Deep Feature Extraction Research Articles

A deep learning LSTM-based approach for AMD classification using OCT images

Age-related macular degeneration (AMD) is an age-related, persistent, painless eye disease that impairs central vision. The central area (macula) of the retina, located at the back of the eye, sustains damage that is the cause of loss of vision. The early detection of AMD can increase the probability of treatment and prevent vision loss. The AMD can be classified into dry and wet AMD based on the absence of neovascularization. This study introduces a new methodology for the classification of AMD using optical coherence tomography (OCT) retinal images. The proposed methodology is based on three stages. The first stage is the data preparation stage for resizing and normalizing the used images. The second stage is the image processing stage for enhancing the image quality as contrast and resolution these enhancements have been checked by the weighted peak signal-to-noise ratio (WPSNR) methodology. The third stage is the deep feature extraction and classification stage, which consists of two sub-models. The first model is MobileNet V1 which has been used as a deep feature extractor. The second model is LSTM (long short-term memory), fed with deep features to classify the AMD stages. A multi-classification with six separate trials has been employed with the proposed methodology, and compared with other models like DenseNet201 and InceptionV3. The proposed model has been tested on a sample of benchmark data with 4005 grayscale images labeled into three classes. The proposed methodology has achieved an accuracy of 98.85%, a sensitivity of 99.09%, and a specificity of 99.1%. To ensure the effectiveness of the proposed methodology, a comparative analysis has been established with previous approaches in the related field, and the results demonstrated the superiority of the proposed system in AMD multi-classification.

Automated Detection of Plant Diseases and Pests Using Advanced Image Processing and Neural Networks

Pests and viral plant diseases have proliferated in agriculture. Rats spread plant diseases. Hard to detect viral infections or insect infestations that impair plant growth and yield. It decreases crop size, quality, and marketability, incurring major economic losses. Strong, disease-resistant crops are needed for agriculture. Bacterial, fungal, and insect/pest diseases destroy 70% and 15% of crops. Above 85% of disease severity destroys crops. Early detection systems track pests and diseases. High-definition equipment is costly for rural farmers. Correct diagnosis permits quick infection severity and type assessment and control. In precision agriculture, neural networks and image processing eliminate human error. Sustainable agriculture and organic agricultural cultivation must start early. The proposed system classifies diseases and pests for plant safety. A healthy, fast-growing plantation is our goal. Early detection of viral diseases and insect infestations boosts crop productivity. Plant disease diagnostics focuses on leaves and buds. Stem and insect infections are covered. The suggested method classifies plant health and disease automatically. Peanut leaves develop Bacterial Blight from contagious illnesses like Cercospora Leaf Peanut. Boll Weevils, European Corn Borers, and Fall Army Worms attack cotton. The conditions are categorised. Image accuracy depends on lighting, resolution, location, and backdrop complexity. Various low-resolution pictures are enriched to better classification. Complex leaf intensity variations cause the system to misdiagnose illness or insect infestation. To avoid misclassification, the system segments the sample image using various methods to find lesions. Multilevel Segmentation (MS) improves Deep neural network feature extraction using approximate outlines. This method protects important pixels, eliminates unneeded pixels, and amplifies the lesion to control segmentation and reduce insufficient and excessive segmentation.

SuPrNet: Super Proxy for 4D occupancy forecasting

Spacetime (4D) occupancy forecasting explicitly models the future development of self-driving vehicles (SDVs) themselves and their surrounding environment, which is highly beneficial for their downstream tasks. Currently, using point cloud forecasting as a proxy to self-supervisedly optimize the 4D occupancy forecasting framework has significant application prospects due to its independence from labeled data. However, existing methods have the following shortcomings: (1) Insufficient mining of spatiotemporal information in the scene. (2) Neglecting the extraction of regional deep features. (3) Inadequate learning ability for spatiotemporal predictive learning. Based on this, we propose SuPrNet, constructing a ”Super Proxy” to address the aforementioned issues. Specifically, SuPrNet first embeds the spatial and temporal information of the current frame into point-wise features through a 4D hybrid representation combined with multi-planar and grid features (4DPG). Then, a Hierarchical Encoder–Decoder Module (HED) with sparsity consideration is used to structure the point cloud and extract deep features. Thanks to the structured feature map representation, SuPrNet proceeds to use a spatiotemporal predictive learning module (STL) to implicitly predict future feature maps. Finally, we decode these feature maps to a specific size as the forecasting of future occupancy. The 4DPG, HED, and STL in the model are designed to address the aforementioned issues, and experiments on Nuscene and KITTI-Odometry show that our method achieves the best 4D occupancy forecasting and point cloud forecasting accuracy, surpassing previous state-of-the-art models by approximately 30%. Additionally, we have expanded and validated the application prospects of SuPrNet through various downstream tasks. Our project is available at https://github.com/AlanLiangC/4DPCF.

A hybrid reliability assessment method based on health index construction and reliability modeling for rolling bearing

AbstractThe assessment of rolling bearing reliability is vital for ensuring mechanical operational safety and minimizing maintenance costs. Due to the difficulty in obtaining data on the performance degradation and failure time of rolling bearings, traditional methods for reliability assessment are challenged. This paper introduces a novel hybrid method for the reliability assessment of rolling bearings, combining the convolutional neural network (CNN)‐convolutional block attention module (CBAM)‐ bidirectional long short‐term memory (BiLSTM) network with the Wiener process. The approach comprises three distinct stages: Initially, it involves acquiring two‐dimensional time‐frequency representations of bearings at various operational phases using Continuous Wavelet Transform. Subsequently, the CNN‐CBAM‐BiLSTM network is employed to establish health index (HI) for the bearings and to facilitate the extraction of deep features, serving as input for the Wiener process. The final stage applies the Wiener process to evaluate the bearings’ reliability, characterizing the HI and quantifying uncertainties. The experiment is performed on bearing degradation data and the results indicate the effectiveness and superiority of the proposed hybrid method.

Deciphering Optimal Radar Ensemble for Advancing Sleep Posture Prediction through Multiview Convolutional Neural Network (MVCNN) Approach Using Spatial Radio Echo Map (SREM).

Assessing sleep posture, a critical component in sleep tests, is crucial for understanding an individual's sleep quality and identifying potential sleep disorders. However, monitoring sleep posture has traditionally posed significant challenges due to factors such as low light conditions and obstructions like blankets. The use of radar technolsogy could be a potential solution. The objective of this study is to identify the optimal quantity and placement of radar sensors to achieve accurate sleep posture estimation. We invited 70 participants to assume nine different sleep postures under blankets of varying thicknesses. This was conducted in a setting equipped with a baseline of eight radars-three positioned at the headboard and five along the side. We proposed a novel technique for generating radar maps, Spatial Radio Echo Map (SREM), designed specifically for data fusion across multiple radars. Sleep posture estimation was conducted using a Multiview Convolutional Neural Network (MVCNN), which serves as the overarching framework for the comparative evaluation of various deep feature extractors, including ResNet-50, EfficientNet-50, DenseNet-121, PHResNet-50, Attention-50, and Swin Transformer. Among these, DenseNet-121 achieved the highest accuracy, scoring 0.534 and 0.804 for nine-class coarse- and four-class fine-grained classification, respectively. This led to further analysis on the optimal ensemble of radars. For the radars positioned at the head, a single left-located radar proved both essential and sufficient, achieving an accuracy of 0.809. When only one central head radar was used, omitting the central side radar and retaining only the three upper-body radars resulted in accuracies of 0.779 and 0.753, respectively. This study established the foundation for determining the optimal sensor configuration in this application, while also exploring the trade-offs between accuracy and the use of fewer sensors.

Framework for psoriasis/molluscum detection in skin images using ResNetV2 variants

BackgroundComputer algorithms are extensively utilized in hospitals for the purpose of achieving expedited and precise disease identification from medical data. The objective of this study is to propose a deep-learning (DL) framework to classify the chosen digital skin image (DSI) database into psoriasis and molluscum with better accuracy. Aims and objectivesAims to develop a DL-tool using the pre-trained ResNetV2 DL-models and verify the performance of the developed tool using a chosen DSI database. Further, confirm the merit of the ResNetV2-based tool against other chosen DL-models. Methods and resultsThis study initially examines the performance of chosen pre-trained DL (PDL) methods using the DSI database using conventional and fused features. Proposed DL-tool consist the following stages; (i) image collection and resizing it 224x224x3 pixels, (ii) deep-features extraction using the selected PDL, (iii) feature reduction and serial features concatenation to get a new features vector, and (iv) the performance evaluation through three-fold cross validation and confirmation. The feature reduction and serial features fusion of this study is performed initially using feature sorting based on its rank and 50% dropout and Particle Swarm Optimization (PSO) based feature reduction. The reduced features of two chosen PDL-models are then considered to obtain the Fused Deep Features (FDF) and the classification task is executed on the DSI data to verify the performance of the developed DL-tool. The experimental outcome of this study confirms that the proposed scheme helps to provide a detection accuracy of >97%, when the K-Nearest Neighbour based classification is executed. ConclusionInvestigational outcome of this study confirms that the proposed DL-tool helps to achieve better detection accuracy when the features of ResNetV2 models are considered to generate the FDF. Further, the accuracy achieved with the PSO based FDF is better compared to the conventional method generated FDF.

A General Image Super-Resolution Reconstruction Technique for Walnut Object Detection Model

Object detection models are commonly used in yield estimation processes in intelligent walnut production. The accuracy of these models in capturing walnut features largely depends on the quality of the input images. Without changing the existing image acquisition devices, this study proposes a super-resolution reconstruction module for drone-acquired walnut images, named Walnut-SR, to enhance the detailed features of walnut fruits in images, thereby improving the detection accuracy of the object detection model. In Walnut-SR, a deep feature extraction backbone network called MDAARB (multilevel depth adaptive attention residual block) is designed to capture multiscale information through multilevel channel connections. Additionally, Walnut-SR incorporates an RRDB (residual-in-residual dense block) branch, enabling the module to focus on important feature information and reconstruct images with rich details. Finally, the CBAM (convolutional block attention module) attention mechanism is integrated into the shallow feature extraction residual branch to mitigate noise in shallow features. In 2× and 4× reconstruction experiments, objective evaluation results show that the PSNR and SSIM for 2× and 4× reconstruction reached 24.66 dB and 0.8031, and 19.26 dB and 0.4991, respectively. Subjective evaluation results indicate that Walnut-SR can reconstruct images with richer detail information and clearer texture features. Comparative experimental results of the integrated Walnut-SR module show significant improvements in mAP50 and mAP50:95 for object detection models compared to detection results using the original low-resolution images.

Preoperative prediction of microvascular invasion risk in hepatocellular carcinoma with MRI: peritumoral versus tumor region

ObjectivesTo explore the predictive performance of tumor and multiple peritumoral regions on dynamic contrast-enhanced magnetic resonance imaging (MRI), to identify optimal regions of interest for developing a preoperative predictive model for the grade of microvascular invasion (MVI).MethodsA total of 147 patients who were surgically diagnosed with hepatocellular carcinoma, and had a maximum tumor diameter ≤ 5 cm were recruited and subsequently divided into a training set (n = 117) and a testing set (n = 30) based on the date of surgery. We utilized a pre-trained AlexNet to extract deep learning features from seven different regions of the maximum transverse cross-section of tumors in various MRI sequence images. Subsequently, an extreme gradient boosting (XGBoost) classifier was employed to construct the MVI grade prediction model, with evaluation based on the area under the curve (AUC).ResultsThe XGBoost classifier trained with data from the 20-mm peritumoral region showed superior AUC compared to the tumor region alone. AUC values consistently increased when utilizing data from 5-mm, 10-mm, and 20-mm peritumoral regions. Combining arterial and delayed-phase data yielded the highest predictive performance, with micro- and macro-average AUCs of 0.78 and 0.74, respectively. Integration of clinical data further improved AUCs values to 0.83 and 0.80.ConclusionCompared with those of the tumor region, the deep learning features of the peritumoral region provide more important information for predicting the grade of MVI. Combining the tumor region and the 20-mm peritumoral region resulted in a relatively ideal and accurate region within which the grade of MVI can be predicted.Clinical relevance statementThe 20-mm peritumoral region holds more significance than the tumor region in predicting MVI grade. Deep learning features can indirectly predict MVI by extracting information from the tumor region and directly capturing MVI information from the peritumoral region.Key PointsWe investigated tumor and different peritumoral regions, as well as their fusion.MVI predominantly occurs in the peritumoral region, a superior predictor compared to the tumor region.The peritumoral 20 mm region is reasonable for accurately predicting the three-grade MVI.Graphical

Optimizing breast cancer diagnosis: combining hybrid architectures through Apache Spark

Early detection and diagnosis of breast cancer are critical for saving lives. This paper addresses two major challenges associated with this task: the vast amount of data processing involved and the need for early detection of breast cancer. To tackle these issues, we developed thirty hybrid architectures by combining five deep learning techniques (Xception, Inception-V3, ResNet50, VGG16, VGG19) as feature extractors and six classifiers (random forest, logistic regression, naive Bayes, gradient-boosted tree, decision tree, and support vector machine) implemented on the Spark framework. We evaluated the performance of these architectures using four classification criteria. The results, analyzed using Scott Knott's statistical test, demonstrated the effectiveness of merging deep learning feature extraction techniques with traditional classifiers for classifying breast cancer into malignant and benign tumors. Notably, the hybrid architecture using logistic regression as the classifier and ResNet50 for feature extraction (RESLR) emerged as the top performer. It achieved impressive accuracy scores of 98.20%, 96.59%, 96.64%, and 94.84% across the Break-His dataset at different magnifications (40X, 100X, 200X, and 400X) respectively. Additionally, RESLR achieved an accuracy of 97.05% on the ICIAR dataset and a remarkable accuracy of 95.31% on the FNAC dataset.

Non-Contact Blood Pressure Estimation From Radar Signals by a Stacked Deformable Convolution Network.

This study introduces a contactless blood pressure monitoring approach that combines conventional radar signal processing with novel deep learning architectures. During the preprocessing phase, datasets suitable for synchronization are created by integrating Kalman filtering, multiscale bandpass filters, and a periodic extraction method in the time domain. These data comprise data on chest micro variations, encapsulating a complex array of physiological and biomedical information reflective of cardiac micromotions. The Radar-based Stacked Deformable convolution Network (RSD-Net) integrates channel and spatial self attention mechanisms within a deformable convolutional framework to enhance feature extraction from radar signals. The network architecture systematically employs deformable convolutions for initial deep feature extraction from individual signals. Subsequently, continuous blood pressure estimation is conducted using self attention mechanisms on feature map from single source coupled with multi-feature map channel attention. The performance of model is corroborated via the open-source dataset procured using a non-invasive 24 GHz six-port continuous wave radar system. The dataset, encompassing readings from 30 healthy individuals subjected to diverse conditions including rest, the Valsalva maneuver, apnea, and tilt-table examinations. It serves to substantiate the validity and resilience of the proposed method in the non-contact assessment of continuous blood pressure. Evaluation metrics reveal Pearson correlation coefficients of 0.838 for systolic and 0.797 for diastolic blood pressure predictions. The Mean Error (ME) and Standard Deviation (SD) for systolic and diastolic blood pressure measurements are -0.32 ±6.14 mmHg and -0.20 ±5.50 mmHg, respectively. The ablation study assesses the contribution of different structural components of the RSD-Net, validating their significance in the overall of model performance.

Deep Double Incomplete Multi-View Multi-Label Learning With Incomplete Labels and Missing Views.

View missing and label missing are two challenging problems in the applications of multi-view multi-label classification scenery. In the past years, many efforts have been made to address the incomplete multi-view learning or incomplete multi-label learning problem. However, few works can simultaneously handle the challenging case with both the incomplete issues. In this article, we propose a new incomplete multi-view multi-label learning network to address this challenging issue. The proposed method is composed of four major parts: view-specific deep feature extraction network, weighted representation fusion module, classification module, and view-specific deep decoder network. By, respectively, integrating the view missing information and label missing information into the weighted fusion module and classification module, the proposed method can effectively reduce the negative influence caused by two such incomplete issues and sufficiently explore the available data and label information to obtain the most discriminative feature extractor and classifier. Furthermore, our method can be trained in both supervised and semi-supervised manners, which has important implications for flexible deployment. Experimental results on five benchmarks in supervised and semi-supervised cases demonstrate that the proposed method can greatly enhance the classification performance on the difficult incomplete multi-view multi-label classification tasks with missing labels and missing views.

A two-stream decision fusion network for cervical pap-smear image classification tasks

Deep learning, especially Convolution Neural Networks (CNNs), has demonstrated superior performance in image recognition and classification tasks. They make complex pattern recognition possible by extracting image features through layers of abstraction. However, despite the excellent performance of deep learning in general image classification, its limitations are becoming apparent in specific domains such as cervical cell medical image classification. This is because although the morphology of cervical cells varies between normal, diseased and cancerous, these differences are sometimes very small and difficult to capture. To solve this problem, we propose a two-stream feature fusion model comprising a manual feature branch, a deep feature branch, and a decision fusion module. Specifically, We process cervical cells through a modified DarkNet backbone network to extract deep features. In order to enhance the learning of deep features, we have devised scale convolution blocks to substitute the original convolution, termed Basic convolution blocks. The manual feature branch comprises a range of traditional features and is linked to a multilayer perceptron. Additionally, we design three decision feature channels trained from both manual and deep features to enhance the model performance in cervical cell classification. Our proposed model demonstrates superior performance when compared to state-of-the-art cervical cell classification models. We establish a 15-category 148762 cervical cytopathology image dataset (CCID). In addition, we additionally conducted experiments on the SIPaKMeD dataset. Numerous experiments show that our proposed model performs excellently compared to state-of-the-art classification models. The outcomes illustrate that our approach can significantly aid pathologists in accurately evaluating cervical smears.

A cloud–edge collaboration based quality-related hierarchical fault detection framework for large-scale manufacturing processes

Against the backdrop of the new-generation intelligent manufacturing and Industrial Internet of Things, manufacturing processes are evolving towards integration, large-scale operations, and complexity, and the requirements for process safety and product quality reliability are constantly improving. However, these processes are generally distinguished by multi-production sub-processes with cooperative coupling, and multi-hierarchical levels integrated automation systems, where any deviation or abnormality occurring within a sub-process or system level will lead to the extensive propagation and evolution of faults, making the comprehensive quality-related fault detection face increasing challenges in practice. In response, a novel cloud–edge collaboration-based quality-related hierarchical fault detection framework is constructed in this paper, which can facilitate interconnected communication between sub-processes and enhance collaborative interactions at different hierarchical levels to improve fault detection accuracy and reliability. First, two quality supervised training mechanisms based on minimal gated unit are introduced for capturing different scale dynamics features on the cloud side and edge side. These methods enable more adequate extraction of quality-related deep hidden features, particularly in the presence of nonlinearity and dynamics. Subsequently, a federated bi-directional knowledge distillation-based strategy is proposed, leveraging the concept of federated learning. Quality knowledge such as prediction and quality-related features from both the cloud side and edge side are integrated into the training steps of federated learning, thereby strengthening the interaction between multiple sub-processes and distinct levels. Furthermore, comprehensive quality-related fault detection strategies are formulated based on the quality-related features of the cloud side and each sub-process on the edge side, respectively. Finally, the proposed framework is deployed into a cloud–edge-device collaborative prototype system based on a real hot strip mill process to demonstrate its effectiveness and applicability.

Hyperspectral image classification using a deep relation network with random replacement data augmentation

ABSTRACT Achieving high classification accuracy under a limited number of training samples remains a key challenge in hyperspectral image (HSI) classification. To address this issue, we propose a random replacement data augmentation method. The proposed method randomly selects a rectangular spatial region in HSI, and the pixels in the selected region are replaced with data from different spectral dimensions randomly to generate new sample, increasing the diversity of training samples effectively. In addition, we also propose an improved deep metric learning-based relation network, which improves model’s ability to extract deep features and fully utilizes spatial information to establish relationships between feature maps and categories effectively. Experiments have shown that the proposed method can achieve more competitive results.

YOLOv7-Branch: A Jujube Leaf Branch Detection Model for Agricultural Robot.

The intelligent harvesting technology for jujube leaf branches presents a novel avenue for enhancing both the quantity and quality of jujube leaf tea, whereas the precise detection technology for jujube leaf branches emerges as a pivotal factor constraining its development. The precise identification and localization of jujube leaf branches using real-time object detection technology are crucial steps toward achieving intelligent harvesting. When integrated into real-world scenarios, issues such as the background noise introduced by tags, occlusions, and variations in jujube leaf morphology constrain the accuracy of detection and the precision of localization. To address these issues, we describe a jujube leaf branch object detection network based on YOLOv7. First, the Polarized Self-Attention module is embedded into the convolutional layer, and the Gather-Excite module is embedded into the concat layer to incorporate spatial information, thus achieving the suppression of irrelevant information such as background noise. Second, we incorporate implicit knowledge into the Efficient Decoupled Head and replace the original detection head, enhancing the network's capability to extract deep features. Third, to address the issue of imbalanced jujube leaf samples, we employ Focal-EIoU as the bounding box loss function to expedite the regression prediction and enhance the localization accuracy of the model's bounding boxes. Experiments show that the precision of our model is 85%, which is increased by 3.5% compared to that of YOLOv7-tiny. The mAP@0.5 value is 83.7%. Our model's recognition rate, recall and mean average precision are superior to those of other models. Our method could provide technical support for yield estimation in the intelligent management of jujube orchards.

Multi-layer adaptive spatial-temporal feature fusion network for efficient food image recognition

Numerous deep learning methods have been developed to tackle the challenges of recognizing food images, including convolutional neural networks, deep feature extraction, and deep feature fusion methods. This research proposes a new architecture called ASTFF-Net that uses deep feature fusion to tackle various challenges in food recognition, including similarity patterns between two categories, multi-object problems, light conditions, camera position, noise objects, and blurred images. ASTFF-Net is a robust and adaptive spatial–temporal fusion network designed to address these challenges effectively. The ASTFF-Net architecture consisted of three networks. In the spatial feature extraction network, the ResNet50 architecture was used to extract robust spatial features, and the reduction operation was utilized to minimize parameter size. Subsequently, the spatial features were passed through a 1D convolution (Conv1D) to fit the features into the recurrent neural networks. In the temporal feature extraction network, the spatial features were given to the long short-term memory, allowing the network to learn from various long sequence patterns. In the adaptive feature fusion network, the robust spatial and temporal features were fused and assigned to the Conv1D, followed by the softmax function. The ASTFF-Net architecture is also intended to decrease the number of network parameters and prevent overfitting problems. Experimental results on four benchmark food image datasets: Food11, UEC Food-100, UEC Food-256, and ETH Food-101, demonstrate that the proposed ASTFF-Nets, particularly ASTFF-NetB3, were more competitive compared with other existing methods.

A Joint Network of Edge-Aware and Spectral-Spatial Feature Learning for Hyperspectral Image Classification.

Hyperspectral image (HSI) classification is a vital part of the HSI application field. Since HSIs contain rich spectral information, it is a major challenge to effectively extract deep representation features. In existing methods, although edge data augmentation is used to strengthen the edge representation, a large amount of high-frequency noise is also introduced at the edges. In addition, the importance of different spectra for classification decisions has not been emphasized. Responding to the above challenges, we propose an edge-aware and spectral-spatial feature learning network (ESSN). ESSN contains an edge feature augment block and a spectral-spatial feature extraction block. Firstly, in the edge feature augment block, the edges of the image are sensed, and the edge features of different spectral bands are adaptively strengthened. Then, in the spectral-spatial feature extraction block, the weights of different spectra are adaptively adjusted, and more comprehensive depth representation features are extracted on this basis. Extensive experiments on three publicly available hyperspectral datasets have been conducted, and the experimental results indicate that the proposed method has higher accuracy and immunity to interference compared to state-of-the-art (SOTA) method.

A self-supervised framework for computer-aided arrhythmia diagnosis

Cardiovascular diseases, including all types of arrhythmias, are the leading cause of death. Deep learning (DL)-based electrocardiography (ECG) diagnosis systems have attracted considerable attention in recent years. However, training DL-based models requires much high-quality labeled data, and labeling ECG records is time-consuming and expensive. In this paper, a high-precision deep network obtained via low-cost annotation, that is, a self-supervised residual convolutional neural network (SSRCNN), was designed. First, a convolutional neural network and residual blocks were designed to construct a deep feature extractor to automatically obtain the feature expression of ECG signals. Next, we developed time- and lead-dimension-based data augmentation methods and designed a pretraining framework based on unlabeled datasets such that the feature extractor could update the weights in the unlabeled samples. Furthermore, through interactions with clinicians, we used a few labeled data to fine-tune the pretrained model and feature classifier to automatically extract deep features and perform effective classification. Finally, we used different open-source datasets to validate the superiority of the SSRCNN. With the support of unlabeled datasets, SSRCNN can effectively reduce clinicians’ workload. Compared with existing methods, the SSRCNN achieves a better diagnostic performance in terms of average accuracy (99.0 %) and average F1-macro (86.83 %) using approximately 25 % of labeled data. Therefore, the SSRCNN has a potential for practical applications in clinical settings.

Machine learning for predicting cognitive decline within five years in Parkinson's disease: Comparing cognitive assessment scales with DAT SPECT and clinical biomarkers.

Parkinson's disease (PD) is an age-related neurodegenerative condition characterized mostly by motor symptoms. Although a wide range of non-motor symptoms (NMS) are frequently experienced by PD patients. One of the important and common NMS is cognitive impairment, which is measured using different cognitive scales. Monitoring cognitive impairment and its decline in PD is essential for patient care and management. In this study, our goal is to identify the most effective cognitive scale in predicting cognitive decline over a 5-year timeframe initializing clinical biomarkers and DAT SPECT. Machine Learning has previously shown superior performance in image and clinical data classification and detection. In this study, we propose to use machine learning with different types of data, such as DAT SPECT and clinical biomarkers, to predict PD-CD based on various cognitive scales. We collected 330 DAT SPECT images and their clinical data in baseline, years 2,3,4, and 5 from Parkinson's Progression Markers Initiative (PPMI). We then designed a 3D Autoencoder to extract deep radiomic features (DF) from DAT SPECT images, and we then concatenated it with 17 clinical features (CF) to predict cognitive decline based on Montreal Cognitive Assessment (MoCA) and The Movement Disorder Society-Unified Parkinson's Disease Rating Scale (MDS-UPDRS-I). The utilization of MoCA as a cognitive decline scale yielded better performance in various years compared to MDS-UPDRS-I. In year 4, the application of the deep radiomic feature resulted in the highest achievement, with a cross-validation AUC of 89.28, utilizing the gradient boosting classifier. For the MDS-UPDRS-I scale, the highest achievement was obtained by utilizing the deep radiomic feature, resulting in a cross-validation AUC of 81.34 with the random forest classifier. The study findings indicate that the MoCA scale may be a more effective predictor of cognitive decline within 5 years compared to MDS-UPDRS-I. Furthermore, deep radiomic features had better performance compared to sole clinical biomarkers or clinical and deep radiomic combined. These results suggest that using the MoCA score and deep radiomic features extracted from DAT SPECT could be a promising approach for identifying individuals at risk for cognitive decline in four years. Future research is needed to validate these findings and explore their utility in clinical practice.

A Novel Mamba Architecture with a Semantic Transformer for Efficient Real-Time Remote Sensing Semantic Segmentation

Real-time remote sensing segmentation technology is crucial for unmanned aerial vehicles (UAVs) in battlefield surveillance, land characterization observation, earthquake disaster assessment, etc., and can significantly enhance the application value of UAVs in military and civilian fields. To realize this potential, it is essential to develop real-time semantic segmentation methods that can be applied to resource-limited platforms, such as edge devices. The majority of mainstream real-time semantic segmentation methods rely on convolutional neural networks (CNNs) and transformers. However, CNNs cannot effectively capture long-range dependencies, while transformers have high computational complexity. This paper proposes a novel remote sensing Mamba architecture for real-time segmentation tasks in remote sensing, named RTMamba. Specifically, the backbone utilizes a Visual State-Space (VSS) block to extract deep features and maintains linear computational complexity, thereby capturing long-range contextual information. Additionally, a novel Inverted Triangle Pyramid Pooling (ITP) module is incorporated into the decoder. The ITP module can effectively filter redundant feature information and enhance the perception of objects and their boundaries in remote sensing images. Extensive experiments were conducted on three challenging aerial remote sensing segmentation benchmarks, including Vaihingen, Potsdam, and LoveDA. The results show that RTMamba achieves competitive performance advantages in terms of segmentation accuracy and inference speed compared to state-of-the-art CNN and transformer methods. To further validate the deployment potential of the model on embedded devices with limited resources, such as UAVs, we conducted tests on the Jetson AGX Orin edge device. The experimental results demonstrate that RTMamba achieves impressive real-time segmentation performance.