The histopathological

 This research investigates and optimizes routing protocols in Internet of Things (IoT) based telecommunication networks. Considering the increasing number of connected devices, resource constraints, and the need for fast and stable data transmission, optimization of routing protocols plays a key role in improving network efficiency, reducing energy consumption, and increasing reliability. In this research, an improved DCUR algorithm with the Whale Optimization algorithm is proposed to select optimal routes and balance load distribution in the IoT network. This method determines the optimal data transmission routes using criteria such as received signal strength (RSS), distance between nodes, energy status, and temperature of nodes. Simulation results show that implementing the optimal protocol reduces data transmission latency, increases network lifetime, reduces

In recent years, intelligent video surveillance systems have played a vital role in enhancing safety and security across various environments. This thesis presents the design and implementation of a comprehensive real-time detection and alerting system capable of identifying specific hazardous objects, including various weapons (pistols, swords, knives), incidents (accidents, fire, smoke), and violence. The proposed method is based on the utilization of two distinct object detection models founded on the YOLO11 (Nano version) architecture. The primary model is used for detecting general objects such as humans and vehicles, while the secondary model—trained on a customized dataset—focuses on the precise identification of hazardous items. This custom dataset was collected, labeled, and integrated using Roboflow tools. The software architecture consists of a processing core based on PyTorch (with CPU or GPU execution capabilities) and OpenCV. To manage concurrent processes, a threading-based system is employed to decouple video processing from the user interface. An advanced Graphical User Interface (GUI) has been developed using the PyQt6 library, allowing for the adjustment of various parameters such as confidence thresholds, input modes (webcam, video/image files), and the display of system logs. A key feature of the system is the multi-frame verification mechanism designed to prevent false alarms. Furthermore, upon threat detection, an immediate audio alert system (including sirens and voice messages) is activated, and the final output video is merged with the original audio and saved. Experimental results demonstrate that the system is capable of identifying hazardous objects in various scenarios with high accuracy and speed, providing a real-time response.   

   Schizophrenia is one of the most complex psychiatric disorders characterized by widespread abnormalities in functional brain connectivity. This study aims to develop a convolutional neural network-based >In this study, four functional connectivity metrics (PLV, PCC, MSC, and MI) were extracted from EEG signals across five frequency bands. Using a novel brain anatomy-based approach, the 19×19 connectivity matrices were reduced to 5×15, achieving over 90% reduction in feature dimensionality while preserving neurologically meaningful information. For >Results demonstrated that in the subject-independent approach, the multi-branch architecture utilizing all four metrics simultaneously achieved 99.92% accuracy, 100% sensitivity, and 99.82% specificity. In the subject-dependent approach, 78.44% accuracy was obtained with 76.70% sensitivity and 80.53% specificity. Mutual Information (MI) showed the best single-metric performance with 99.43% accuracy in subject-independent evaluation, while Phase Locking Value (PLV) demonstrated the best performance with 79.98% accuracy in subject-dependent evaluation. The innovations of this research include the anatomy-based dimensionality reduction method, comprehensive comparison of four connectivity metrics, multi-branch architecture for simultaneous processing of spatial-frequency information, and dual evaluation paradigm. The proposed system demonstrates high potential for use as a diagnostic aid in early schizophrenia detection, other psychiatric disorders, and brain-computer interface applications. Schizophrenia is one of the most complex psychiatric disorders characterized by widespread abnormalities in functional brain connectivity. This study aims to develop a convolutional neural network-based >In this study, four functional connectivity metrics (PLV, PCC, MSC, and MI) were extracted from EEG signals across five frequency bands. Using a novel brain anatomy-based approach, the 19×19 connectivity matrices were reduced to 5×15, achieving over 90% reduction in feature dimensionality while preserving neurologically meaningful information. For >Results demonstrated that in the subject-independent approach, the multi-branch architecture utilizing all four metrics simultaneously achieved 99.92% accuracy, 100% sensitivity, and 99.82% specificity. In the subject-dependent approach, 78.44% accuracy was obtained with 76.70% sensitivity and 80.53% specificity. Mutual Information (MI) showed the best single-metric performance with 99.43% accuracy in subject-independent evaluation, while Phase Locking Value (PLV) demonstrated the best performance with 79.98% accuracy in subject-dependent evaluation. The innovations of this research include the anatomy-based dimensionality reduction method, comprehensive comparison of four connectivity metrics, multi-branch architecture for simultaneous processing of spatial-frequency information, and dual evaluation paradigm. The proposed system demonstrates high potential for use as a diagnostic aid in early schizophrenia detection, other psychiatric disorders, and brain-computer interface applications.

   With the rapid development of smart cities and the growing need for accurate and real-time analysis of road infrastructure, the design of AI-based systems capable of perceiving, analyzing, and recording environmental data has become increasingly crucial. In this regard, the present study focuses on the design and implementation of an innovative system for the automatic detection, tracking, and localization of traffic signs. This system not only pushes technical boundaries but also makes a significant contribution to the localization of traffic-related data. The advanced YOLOv9 model is employed for precise traffic sign detection, while the powerful ByteTrack algorithm ensures continuous tracking. What truly distinguishes this research is the novel application of federated learning using the FedAvg algorithm—implemented for the first time in the domain of traffic sign recognition. This method enables the training of models on heterogeneous datasets, including two distinct subsets, DFG and Mapillary, without requiring physical data aggregation. This approach not only preserves data privacy but also significantly enhances the generalization capability of the model. On the data side, the study introduces a rich and unprecedented dataset comprising 14,111 images and over 19,000 traffic sign instances across 118 classes. The data was collected over two years in varying temporal conditions (morning, noon, evening, night) and all four seasons, spanning urban, rural, and interurban areas across the country using mobile phone cameras. The images were meticulously annotated using the Makesense tool in both YOLO (.txt) and Pascal VOC (.xml) formats. The system’s performance, evaluated through 6-Fold Cross Validation, demonstrates its high accuracy, achieving a remarkable mAP50 of 95.66%. This not only reflects the model's robustness in real-world conditions but also shows a clear advantage over various versions of YOLO from v5 to v11. The initial idea for this project stemmed from a proposal by Tehran Municipality to design a digital map of traffic signs. However, the outcomes of this research go far beyond a municipal application and offer valuable tools for navigation systems, intelligent vehicles, spatial analytics, and the development of a national traffic sign map in Iran. This work presents a seamless integration of cutting-edge technology, locally-driven data, and modern AI architectures to pave the way for a smarter and safer future on the country’s roads.

Sleep is one of the fundamental human needs that significantly impacts physical and mental health. Stress during sleep can lead to sleep disorders and related health issues, making accurate prediction of sleep stress particularly important. This thesis explores the detection of sleep stress using deep learning, specifically focusing on LSTM and GRU recurrent neural network models, as well as a hybrid model combining the two. The aim of this research is to provide an efficient and accurate model for predicting and detecting sleep stress based on the SaYoPillow dataset. We used 10-fold Cross Validation. Various models were evaluated, and the results showed that the hybrid Bidirectional LSTM-GRU model achieved the best performance with an accuracy of 1.00, precision of 1.00, recall of 1.00, and an F1 score of 1.00, outperforming individual LSTM and GRU, and MLP models in detecting all 5 levels of sleep stress. The use of a confusion matrix and evaluation metrics such as accuracy, precision, recall, and F1 score demonstrated that the hybrid model not only has high accuracy in detecting positive cases but also reduces errors related to identifying negative cases. This research highlights that deep learning models, particularly the hybrid Bidirectional LSTM-GRU model, can be effective tools for detecting sleep stress, thereby contributing to improved sleep quality and overall health. The development of these models can assist healthcare professionals in providing appropriate preventive and therapeutic strategies for managing sleep stress.

In linear projects with long paths, such as roads and railways, encountering unstable infrastructures such as sedimentary layers and soft soils is inevitable. This issue leads to problems such as excessive settlement, reduced load-bearing capacity, and the destruction of pavement layers. This research will examine the impact of implementing deep stone heel structures on the sides of existing pavement sections at various distances and geometric dimensions in reducing the amount of settlement. For this purpose, two-dimensional modeling was carried out using numerical modeling methods in the Midas GTX NX software. The results of various models were compared in three general scenarios: a section without the use of deep stone heel structures, a section with deep stone heel structures of variable dimensions, and a section with deep stone heel structures of variable dimensions combined with geosynthetic enclosures, under three different heights of soft layers of 2, 4, and 7 meters.  

Abstract Research Aim: Tailings dams are built in the mining industry to manage and store the residual waste materials, known as tailings, that remain after the extraction and beneficiation of metallic and non-metallic ores. Tailings often contain toxic materials. This research involves the simulation of a localized tailings dam breach with a water pond, the investigation of the hydraulic characteristics of the flow, and the comparison of a tailings dam breach with and without lateral slopes. It also investigates the impact of different water surface elevations, locations, and breach widths on the flood and the volume of eroded sediment in a tailings dam's local failure. Research method: In this research, the two-dimensional HEC-RAS software, which utilizes shallow water equations, was used for the numerical simulation of laboratory tailings dam failure cases. For the validation of the simulation results against the laboratory data, the Mean Absolute Relative Error (MARE) and the coefficient of determination (R²) were employed. Findings: This research was conducted to investigate the pattern of bed scour, the amount of sediment transport, changes in the hydraulic quantities of the flow, and information about the floods resulting from local failures under different water surface elevations, locations, and breach widths. Conclusion: The Wu as the sediment transport function was most compatible with the experimental results. In the case of local failure far from the abutment, the shear stress and the Froude number were maximized after the localized dam breach in the downstream flow, which decreased over time with the reduction of the water surface elevation and began to increase at the edge of the scoured area. The side slope increased the scour by a factor of 1.3. Unlike the breach location and the lateral slope, the breach width and the water surface elevation significantly impacted the hydrograph pattern of the outflow resulting from the tailings dam failure. Ultimately, HEC-RAS was able to perform the simulation with relatively high accuracy, which could serve as an alternative to costly and time-consuming experiments. Keywords: Tailings dam, Dam break, Scour, Numerical simulation, HEC-RAS 2D

   Parkinson's disease is a brain disorder caused by damage to dopamine producing cells in the brain. People with Parkinson's disease have symptoms such as tremors and slowness of movement, which makes it difficult for these people to control their movements. Parkinson's is usually diagnosed based on tests done by a neurologist. Actions such as; Analysis of the patient's medical history, examination of symptoms, neurological and physical examination. Therefore, the identification of Parkinson's disease is a long-term process that always requires the availability of all the patient's information (history) and their careful study in each session. Therefore, according to the conditions and problems that exist in this field, misdiagnosis is among the possibilities according to its risks. One of the solutions used today to prevent such mistakes is the use of automatic machine learning detection systems. Considering the mentioned issues and problems, this study tests a two-way LSTM model with two activation functions, Softsign and Tanh, for the automatic diagnosis of Parkinson's disease based on the gait analysis of PD people. The raw data of VGRF signals obtained from the Physionet database were tested in the proposed model to classify PD and healthy subjects. Experiments show the high efficiency of the proposed method in diagnosing Parkinson's disease based on the analysis of movement signals related to people's walking. The proposed algorithm achieved 97.1% accuracy. Among the methods investigated in this study, the presented method has obtained the best performance in the diagnosis of Parkinson's disease using movement signals related to walking. These results show that this model can learn efficient features from existing data that can be useful in clinical diagnosis.