will be useful for further exploration on dielectric characteristics and spectral features of rocks.
RESUMEN
The spectral absorption features are very similar between some minerals, especially hydrothermal alteration minerals related to mineralization, and they are also influenced by other factors such as spectral mixture. As a result, many of the spectral identification approaches for the minerals with similar spectral absorption features are prone to confusion and misjudgment. Therefore, to solve the phenomenon of "same mineral has different spectrums, and same spectrum belongs to different minerals", this paper proposes an accurate approach to hyperspectral mineral identification based on naive bayesian classification model. By testing and analyzing muscovite and kaolinite, the two typical alteration minerals, and comparing this approach with spectral angle matching, binary encoding and spectral feature fitting, the three popular spectral identification approaches, the results show that this approach can make more obvious differences among different minerals having similar spectrums, and has higher classification accuracy, since it is based on the position of absorption feature, absorption depth and the slope of continuum.
RESUMEN
The Moon may be considered as the frontier base for the deep space exploration. The spectral analysis is one of the key techniques to determine the lunar surface rock and mineral compositions. But the lunar topographic relief is more remarkable than that of the Earth. It is necessary to conduct the topographic correction for lunar spectral data before they are used to retrieve the compositions. In the present paper, a lunar Sandmeier model was proposed by considering the radiance effect from the macro and ambient topographic relief. And the reflectance correction model was also reduced based on the Sandmeier model. The Spectral Profile (SP) data from KAGUYA satellite in the Sinus Iridum quadrangle was taken as an example. And the digital elevation data from Lunar Orbiter Laser Altimeter are used to calculate the slope, aspect, incidence and emergence angles, and terrain-viewing factor for the topographic correction Thus, the lunar surface reflectance from the SP data was corrected by the proposed model after the direct component of irradiance on a horizontal surface was derived. As a result, the high spectral reflectance facing the sun is decreased and low spectral reflectance back to the sun is compensated. The statistical histogram of reflectance-corrected pixel numbers presents Gaussian distribution Therefore, the model is robust to correct lunar topographic effect and estimate lunar surface reflectance.
RESUMEN
Language bias stands as a noteworthy concern in visual question answering (VQA), wherein models tend to rely on spurious correlations between questions and answers for prediction. This prevents the models from effectively generalizing, leading to a decrease in performance. In order to address this bias, we propose a novel modality fusion collaborative de-biasing algorithm (CoD). In our approach, bias is considered as the model's neglect of information from a particular modality during prediction. We employ a collaborative training approach to facilitate mutual modeling between different modalities, achieving efficient feature fusion and enabling the model to fully leverage multimodal knowledge for prediction. Our experiments on various datasets, including VQA-CP v2, VQA v2, and VQA-VS, using different validation strategies, demonstrate the effectiveness of our approach. Notably, employing a basic baseline model resulted in an accuracy of 60.14% on VQA-CP v2.
RESUMEN
BACKGROUND: Mastering the spatial distribution and planting area of paddy can provide a scientific basis for monitoring rice production, and planning grain production layout. Previous remote sensing studies on paddy concentrated in the plain areas with large-sized fields, ignored the fact that rice is also widely planted in vast hilly regions. In addition, the land cover types here are diverse, rice fields are characterized by a scattered and fragmented distribution with small- or medium-sized, which pose difficulties for high-precision rice recognition. METHODS: In the paper, we proposed a solution based on Sentinel-1 SAR, Sentinel-2 MSI, DEM, and rice calendar data to focus on the rice fields identification in hilly areas. This solution mainly included the construction of rice feature dataset at four crucial phenological periods, the generation of rice standard spectral curve, and the proposal of spectral similarity algorithm for rice identification. RESULTS: The solution, integrating topographical and rice phenological characteristics, manifested its effectiveness with overall accuracy exceeding 0.85. Comparing the results with UAV, it presented that rice fields with an area exceeding 400 m2 (equivalent to 4 pixels) exhibited a recognition success rate of over 79%, which reached to 89% for fields exceeding 800 m2. CONCLUSIONS: The study illustrated that the proposed solution, integrating topographical and rice phenological characteristics, has the capability for charting various rice field sizes with fragmented and dispersed distribution. It also revealed that the synergy of Sentinel-1 SAR and Sentinel-2 MSI data significantly enhanced the recognition ability of rice paddy fields ranging from 400 m2 to 2000 m2.
RESUMEN
Background and objectives: Glioblastoma (GBM) and brain metastasis (MET) are the two most common intracranial tumors. However, the different pathogenesis of the two tumors leads to completely different treatment options. In terms of magnetic resonance imaging (MRI), GBM and MET are extremely similar, which makes differentiation by imaging extremely challenging. Therefore, this study explores an improved deep learning algorithm to assist in the differentiation of GBM and MET. Materials and methods: For this study, axial contrast-enhanced T1 weight (ceT1W) MRI images from 321 cases of high-grade gliomas and solitary brain metastasis were collected. Among these, 251 out of 270 cases were selected for the experimental dataset (127 glioblastomas and 124 metastases), 207 cases were chosen as the training dataset, and 44 cases as the testing dataset. We designed a new deep learning algorithm called SCAT-inception (Spatial Convolutional Attention inception) and used five-fold cross-validation to verify the results. Results: By employing the newly designed SCAT-inception model to predict glioblastomas and brain metastasis, the prediction accuracy reached 92.3%, and the sensitivity and specificity reached 93.5 and 91.1%, respectively. On the external testing dataset, our model achieved an accuracy of 91.5%, which surpasses other model performances such as VGG, UNet, and GoogLeNet. Conclusion: This study demonstrated that the SCAT-inception architecture could extract more subtle features from ceT1W images, provide state-of-the-art performance in the differentiation of GBM and MET, and surpass most existing approaches.
RESUMEN
Background: High-grade gliomas (HGG) and solitary brain metastases (SBM) are two common types of brain tumors in middle-aged and elderly patients. HGG and SBM display a high degree of similarity on magnetic resonance imaging (MRI) images. Consequently, differential diagnosis using preoperative MRI remains challenging. This study developed deep learning models that used pre-operative T1-weighted contrast-enhanced (T1CE) MRI images to differentiate between HGG and SBM before surgery. Methods: By comparing various convolutional neural network models using T1CE image data from The First Medical Center of the Chinese PLA General Hospital and The Second People's Hospital of Yibin (Data collection for this study spanned from January 2016 to December 2023), it was confirmed that the GoogLeNet model exhibited the highest discriminative performance. Additionally, we evaluated the individual impact of the tumoral core and peritumoral edema regions on the network's predictive performance. Finally, we adopted a slice-based voting method to assess the accuracy of the validation dataset and evaluated patient prediction performance on an additional test dataset. Results: The GoogLeNet model, in a five-fold cross-validation using multi-plane T1CE slices (axial, coronal, and sagittal) from 180 patients, achieved an average patient accuracy of 92.78%, a sensitivity of 95.56%, and a specificity of 90.00%. Moreover, on an external test set of 29 patients, the model achieved an accuracy of 89.66%, a sensitivity of 90.91%, and a specificity of 83.33%, with an area under the curve of 0.939 [95% confidence interval (CI): 0.842-1.000]. Conclusions: GoogLeNet performed better than previous methods at differentiating HGG from SBM, even for core and peritumoral edema in both. HGG and SBM could be fast screened using this end-to-end approach, improving workflow for both tumor treatments.
RESUMEN
Nowadays, Federated Learning (FL) over Internet of Medical Things (IoMT) devices has become a current research hotspot. As a new architecture, FL can well protect the data privacy of IoMT devices, but the security of neural network model transmission can not be guaranteed. On the other hand, the sizes of current popular neural network models are usually relatively extensive, and how to deploy them on the IoMT devices has become a challenge. One promising approach to these problems is to reduce the network scale by quantizing the parameters of the neural networks, which can greatly improve the security of data transmission and reduce the transmission cost. In the previous literature, the fixed-point quantizer with stochastic rounding has been shown to have better performance than other quantization methods. However, how to design such quantizer to achieve the minimum square quantization error is still unknown. In addition, how to apply this quantizer in the FL framework also needs investigation. To address these questions, in this paper, we propose FedMSQE - Federated Learning with Minimum Square Quantization Error, that achieves the smallest quantization error for each individual client in the FL setting. Through numerical experiments in both single-node and FL scenarios, we prove that our proposed algorithm can achieve higher accuracy and lower quantization error than other quantization methods.