Test Dataset Research Articles

Optical convolution operations with optical neural networks for incoherent color image recognition

This study introduces optical neural networks (ONNs) designed to accelerate optical convolution operations using a red, green, and blue (RGB) pixel array integrated into conventional display technology. In a proof-of-concept demonstration, we initially employed a rank-4 kernel for a normal convolution network, which was integrated with a fully connected layer, to accurately classify color images across five fruit categories. Following seven epochs of on-system iterative training for a 3,000 training dataset, the ONN achieved 96% classification accuracy and maintained robust performance on an unseen 1,000 test dataset. Our analysis also showed its potential for efficient operation, with a classification accuracy exceeding 94% using an average less than 34 aJ of optical energy per MAC operation. Additionally, we demonstrated depth-wise convolution with a rank-3 kernel, recurring the system to spectrally resolve the signals into independent R, G, and B channels. This architecture enabled the successful classification of complex patterns containing three MNIST handwritten digits encoded in RGB. Our strategy contributes significantly to optical computing and neuromorphic vision, facilitating efficient recognition of real-world, multi-color, and incoherent light images.

Abstract TMP54: Development and testing of a fully automated tool for the detection, segmentation, and characterization of cervical carotid atherosclerotic disease

Background: Rapid, accurate diagnosis and characterization of carotid atherosclerosis can help prevent disabling strokes. Although carotid plaques can be identified on CT angiography (CTA), interpretation is challenging for frontline physicians. Quantification of plaque volume/composition requires much manual effort. We developed and tested a fully automated tool for segmenting carotid lumens and delineating atherosclerotic plaque, distinguishing between calcific and hypodense components. Methods: We used 528 consecutive cases with CTA head/neck from a population of 7,745 patients with ischemic stroke and transient ischemic attack in an entire province (Alberta) presenting from 1-April-2016 to 31-March-2017. Trained readers supervised by a radiologist manually segmented right and left carotid artery lumens from 3 vertebral bodies below the bulb to 3 above, and segmented regions of carotid atherosclerotic plaque, regardless of degree of stenosis, separately labelling calcific and hypodense components. Cases were split 80/20 between training and testing datasets. The fully automated pipeline included coarse-scale detection of regions of interest, a two-stream U-shaped network for detection and segmentation of lumens and plaques, and a geometry-based inference algorithm to distinguish left and right labels. We evaluated plaque detection using diagnostic performance measures and segmentation using the Dice coefficient. Results: 422 cases were used for training and 106 for testing. In testing data, the fully automated tool achieved excellent performance for bilateral segmentation of carotid lumens (Dice 0.91, 95%CI:0.90-0.92, e.g. Figure 1 ). For detection of calcific and hypodense plaque components, respectively, the model achieved sensitivity of 96.5% (95%CI:89.3-99.1%) and 97.3% (89.6-99.5%), specificity of 95.2% (74.1-99.8%) and 75.8% (57.4-88.3%), positive predictive value of 98.8% (92.6-99.9%) and 89.9% (80.5-95.2%), negative predictive value of 87.0% (65.3-96.6%) and 92.6% (74.2-98.7%), and accuracy of 96.2% (90.1-98.8%) and 90.6% (82.9-95.1%, e.g. Figure 2 ). Dice score was 0.73 (0.69-0.76) for calcific plaque and 0.53 (0.49-0.57) for hypodense plaque. Conclusions: Our fully automated tool achieved good performance for detection and segmentation of carotid plaques. Calcific and hypodense plaque volumes can be automatically generated from these labels. Efforts are underway to further optimize the specificity and segmentation performance for hypodense plaques.

Terrain classification and rock abundance analysis at Utopia Planitia using Zhurong image data based on deep learning algorithms

The complexity of image scene information presents challenges for the trafficability assessment and path planning of Mars rovers. To ensure the operational safety of Mars rovers and extract terrain features from complex image scenes, this paper develops an end-to-end deep learning model, using the deep convolutional neural networks ResNet50 and DeepLabV3 + as the framework, with images from the Zhurong rover’s navigation camera as the training and test datasets. A deep learning model suitable for classification and segmentation of terrain in the Mars Utopia Planitia region has been established and applied to planetary geology research. The classification accuracy of model exceeds 83.90 % and segmentation accuracy exceeds 80 %. Subsequently, an analysis of 1309 raw images from the navigation camera yielded 203,744 individual estimates of rock abundance, the model evaluates the rock abundance in the Utopia Planitia region, where the Zhurong rover is located, at 10.94 %. The terrain classification model proposed in this study provides both engineering and scientific value for future rovers on the Utopia Planitia.

Application of perceptual graph neural network intelligent algorithm in software testing platform

In the research on algorithm application in software testing platform, the traditional perceptual graph neural network (PGNN) intelligent algorithm has high computational complexity when processing large-scale and complex software testing data. At the same time, it lacks sufficient ability to automatically select and optimize the code’s syntax structure, execution path, test coverage and other features, resulting in poor test case generation effect. This paper constructed an improved PGNN model, which enhanced the modeling ability of complex code structures and execution paths by adding graph convolution layers and introducing attention mechanisms. It also used PGNNs to integrate and optimize features such as grammatical structure, execution path, and test coverage in software codes. Matrix decomposition and sparsification techniques are used to optimize the computational complexity of the PGNN model and achieve rapid processing of large-scale software test data. Based on the improved PGNN model, training is performed on the software test data set, and the model is tuned in combination with a variety of optimization algorithms to improve the overall performance of the test platform. The experimental results show that in terms of the effectiveness of test case generation, the coverage rate, error detection rate, missed rate and false positive rate are 0.87, 0.92, 0.01 and 0.02 respectively.

Abstract TP167: Segmentation of leukoaraiosis on noncontrast brain using CT-MRI paired data without human annotation

Purpose: White matter hyperintensities (WMH), also known as leukoaraiosis (LA), are common brain abnormalities in elderly individuals. Evaluating LA on CT is challenging due to the less distinguishable hypoattenuation against white matter. We aimed to develop and validate a segmenting algorithm of LA using CT-MRI paired data. Methods: We included 744 CT-MRI paired dataset of patients with ischemic stroke from 4 stroke centers. An external test set comprised 411 patients from non-overlapping hospitals. WMH on MRI was segmented using validated software and the segmentation mask was registered onto NCCT space. We compared predicted LA versus ground-truth registered LA and WMH on MRI using Dice similarity coefficient (DSC) and concordance correlation coefficient (ρ). Results: Mean age (SD) for training and external test datasets were 68.1 (SD 12.7) and 69.2 (SD 13.5) years and 33.2% and 47.9% were female, respectively. In the internal validation dataset, the algorithm achieved a DSC of 0.53, with a volumetric correlation (ρ) of 0.848 with registered LA volume on CT. External validation showed a DSC of 0.527, with ρ values of 0.919 and 0.760 for predicted LA volumes compared to registered LA and WMH volumes on MRI, respectively. Subgroup analysis demonstrated consistent performance across different CT vendors and infarct volumes. Conclusion: Our deep learning algorithm offers a significant advancement in LA segmentation on CT, bridging the gap between CT and MRI assessments. It improves patient care by providing a consistent, accessible, and accurate method for evaluating LA, supporting both clinical practice and large-scale research.

Efficient proxy for time-lapse seismic forward modeling using a U-net encoder-decoder approach

The time-lapse seismic (4D seismic) forward modeling provides crucial data for calibrating reservoir models through different data assimilation algorithms. Unfortunately, the traditional 4D seismic forward-modeling methodology is time-expensive and entails significant computational resource consumption. To address these drawbacks, in this work, our goal is to develop a proxy model for the 4D seismic forward modeling using a class of machine learning algorithm named U-Net encoder–decoder. We applied the developed proxy model to a benchmark carbonate reservoir using an ensemble of reservoir simulation models from UNISIM IV dataset (a synthetic benchmark based on real data of a Brazilian pre-salt field). Moreover, we aim to introduce seminal strategies for interpreting the proposed proxy model operation, its outputs, and possible correlations between input and output variables. To achieve this, we trained and tested two versions of U-net-based models and applied methods for explainable artificial intelligence, such as Gradient-weighted Class Activation Mapping (Grad-CAM) and Forward Feature Selection. The experiments showed good results when applied to the test dataset. The correlation coefficient (R2) values were in the range of 0.7 to 0.9, showing the efficiency of the proxy model to replace the 4D seismic forward modeling. Through qualitative analysis, it was possible to identify which input properties and regions of the reservoir are more relevant for the model’s inference. These results are a step toward robust, explainable machine learning-based proxy forward modeling.

Prediction of reduced left ventricular ejection fraction using atrial fibrillation or flutter electrocardiograms: A machine-learning study.

Although the evaluation of left ventricular ejection fraction (LVEF) in patients with atrial fibrillation (AF) or atrial flutter (AFL) is crucial for appropriate medical management, the prediction of reduced LVEF (<50%) with AF/AFL electrocardiograms (ECGs) lacks evidence. This study aimed to investigate deep-learning approaches to predict reduced LVEF (<50%) in patients with AF/AFL ECGs and easily obtainable clinical information. Patients with 12-lead ECGs of AF/AFL and echocardiography were divided into those with LVEF <50% and ≥50%. A convolutional neural networks-based model customized to the study (AFibEFNet) and other deep-learning models were investigated. Electrocardiogram signals, ECG features, and clinical features (demographic information, comorbidities, blood cell counts, and blood test results) were collected for training. A hold-out test dataset was constructed using a different recruitment period. Five-fold cross-validation and calibration plots were used to evaluate performance. A total of 15,683 patients were analyzed (mean age, 70.0 ± 11.7 years; 61.2% men), with 82.2% having LVEF ≥50% and 17.8% having LVEF < 50%. Among the learning models, the AFibEFNet outperformed other models regarding area under the receiver operating characteristic curve (AUROC), area under the precision-recall curve (AUPRC), and F1-score. Using ECG signals alone, the AFibEFNet model predicted reduced LVEF with AUROC of 0.798 (95% confidence interval [CI], 0.767-0.829) and AUPRC of 0.508 (95% CI, 0.434-0.564). For the AFibEFNet model, additional training with ECG and clinical features significantly improved AUROC (0.816 vs. 0.798, p = 0.04) and AUPRC (0.547 vs. 0.508, p < 0.001). The AFibEFNet model primarily focused on the R-wave, QRS onset and offset, and T-wave in ECG signals. Among the patients with AF/AFL, machine learning may predict reduced LVEF with 12-lead ECGs of AF/AFL.

An AttSDNet model for multi-scale feature perception enhanced remote sensing classification of coastal salt-marsh wetlands.

Coastal salt-marsh wetlands have important ecological value, and play an important role in coastal blue carbon sink. However, under the influence of various external and natural factors, coastal wetland ecosystems worldwide have severely degraded, leading to biodiversity loss and ecological damage. Based on satellite remote sensing data and deep learning methods, it is an effective means to quickly monitor the spatial distribution of coastal wetlands, which is very important for the protection and restoration of coastal wetlands. The U-Net deep learning framework, because of its low data requirements, fast training speed, and efficient architectural design, has seen rapid development and widespread application in the field of image segmentation. However, applying the classic U-Net architecture to the classification of coastal wetland images, which have rich and complex cover types. It struggles to effectively capture the spatial dependencies and multi-scale feature information present in remote sensing images. To address this issue, this study introduces an enhanced U-Net model that integrates attention mechanisms and multi-scale feature extraction. This model employs stacked dilated convolutions to improve the U-Net's single receptive field limitation, thereby enhancing the model's ability to learn the multi-scale features of typical land covers in complex coastal wetlands. Furthermore, a combined channel-spatial attention mechanism module is incorporated to strengthen the extraction and learning of spectral and spatial features of remote sensing image land covers. This highlights the feature of small-scale land covers that are difficult to capture. Remote sensing image classification was conducted using Sentinel-2 optical imagery on the coastal wetlands of the Yellow River Estuary and Jiaozhou Bay located in Shandong Peninsula, China. An independent test dataset was used to validate the model's accuracy, and comparative experiments were conducted with several existing classification methods. The results show that the proposed model achieved the highest classification accuracy in coastal wetland remote sensing image classification compared to SVM, VGG, FCN, U-Net, ResU-Net, and SDU-Net models. The overall accuracy of the two study areas is 92.73% and 98.69%, and the MIoU is 77.68% and 83.76%, respectively. For different scales of land cover types, such as larger-scale distributions of Tamarix chinensis and ponds, the improved model's MIoU increased by 17.72% and 5.45%, respectively. For elongated structures like artificial roads and tidal channels, the MIoU improved by 9.82% and 5.41%. The proposed method effectively extracts and learns the remote sensing feature information of land cover targets at different scales, enhances the classification accuracy of large-scale land covers, and effectively addresses the issues of detail loss in small target classification and disconnection in linear land cover classification. It provides a more accurate and robust technical method for coastal wetland remote sensing classification, offering a solid data foundation for analyzing the distribution of typical land covers. Additionally, it has significant implications for efficiently monitoring biodiversity and protecting the ecological environment in coastal wetlands.

Deep Learning Segmentation-Based Bone Removal from Computed Tomography of the Brain Improves Subdural Hematoma Detection

Timely identification of intracranial blood products is clinically impactful, however the detection of subdural hematoma (SDH) on non-contrast CT scans of the head (NCCTH) is challenging given interference from the adjacent calvarium. This work explores the utility of a NCCTH bone removal algorithm for improving SDH detection. A deep learning segmentation algorithm was designed/trained for bone removal using 100 NCCTH. Segmentation accuracy was evaluated on 15 NCCTH from the same institution and 22 NCCTH from an independent external dataset using quantitative overlap analysis between automated and expert manual segmentations. The impact of bone removal on detecting SDH by junior radiology trainees was evaluated with a reader study comparing detection performance between matched cases with and without bone removal applied. Average Dice overlap between automated and manual segmentations from the internal and external test datasets were 0.9999 and 0.9957, which was superior to other publicly available methods. Among trainee readers, SDH detection was statistically improved using NCCTH with and without bone removal applied compared to standard NCCTH alone (P value <0.001). Additionally, 12/14 (86 %) of participating trainees self-reported improved detection of extra axial blood products with bone removal, and 13/14 (93 %) indicated that they would like to have access to NCCTH bone removal in the on-call setting. Deep learning segmentation-based NCCTH bone removal is rapid, accurate, and improves detection of SDH among trainee radiologists when used in combination with standard NCCTH. This study highlights the potential of bone removal for improving confidence and accuracy of SDH detection.

Application of a methodological framework for the development and multicenter validation of reliable artificial intelligence in embryo evaluation

BackgroundArtificial intelligence (AI) models analyzing embryo time-lapse images have been developed to predict the likelihood of pregnancy following in vitro fertilization (IVF). However, limited research exists on methods ensuring AI consistency and reliability in clinical settings during its development and validation process. We present a methodology for developing and validating an AI model across multiple datasets to demonstrate reliable performance in evaluating blastocyst-stage embryos.MethodsThis multicenter analysis utilizes time-lapse images, pregnancy outcomes, and morphologic annotations from embryos collected at 10 IVF clinics across 9 countries between 2018 and 2022. The four-step methodology for developing and evaluating the AI model include: (I) curating annotated datasets that represent the intended clinical use case; (II) developing and optimizing the AI model; (III) evaluating the AI’s performance by assessing its discriminative power and associations with pregnancy probability across variable data; and (IV) ensuring interpretability and explainability by correlating AI scores with relevant morphologic features of embryo quality. Three datasets were used: the training and validation dataset (n = 16,935 embryos), the blind test dataset (n = 1,708 embryos; 3 clinics), and the independent dataset (n = 7,445 embryos; 7 clinics) derived from previously unseen clinic cohorts.ResultsThe AI was designed as a deep learning classifier ranking embryos by score according to their likelihood of clinical pregnancy. Higher AI score brackets were associated with increased fetal heartbeat (FH) likelihood across all evaluated datasets, showing a trend of increasing odds ratios (OR). The highest OR was observed in the top G4 bracket (test dataset G4 score ≥ 7.5: OR 3.84; independent dataset G4 score ≥ 7.5: OR 4.01), while the lowest was in the G1 bracket (test dataset G1 score < 4.0: OR 0.40; independent dataset G1 score < 4.0: OR 0.45). AI score brackets G2, G3, and G4 displayed OR values above 1.0 (P < 0.05), indicating linear associations with FH likelihood. Average AI scores were consistently higher for FH-positive than for FH-negative embryos within each age subgroup. Positive correlations were also observed between AI scores and key morphologic parameters used to predict embryo quality.ConclusionsStrong AI performance across multiple datasets demonstrates the value of our four-step methodology in developing and validating the AI as a reliable adjunct to embryo evaluation.

A deep learning model for assistive decision-making during robot-aided rehabilitation therapies based on therapists’ demonstrations

BackgroundA promising approach to improving motor recovery during rehabilitation is the use of robotic rehabilitation devices. These robotic devices provide tools to monitor the patient’s recovery progress while providing highly standardized and intensive therapy. A major challenge in using these robotic devices is the ability to decide when to assist the user. In this context, we propose a Deep Learning-based solution that can learn from a therapist’s criteria when a patient needs assistance during robot-aided rehabilitation therapy.MethodsAn experimental session was conducted with diverse people who suffered from neurological conditions. The participants used an upper limb rehabilitation robot to play a point-to-point game. A therapist supervised the robot-aided rehabilitation exercises and assisted the participants when considered necessary. This assistance provided by the therapist was detected to label those trajectories that were assisted to train a Deep Learning model that learns from the therapist when to assist. A series of transformations have been applied to the trajectories performed by the participants to generalize the method. Furthermore, the trajectory data was divided into sequences to be introduced to the model and continuously infer whether the user needs assistance. The data acquired during the experimental sessions have been divided into two datasets to train and evaluate the model: intra-participants (80% training, 20% validation) and test participants. The architecture of the Deep Learning model is conceived to perform time-series classification. It consists of diverse one-dimensional convolutional layers, a convolutional attention mechanism, and a Global Average Pooling layer. In addition, the output layer has one neuron with the sigmoid activation function, whose output can be interpreted as a probability of assistance. The model proposed in this study has been evaluated according to different metrics. In addition, the impact of applying fine-tuning to adapt the assistance to each user has been evaluated with the test participants.ResultsThe proposed model achieved an accuracy of 91.39% and an F1-Score of 75.15% with the validation dataset during a sequence-to-sequence evaluation, surpassing other state-of-the-art architectures. When evaluating the trajectories collected in the test dataset, the method proposed achieved an accuracy of 76.09% and an F1-Score of 74.42% after applying fine-tuning to each participant.ConclusionsThe results achieved by our Deep Learning-based method show the feasibility of learning assistance decision-making from experimented therapists. Furthermore, fine-tuning can be applied to personalize the assistance to each user and improve the accuracy of the method presented when deciding whether to assist with the rehabilitation robot.

DeepLabv3 + method for detecting and segmenting apical lesions on panoramic radiography

ObjectiveThis study aimed to apply the DeepLabv3 + model and compare it with the U-Net model in terms of detecting and segmenting apical lesions on panoramic radiography.Methods260 panoramic images that contain apical lesions in different regions were collected and randomly divided into training and test datasets. All images were manually annotated for apical lesions using Computer Vision Annotation Tool software by two independent dental radiologists and a master reviewer. The DeepLabv3 + model, one of the state-of-the-art deep semantic segmentation models, was utilized using Python programming language and the TensorFlow library and applied to the prepared datasets. The model was compared with the U-Net model applied to apical lesions and other medical image segmentation problems in the literature.ResultsThe DeepLabv3 + and U-Net models were applied to the same datasets with the same hyper-parameters. The AUC and recall results of the DeepLabv3 + were 29.96% and 61.06% better than the U-Net model. However, the U-Net model gets 69.17% and 25.55% better precision and F1-score results than the DeepLabv3 + model. The difference in the IoU results of the models was not statistically significant.ConclusionsThis paper comprehensively evaluated the DeepLabv3 + model and compared it with the U-Net model. Our experimental findings indicated that DeepLabv3 + outperforms the U-Net model by a substantial margin for both AUC and recall metrics. According to those results, for detecting apical lesions, we encourage researchers to use and improve the DeepLabv3 + model.Clinical relevanceThe DeepLabv3 + model has the poten tial to improve clinical diagnosis and treatment planning and save time in the clinic.

Improving permeability modeling using artificial neural networks in Lower Safa sand reservoir, JG field, Abu El-Gharadig basin, Western Desert, Egypt

Predicting permeability along and between wells to obtain the 3D spatial distribution is challenging in reservoir modeling. Permeability measurements using core sampling are time-consuming and accord too limited subsurface coverage. This study presents a systematic approach based on artificial neural networks (ANNs) to integrate core data and well logs to overcome these challenges and construct a robust 3D permeability model for the Lower Safa sand reservoir, JG field, Abu El-Gharadig basin, Egypt. The routine core analysis was performed to determine the reservoir quality and heterogeneity. Core-log depth match was performed, and the influential well logs, including effective porosity, bulk density, and deep resistivity logs, were selected to build the permeability model. ANN is applied to integrate the core permeability with the logs to predict permeability logs in the two studied wells. The dataset is randomly split into training, validation, and testing sets to evaluate the model's performance, and different hyperparameters are tuned to improve the prediction accuracy. A detailed comparison between the conventional method and the proposed ANN approach is provided in this study. Results demonstrate that the proposed ANN model was able to improve the permeability prediction performance by capturing the complex relationships between permeability and well logging data. The model achieves outstanding performance, where the coefficient of determination (R2) between the predicted permeability and core permeability is 0.90, 0.91, and 0.88 for the training, validation, and testing datasets, respectively. The developed model was used to predict the permeability logs along the reservoir intervals of the two studied wells. Ultimately, the Sequential Gaussian Simulation algorithm was used to populate the predicted logs in 3D. The outcomes of the study will aid users of deep learning to make informed choices on the appropriate ANN models to use in clastic reservoir characterization for more accurate permeability prediction with limited available data.

ANN-assisted comprehensive screening of silica gel-alunite composite sorbent system for efficient adsorption of toxic nickel ions: Batch and continuous mode water treatment applications.

Through batch and fixed-bed column operations, nickel ions were extracted from a contaminated aqueous media by adsorption onto silica gel-immobilized alunite (Sg@Aln). A three-layer backward-propagating network with an ideal pattern of 5-10-1 and 4-10-1 was used to train and validate an artificial neural network (ANN) model for process modeling and optimization in batch and continuous systems, respectively. For the test dataset, the model outputs of the model pointed out a satisfactory alignment between the anticipated and experimental response. The Sg@Aln dosage and contact time were recorded as the most relevant parameters in Ni2+ elimination. The Sg@Aln-metal interactions were also characterized using a variety of instrumental approaches. The maximum Ni2+ adsorption was achieved by utilizing 2g/L of the adsorbent at a solution pH of 5.0 after 10min of contact time, equating to 89.11%. The data corresponded well with the non-linear shape of the Langmuir isotherm (R2=0.99), and the computed maximal adsorption capacity was 96.01mg/g (1.64×10-3mol/g) at 25°C. Kinetic analysis reveals that the adsorption process is consistent with the pseudo-second-order model, with R2=0.9998. Thermodynamic findings indicated endothermicity, spontaneity, and adsorption favorability. Sg@Aln could remove 41.23mg/g and 33.20mg/g of Ni2+ from actual wastewater in batch and continuous processes, respectively. While the Sg@Aln column's breakthrough curve is consistent with Chu's simplistic model, the breakthrough capacity was 69.35mg/g. Overall, the results might open new possibilities for treating metal pollution in the aquatic environment.

Neuro-computational simulation of blood flow loaded with gold and maghemite nanoparticles inside an electromagnetic microchannel under rapid and unexpected change in pressure gradient

ABSTRACT In cardiovascular research, electromagnetic fields generated by Riga plates are utilized to study or manipulate blood flow dynamics, which is particularly crucial in developing treatments for conditions such as arterial plaque deposition and understanding blood behavior under varied flow conditions. This research predicts the flow patterns of blood enhanced with gold and maghemite nanoparticles (gold-maghemite/blood) in an electromagnetic microchannel influenced by Riga plates with a temperature gradient that decays exponentially, under sudden changes in pressure gradient. The flow modeling includes key physical influences like radiation heat emission and Darcy drag forces in porous media, with the flow mathematically represented through unsteady partial differential equations solved using the Laplace transform (LT) method. Results, including shear stress (SS) and rate of heat transfer (RHT), are graphically detailed, demonstrating changes in blood velocity profile with modifications in the Hartmann number and the width of electrodes, and differences in temperature and RHT between hybrid nano-blood (HNB) and nano-blood (NB). The key results indicate an increase in blood velocity distribution with higher modified Hartmann number, and a decrease with wider electrodes. Temperature is elevated in both hybrid nano-blood (HNB) and nano-blood (NB). Notably, HNB with gold and maghemite enhances heat transmission in the flow. Furthermore, an artificial intelligence-driven methodology employing an artificial neural network (ANN) has been incorporated to facilitate rapid and precise evaluations of SS and RHT, demonstrating remarkable predictive accuracy. The proposed algorithm exhibits outstanding accuracy, achieving 99.998% on the testing dataset and 96.843% during cross-validation for predicting SS, and 100% on the testing dataset, and 95.008% during cross-validation for predicting RHT. The implementation of nanotechnology with artificial intelligence promises new tools for doctors and surgeons, potentially transforming patient care in fields such as oncology, cardiology, and radiology. This model also facilitates the generation of precise electromagnetic fields to guide drug-loaded magnetic nanoparticles for applications in targeted drug delivery, hyperthermia treatment, MRI contrast enhancement, blood flow monitoring, cancer treatment, and controlled drug release.

Artificial intelligence for segmentation and classification in lumbar spinal stenosis: an overview of current methods.

Lumbar spinal stenosis (LSS) is a frequently occurring condition defined by narrowing of the spinal or nerve root canal due to degenerative changes. Physicians use MRI scans to determine the severity of stenosis, occasionally complementing it with X-ray or CT scans during the diagnostic work-up. However, manual grading of stenosis is time-consuming and induces inter-reader variability as a standardized grading system is lacking. Machine Learning (ML) has the potential to aid physicians in this process by automating segmentation and classification of LSS. However, it is unclear what models currently exist to perform these tasks. A systematic review of literature was performed by searching the Cochrane Library, Embase, Emcare, PubMed, and Web of Science databases for studies describing an ML-based algorithm to perform segmentation or classification of the lumbar spine for LSS. Risk of bias was assessed through an adjusted version of the Newcastle-Ottawa Quality Assessment Scale that was more applicable to ML studies. Qualitative analyses were performed based on type of algorithm (conventional ML or Deep Learning (DL)) and task (segmentation or classification). A total of 27 articles were included of which nine on segmentation, 16 on classification and 2 on both tasks. The majority of studies focused on algorithms for MRI analysis. There was wide variety among the outcome measures used to express model performance. Overall, ML algorithms are able to perform segmentation and classification tasks excellently. DL methods tend to demonstrate better performance than conventional ML models. For segmentation the best performing DL models were U-Net based. For classification U-Net and unspecified CNNs powered the models that performed the best for the majority of outcome metrics. The number of models with external validation was limited. DL models achieve excellent performance for segmentation and classification tasks for LSS, outperforming conventional ML algorithms. However, comparisons between studies are challenging due to the variety in outcome measures and test datasets. Future studies should focus on the segmentation task using DL models and utilize a standardized set of outcome measures and publicly available test dataset to express model performance. In addition, these models need to be externally validated to assess generalizability.

Quality Grading of Oudemansiella raphanipes Using Three-Teacher Knowledge Distillation with Cascaded Structure for LightWeight Neural Networks

Oudemansiella raphanipes is valued for its rich nutritional content and medicinal properties, but traditional manual grading methods are time-consuming and labor-intensive. To address this, deep learning techniques are employed to automate the grading process, and knowledge distillation (KD) is used to enhance the accuracy of a small-parameter model while maintaining a low resource occupation and fast response speed in resource-limited devices. This study employs a three-teacher KD framework and investigates three cascaded structures: the parallel model, the standard series model, and the series model with residual connections (residual-series model). The student model used is a lightweight ShuffleNet V2 0.5x, while the teacher models are VGG16, ResNet50, and Xception. Our experiments show that the cascaded structures result in improved performance indices, compared with the traditional ensemble model with equal weights; in particular, the residual-series model outperforms the other models, achieving a grading accuracy of 99.7% on the testing dataset with an average inference time of 5.51 ms. The findings of this study have the potential for broader application of KD in resource-limited environments for automated quality grading.

Downscaled gridded global dataset for gross domestic product (GDP) per capita PPP over 1990-2022.

We present a comprehensive gridded GDP per capita dataset downscaled to the admin 2 level (43,501 units) covering 1990-2022. It updates existing outdated datasets, which use reported subnational data only up to 2010. Our dataset, which is based on reported subnational GDP per capita data from 89 countries and 2,708 administrative units, employs various novel methods for extrapolation and downscaling. Downscaling with machine learning algorithms showed high performance (R2 = 0.79 for cross-validation, R2 = 0.80 for the test dataset) and accuracy against reported datasets (Pearson R = 0.88). The dataset includes reported and downscaled annual data (1990-2022) for three administrative levels: 0 (national; reported data for 237 administrative units), 1 (provincial; reported data for 2,708 administrative units for 89 countries), and 2 (municipality; downscaled data for 43,501 administrative units). The dataset has a higher spatial resolution and wider temporal range than the existing data do and will thus contribute to global or regional spatial analyses such as socioenvironmental modelling and economic resilience evaluation. The data are available at https://doi.org/10.5281/zenodo.10976733 .

Large blood vessel segmentation in quantitative DCE-MRI of brain tumors: A Swin UNETR approach.

Brain tumor growth is associated with angiogenesis, wherein the density of newly developed blood vessels indicates tumor progression and correlates with the tumor grade. Quantitative dynamic contrast enhanced-magnetic resonance imaging (DCE-MRI) has shown potential in brain tumor grading and treatment response assessment. Segmentation of large-blood-vessels is crucial for automatic and accurate tumor grading using quantitative DCE-MRI. Traditional manual and semi-manual rule-based large-blood-vessel segmentation methods are time-intensive and prone to errors. This study proposes a novel deep learning-based technique for automatic large-blood-vessel segmentation using Swin UNETR architectures and comparing it with U-Net and Attention U-Net architectures. The study employed MRI data from 187 brain tumor patients, with training, validation, and testing datasets sourced from two centers, two vendors, and two field-strength magnetic resonance scanners. To test the generalizability of the developed model, testing was also carried out on different brain tumor types, including lymphoma and metastasis. Performance evaluation demonstrated that Swin UNETR outperformed other models in segmenting large-blood-vessel regions (achieving Dice scores of 0.979, and 0.973 on training and validation sets, respectively, with test set performance ranging from 0.835 to 0.982). Moreover, most quantitative parameters showed significant differences (p < 0.05) between with and without large-blood-vessel. After large-blood-vessel removal, using both ground truth and predicted masks, the values of parameters in non-vascular tumoral regions were statistically similar (p > 0.05). The proposed approach has potential applications in improving the accuracy of automatic grading of tumors as well as in treatment planning.

Determining the status of tertiary lymphoid structures in invasive pulmonary adenocarcinoma based on chest CT radiomic features

ObjectivesThe aim of this study was to determine the status of tertiary lymphoid structures (TLSs) using radiomic features in patients with invasive pulmonary adenocarcinoma (IA).MethodsIn this retrospective study, patients with IA from November 2015 to March 2024 were recruited from two independent centers (center 1, training and internal test data set; center 2, external test data set). TLS was divided into two groups according to hematoxylin-eosin staining. Radiomic features were extracted, and support vector machine (SVM) were implemented to predict the status of TLSs. Receiver operating characteristic (ROC) curves were used to analyze diagnostic performance. Furthermore, visual assessments of the test set were also conducted by two thoracic radiologists and compared with the radiomics results.ResultsA total of 456 patients were included (training data set, n = 278; internal test data set, n = 115; external test data set, n = 63). The area under the curve (AUC) of the radiomics model on the validation set, the internal test set, and the external test set were 0.781 (95% confidence interval (CI): 0.659–0.905;), 0.804 (95% CI: 0.723–0.884;) and 0.747 (95% CI: 0.621–0.874;), respectively. In the visual assessments, the mean CT value and air bronchogram were important indicators of TLS, the AUC was 0.683. In the external test set, the AUC of the clinical model was 0.632.ConclusionsThe radiomics model has a higher AUC than the clinical model and effectively discriminates TLSs in patients with IA.Critical relevance statementThis study demonstrates that the radiomics-based model can differentiate TLSs in patients with IA. As a non-invasive biomarker, it enhances our understanding of tumor prognosis and management.Key PointsTLSs are closely related to favorable clinical outcomes in non-small cell lung cancer.Radiomics from Chest CT predicted TLSs in patients with IA.This study supports individualized clinical decision-making for patients with IA.Graphical