Jasem Almotiri

Neuropsychiatric disorders including attention-deficit/hyperactivity disorder (ADHD), bipolar disorder (BD), and schizophrenia (SZ) exhibit significant clinical heterogeneity and overlapping symptom profiles, which severely limits the diagnostic specificity of conventional symptom-based assessments. While structural magnetic resonance imaging (sMRI) provides objective insights into anatomical brain alterations, existing computational frameworks often rely on single-plane or binary patient-control analyses, lack anatomical interpretability, or risk data leakage through inconsistent validation protocols. To address these challenges, this study proposes a segmentation-guided AI framework for medical image processing and recognition that fuses FastSurfer-derived regional handcrafted features encompassing Fourier shape descriptors, two-dimensional wavelet and second-order local directional gradient pattern (LDGP) textures, and ROI-filtered XFeat keypoints with complementary multi-view (axial, sagittal, coronal) EfficientNetB0 deep representations extracted from whole-brain T1-weighted sMRI. The segmentation stage provides anatomically defined regions for radiomic feature extraction, while explainable representations maintain anatomical traceability. A key contribution of this work is its dual evaluation strategy within a unified structural feature space: it addresses both broad transdiagnostic four-class classification (Healthy Control (HC) vs. ADHD vs. BD vs. SZ) and focused binary patient-control contrasts (SZ vs. HC and BD vs. HC). To ensure rigorous evaluation, feature fusion, feature standardisation, Salp Swarm Optimizer (SSO) feature selection, hyperparameter selection, and early stopping are performed exclusively within the training partitions of a participant-level nested cross-validation (five outer and three inner folds), and held-out participants are used only once for evaluation. Evaluated on the UCLA Consortium for Neuropsychiatric Phenomics (CNP) dataset, the framework achieved 79.3% accuracy in four-class transdiagnostic classification. In focused binary tasks, the framework reached 84.3% accuracy for SZ versus HC and 82.7% accuracy for BD versus HC. These results indicate that combining segmentation-guided radiomic descriptors with multi-view deep representations provides a promising research framework for structural MRI analysis of neuropsychiatric disorders, while external validation is required before clinical application.

Background/Objectives: Lung cancer is still one of the top cancer mortality causes around the world, and there is a need for an accurate and clinically reliable diagnostic tool. While Computed Tomography (CT) imaging is very useful for evaluation of pulmonary nodules and tumor morphology, its interpretation is complicated by inter-patient variability, imaging artifacts, low tissue contrast, and tumor heterogeneity. Although Computer-Aided Diagnosis (CAD) systems have enhanced the diagnostic process, handcrafted feature-based approaches often fail to capture complex tumor characteristics, and numerous deep learning systems lack clinical interpretability. To tackle these challenges, this study suggests a unified diagnostic approach to characterize the tumor comprehensively. Methods: Lung window intensity clipping and the MedSAM foundation model are used to segment the tumor regions. After segmentation, handcrafted texture, shape, morphology and keypoint features are extracted in addition to deep features extracted by ResNet50. Particle Swarm Optimization (PSO) is used to select and refine the features, followed by an LSTM network that learns the sequential relationships among features for histological subtype classification. Results: It was observed that the proposed approach outperformed the benchmark approaches by attaining a higher accuracy of 93.70% and 94.70% on the Lung-PET-CT-Dx and LIDC-IDRI datasets, respectively. The ablation analysis supports the contribution of each module, clearly showing the progressive improvement of the overall classification performance obtained by integrating the complementary modules. Conclusions: The proposed framework effectively incorporated MedSAM-based tumor segmentation, radiomic feature analysis, and deep feature representation and sequential dependency modeling all in a single diagnostic workflow for lung cancer evaluation and diagnosis. These results prove its feasibility for explainable computer-aided diagnosis and decision support for lung cancer evaluation.

Wearable inertial sensing and markerless vision are increasingly integrated to enable objective assessment of locomotor and postural function for sports telerehabilitation, intelligent physiotherapy, and athlete performance monitoring. Before such multimodal systems can be translated to clinical practice, their fusion, optimization, and temporal modeling strategies require validation under controlled conditions with reliable ground truth. This study presents a unified multimodal framework that hierarchically integrates inertial measurement unit (IMU) signals and RGB visual information through kernelized representation learning, adaptive multimodal fusion, evolutionary feature optimization, and deep temporal classification. The IMU branch employs Kernelized Extreme Learning Machine (KELM) denoising, Kernelized Canonical Correlation Fusion (KCCF), entropy-guided adaptive windowing, and complementary time-series descriptors (MINIROCKET, TS-CHIEF, and r-STSF). Concurrently, the RGB branch combines anisotropic diffusion filtering, HRNet-based silhouette extraction, DensePose R-CNN, Mesh Graphormer, and Multi-Model Pose-Flow Fusion (MPFF) to learn robust visual representations. Both modalities are integrated through Weighted Canonical Feature Fusion (WCFF) and optimized using a Genetic Algorithm for feature selection and adaptive modality weighting before temporal modeling with cluster-based alignment, Gaussian Process Sequence Modeling, and DeepConvLSTM. As the selected benchmarks do not provide complete inertial recordings, the inertial modality is established according to the adopted experimental protocol to support multimodal fusion analysis. Under 5-fold subject-independent cross-validation, the framework achieves accuracies of 86.56 ± 0.31% on SoccerDiffusion and 88.04 ± 0.25% on HumanoidRobotPose. Although evaluated on humanoid robotic benchmarks, the proposed framework provides a methodological basis for future wearable-enabled clinical movement assessment, remote rehabilitation, and athlete monitoring, while validation on synchronized human inertial-visual datasets remains an important direction for future research.

Hepatocellular carcinoma (HCC) is a leading cause of cancer-related mortality worldwide, so its detection and monitoring are critical. However, contrast-enhanced magnetic resonance imaging (CE-MRI) is particularly vulnerable to complex, unstructured noise, which compromises image quality and diagnostic accuracy. This study proposes the use of NLE-ANSNet, a deep learning-based denoizing framework that integrates multilevel noise level estimators (NLEs) and adaptive noise scaling (ANS) within residual blocks. The model performs progressive, stagewise noise suppression at multiple feature depths, dynamically adjusting normalization based on localized noise estimates. This enables context-aware denoizing, preserving fine anatomical details. To simulate clinically realistic conditions, we developed a hybrid noise simulation framework that combines Gaussian, Poisson, and Rician noise at the pixel level. This framework aims to approximate a balanced noise distribution for evaluation purposes, with both mean and median noise levels reported to enhance evaluation robustness and prevent bias from extreme cases. NLE-ANSNet achieves a PSNR of 34.01 dB and an SSIM of 0.9393, surpassing those of state-of-the-art models. The method aims to support diagnostic reliability by preserving image structure and intensity fidelity in CE-MRI interpretation. In addition to quantitative analysis, a qualitative assessment was conducted to visually compare denoizing outputs across models, further demonstrating NLE-ANSNet’s superior ability to suppress noise while preserving diagnostically critical information. Unlike previous approaches, this study introduces a denoizing framework that combines multilevel noise estimation and adaptive noise scaling specifically tailored for CE-MRI in HCC under hybrid noise conditions—a clinically relevant and underexplored area. Overall, this study supports improved clinical decision making in HCC management.

The evolving field of Alzheimer’s disease (AD) diagnosis has greatly benefited from deep learning models for analyzing brain magnetic resonance (MR) images. This study introduces Dynamic GradNet, a novel deep learning model... | Find, read and cite all the research you need on Tech Science Press

Emotion charting using multimodal signals has gained great demand for stroke-affected patients, for psychiatrists while examining patients, and for neuromarketing applications. Multimodal signals for emotion charting include electrocardiogram (ECG) signals, electroencephalogram (EEG) signals, and galvanic skin response (GSR) signals. EEG, ECG, and GSR are also known as physiological signals, which can be used for identification of human emotions. Due to the unbiased nature of physiological signals, this field has become a great motivation in recent research as physiological signals are generated autonomously from human central nervous system. Researchers have developed multiple methods for the classification of these signals for emotion detection. However, due to the non-linear nature of these signals and the inclusion of noise, while recording, accurate classification of physiological signals is a challenge for emotion charting. Valence and arousal are two important states for emotion detection; therefore, this paper presents a novel ensemble learning method based on deep learning for the classification of four different emotional states including high valence and high arousal (HVHA), low valence and low arousal (LVLA), high valence and low arousal (HVLA) and low valence high arousal (LVHA). In the proposed method, multimodal signals (EEG, ECG, and GSR) are preprocessed using bandpass filtering and independent components analysis (ICA) for noise removal in EEG signals followed by discrete wavelet transform for time domain to frequency domain conversion. Discrete wavelet transform results in spectrograms of the physiological signal and then features are extracted using stacked autoencoders from those spectrograms. A feature vector is obtained from the bottleneck layer of the autoencoder and is fed to three classifiers SVM (support vector machine), RF (random forest), and LSTM (long short-term memory) followed by majority voting as ensemble classification. The proposed system is trained and tested on the AMIGOS dataset with k-fold cross-validation. The proposed system obtained the highest accuracy of 94.5% and shows improved results of the proposed method compared with other state-of-the-art methods.

In the recent past, a huge number of cameras have been placed in a variety of public and private areas for the purposes of surveillance, the monitoring of abnormal human actions, and traffic surveillance. The detection and recognition of abnormal activity in a real-world environment is a big challenge, as there can be many types of alarming and abnormal activities, such as theft, violence, and accidents. This research deals with accidents in traffic videos. In the modern world, video traffic surveillance cameras (VTSS) are used for traffic surveillance and monitoring. As the population is increasing drastically, the likelihood of accidents is also increasing. The VTSS is used to detect abnormal events or incidents regarding traffic on different roads and highways, such as traffic jams, traffic congestion, and vehicle accidents. Mostly in accidents, people are helpless and some die due to the unavailability of emergency treatment on long highways and those places that are far from cities. This research proposes a methodology for detecting accidents automatically through surveillance videos. A review of the literature suggests that convolutional neural networks (CNNs), which are a specialized deep learning approach pioneered to work with grid-like data, are effective in image and video analysis. This research uses CNNs to find anomalies (accidents) from videos captured by the VTSS and implement a rolling prediction algorithm to achieve high accuracy. In the training of the CNN model, a vehicle accident image dataset (VAID), composed of images with anomalies, was constructed and used. For testing the proposed methodology, the trained CNN model was checked on multiple videos, and the results were collected and analyzed. The results of this research show the successful detection of traffic accident events with an accuracy of 82% in the traffic surveillance system videos.

Epilepsy is a common brain disorder that causes patients to face multiple seizures in a single day. Around 65 million people are affected by epilepsy worldwide. Patients with focal epilepsy can be treated with surgery, whereas generalized epileptic seizures can be managed with medications. It has been noted that in more than 30% of cases, these medications fail to control epileptic seizures, resulting in accidents and limiting the patient’s life. Predicting epileptic seizures in such patients prior to the commencement of an oncoming seizure is critical so that the seizure can be treated with preventive medicines before it occurs. Electroencephalogram (EEG) signals of patients recorded to observe brain electrical activity during a seizure can be quite helpful in predicting seizures. Researchers have proposed methods that use machine and/or deep learning techniques to predict epileptic seizures using scalp EEG signals; however, prediction of seizures with increased accuracy is still a challenge. Therefore, we propose a three-step approach. It includes preprocessing of scalp EEG signals with PREP pipeline, which is a more sophisticated alternative to basic notch filtering. This method uses a regression-based technique to further enhance the SNR, with a combination of handcrafted, i.e., statistical features such as temporal mean, variance, and skewness, and automated features using CNN, followed by classification of interictal state and preictal state segments using LSTM to predict seizures. We train and validate our proposed technique on the CHB-MIT scalp EEG dataset and achieve accuracy of 94%, sensitivity of 93.8%, and 91.2% specificity. The proposed technique achieves better sensitivity and specificity than existing methods.

The most important and emerging characteristic of Wireless Body Area Networks (WBANs), which differentiates them from other wired and wireless area networks, is mobility. Therefore, the routing protocols for WBAN are designed in such a way that they can deal with dynamic changes in topology and provide maximum throughput, packet delivery ratio, average end-to-end delay, and minimum energy consumption. Thus, achieving optimal values for every performance parameter becomes a big challenge. This work investigates the performance of three separate path discovery protocols, such as Destination-Sequenced Distance-Vector Routing (DSDV), Ad Hoc On-demand Distance Vector (AODV), and Ad Hoc On-demand Multipath Distance Vector Routing protocol (AOMDV), for two different mobility models with a fixed-positioned sink. During experimentation, the AOMDV routing protocol achieves a high packet delivery ratio (PDR), average end-to-end delay, and throughput as compared to other routing protocols.

A population explosion has resulted in garbage generation on a large scale. The process of proper and automatic garbage collection is a challenging and tedious task for developing countries. This paper proposes a deep learning-based intelligent garbage detection system using an Unmanned Aerial Vehicle (UAV). The main aim of this paper is to provide a low-cost, accurate and easy-to-use solution for handling the garbage effectively. It also helps municipal corporations to detect the garbage areas in remote locations automatically. This automation was derived using two Convolutional Neural Network (CNN) models and images of solid waste were captured by the drone. Both models were trained on the collected image dataset at different learning rates, optimizers and epochs. This research uses symmetry during the sampling of garbage images. Homogeneity regarding resizing of images is generated due to the application of symmetry to extract their characteristics. The performance of two CNN models was evaluated with the state-of-the-art models using different performance evaluation metrics such as precision, recall, F1-score, and accuracy. The CNN1 model achieved better performance for automatic solid waste detection with 94% accuracy.