Sakthivel Balu
Sakthivel Balu has a total of 11 co-authors. Here are the top co-authors by number of shared publications:
- Banu Priya Ganapathy Raman1 shared publication
- Duraichi Natarajan
Department of Electronics and Communication Engineering, Saveetha School of Engineering, Saveetha Institute of Medical and Technical Sciences, Saveetha University, Chennai, India
1 shared publication - Duraichi Natarajan1 shared publication
G. Narmadha
1 shared publicationM. Jeyalakshmi
1 shared publicationM. Ponnrajakumari
1 shared publication
Latest Publications
- 31 August 2026
- Published by Bentham Science Publishers Ltd. in Current Medical Imaging
- Vol. 22, 1-17
Multimodal Osteoporosis Assessment using X-ray Imaging and Clinical Data with a Hybrid Deep Learning Model
Introduction/Background: Osteoporosis is a systemic skeletal disorder characterised by low bone mass. The classification typically includes three stages: normal, osteopenia, and osteoporosis. Early detection is essential for effective treatment and reducing fracture risk. This detection is possible through X-ray imaging and clinical assessments, such as measuring T-score and SoS, etc. This study aims to enhance the accuracy of osteoporosis assessment by combining both imaging and clinical data using advanced AI techniques. Materials and Methods: A new multimodal approach is proposed for osteoporosis assessment based on imaging and clinical data. The dataset is collected from Begum Shafia Nazir Husain Trust Hospital in Chennai, as well as from various rural clinic camps organized across different regions. For X-ray image classification, a new DL model, the Dynamic Topological Convolutional Transformer (DTCT), is proposed. For clinical data processing, a hybrid model combining Self-Organising Maps (SOM) with Graph Neural Networks (GNN) is applied. The approach is evaluated on a dataset collected from the Begum Shafia Nazir Husain Trust Hospital. Results: The DTCT-based classification achieves the highest accuracy of 96% for multi-class classification. The SOM-combined GNN model achieves a maximum accuracy of 95% in classification. It achieves a robust framework for integrating different data modalities in medical diagnostics. Discussion: The proposed method provides high classification accuracy, improved interpretability, robust performance against noise and variations, and a scalable architecture for efficient adaptation to diverse datasets. Conclusion: The proposed approach combines imaging and clinical data for an accurate osteoporosis assessment with superior performance over existing techniques.
- 31 August 2026
- Published by Bentham Science Publishers Ltd. in Current Medical Imaging
- Vol. 22, 1-17
Multimodal Osteoporosis Assessment using X-ray Imaging and Clinical Data with a Hybrid Deep Learning Model
Introduction/Background: Osteoporosis is a systemic skeletal disorder characterised by low bone mass. The classification typically includes three stages: normal, osteopenia, and osteoporosis. Early detection is essential for effective treatment and reducing fracture risk. This detection is possible through X-ray imaging and clinical assessments, such as measuring T-score and SoS, etc. This study aims to enhance the accuracy of osteoporosis assessment by combining both imaging and clinical data using advanced AI techniques. Materials and Methods: A new multimodal approach is proposed for osteoporosis assessment based on imaging and clinical data. The dataset is collected from Begum Shafia Nazir Husain Trust Hospital in Chennai, as well as from various rural clinic camps organized across different regions. For X-ray image classification, a new DL model, the Dynamic Topological Convolutional Transformer (DTCT), is proposed. For clinical data processing, a hybrid model combining Self-Organising Maps (SOM) with Graph Neural Networks (GNN) is applied. The approach is evaluated on a dataset collected from the Begum Shafia Nazir Husain Trust Hospital. Results: The DTCT-based classification achieves the highest accuracy of 96% for multi-class classification. The SOM-combined GNN model achieves a maximum accuracy of 95% in classification. It achieves a robust framework for integrating different data modalities in medical diagnostics. Discussion: The proposed method provides high classification accuracy, improved interpretability, robust performance against noise and variations, and a scalable architecture for efficient adaptation to diverse datasets. Conclusion: The proposed approach combines imaging and clinical data for an accurate osteoporosis assessment with superior performance over existing techniques.
- Open Access
- 25 April 2026
- Published by Polskie Naukowo-Techniczne Towarzystwo Eksploatacyjne in Eksploatacja i Niezawodność – Maintenance and Reliability
reliability-centered approach for solar panel fault detection and classification via autonomous drone system in solar farms
In this work, a drone-based system is implemented for autonomous real-time fault detection in solar farms for improved reliability of power generation. To capture and analyze high-dimensional features from aerial images of solar panels, three powerful feature-extracting models are used. Then, for feature optimization, a new modified hybrid metaheuristic called Red Panda-Dhole Optimization (RPDO) is proposed. For accurate classification, a residual boosting-based Long Short-Term Memory (LSTM) architecture is proposed. The developed drone, based on the STM32F427VG microcontroller, follows a predefined flight path autonomously and captures images of the solar farm at regular intervals. These images are transmitted to a standalone system where they are processed and classified into different fault categories using a trained model. The proposed model is trained and tested on inspection videos recorded with drones in solar farms. The model achieved the highest accuracy of 96.2% for the classification of solar panel status into Bird-drop, Clean, Dusty, Electrical-damage, and Physical-damage
- 25 April 2026
- Published by Polskie Naukowo-Techniczne Towarzystwo Eksploatacyjne in Eksploatacja i Niezawodność – Maintenance and Reliability
Open Accessreliability-centered approach for solar panel fault detection and classification via autonomous drone system in solar farms
In this work, a drone-based system is implemented for autonomous real-time fault detection in solar farms for improved reliability of power generation. To capture and analyze high-dimensional features from aerial images of solar panels, three powerful feature-extracting models are used. Then, for feature optimization, a new modified hybrid metaheuristic called Red Panda-Dhole Optimization (RPDO) is proposed. For accurate classification, a residual boosting-based Long Short-Term Memory (LSTM) architecture is proposed. The developed drone, based on the STM32F427VG microcontroller, follows a predefined flight path autonomously and captures images of the solar farm at regular intervals. These images are transmitted to a standalone system where they are processed and classified into different fault categories using a trained model. The proposed model is trained and tested on inspection videos recorded with drones in solar farms. The model achieved the highest accuracy of 96.2% for the classification of solar panel status into Bird-drop, Clean, Dusty, Electrical-damage, and Physical-damage
- Open Access
- 10 May 2025
- Published by Elsevier in Results in Engineering
- Vol. 26
Enhancing spectrum prediction in cognitive radio networks using an optimized generative adversarial network
- 10 May 2025
- Published by Elsevier in Results in Engineering
- Vol. 26
Open AccessEnhancing spectrum prediction in cognitive radio networks using an optimized generative adversarial network
- 1 January 2025
- Published by Institutul Național de Cercetare Dezvoltare Agricolă Fundulea in Romanian Agriculture Research
- Vol. 42, 991-1004
Advancing Agriculture 6.0: A Novel Residual Recurrent Unit for Crop Recommendation Using IoT Data
In the era of smart technologies, Agriculture 6.0 utilized the most advanced technologies like artificial intelligence to enhance smart farming systems. It processed an optimal crop selection to maximize agricultural productivity based on specific environmental and soil conditions. In this work, the new Deep Learning (DL) method is presented such as the Residual Recurrent Unit (RRU) Model which has two key modules Adaptive Scaling in the Transform Module and a Residual Connection Module. These modules optimized the model's sensitivity to varying inputs and also allowed untransformed data to directly influence the cell state to improve gradient flow. These features enhance the model performance by similar past and present inputs effectively. To further optimize performance, Siberian Tiger Optimization (STO) was used to tune the hyperparameter and attain the best possible result. The experimental result of the novel RRU model validated for both with and without STO based on various metrics like Accuracy, Precision, Recall, F1-Score, ROC, Loss, Cohen's Kappa and Matthews Correlation Coefficient etc. based on the validation, the proposed RRU model has achieved a superior performance in all aspects than traditional methods. Also, this work included a hardware setup and collected data from a real field to investigate the model’s performances based on soil and environmental factors. Therefore, real data results based on the novel RRU model provided an accurate crop recommendation and it is majorly supportive of precision agriculture with effectiveness in future.
- 1 January 2025
- Published by Institutul Național de Cercetare Dezvoltare Agricolă Fundulea in Romanian Agriculture Research
- Vol. 42, 991-1004
Advancing Agriculture 6.0: A Novel Residual Recurrent Unit for Crop Recommendation Using IoT Data
In the era of smart technologies, Agriculture 6.0 utilized the most advanced technologies like artificial intelligence to enhance smart farming systems. It processed an optimal crop selection to maximize agricultural productivity based on specific environmental and soil conditions. In this work, the new Deep Learning (DL) method is presented such as the Residual Recurrent Unit (RRU) Model which has two key modules Adaptive Scaling in the Transform Module and a Residual Connection Module. These modules optimized the model's sensitivity to varying inputs and also allowed untransformed data to directly influence the cell state to improve gradient flow. These features enhance the model performance by similar past and present inputs effectively. To further optimize performance, Siberian Tiger Optimization (STO) was used to tune the hyperparameter and attain the best possible result. The experimental result of the novel RRU model validated for both with and without STO based on various metrics like Accuracy, Precision, Recall, F1-Score, ROC, Loss, Cohen's Kappa and Matthews Correlation Coefficient etc. based on the validation, the proposed RRU model has achieved a superior performance in all aspects than traditional methods. Also, this work included a hardware setup and collected data from a real field to investigate the model’s performances based on soil and environmental factors. Therefore, real data results based on the novel RRU model provided an accurate crop recommendation and it is majorly supportive of precision agriculture with effectiveness in future.