Soroush Baseri Saadi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Soroush Baseri Saadi has a total of 28 co-authors. Here are the top co-authors by number of shared publications:
- Reinhilde Jacobs
OMFS-IMPATH Research Group, Department of Imaging and Pathology, Catholic University Leuven, Leuven, Belgium
8 shared publications - Jonas Ver Berne
OMFS-IMPATH Research Group, Department of Imaging and Pathology, Catholic University Leuven , Leuven 3000
5 shared publications - Peter Claes
Department of Human Genetics, KU Leuven, Leuven, Belgium
5 shared publications - Rocharles Cavalcante Fontenele
Department of Stomatology, Public Health and Forensic Dentistry, Division of Oral Radiology, Ribeirao Preto School of Dentistry, University of Sao Paulo (USP), Ribeirão Preto, Brazil; Center of Excellence in Genomics and Precision Dentistry, Faculty of Dentistry, Chulalongkorn University, Bangkok, Thailand
4 shared publications - Malika Bendechache
University of Galway
3 shared publications - Ramin Ranjbarzadeh
University of Galway
3 shared publications
Latest Publications
- Open Access
- 5 November 2026
- Published by Springer Nature in Journal of Imaging Informatics in Medicine
- p. 1-17
Panognostics-Plus: An Integrated Multi-Task AI Framework for Automated Panoramic Radiograph Interpretation with Dental Charting and Structured Reporting—Development and Technical Validation Study
Dental charting from panoramic radiographs is essential for clinical assessment, yet remains time-consuming. This study presents Panognostics-Plus, an automated cloud-based AI framework for panoramic radiograph analysis integrating dental charting and structured radiological reporting. Panognostics-Plus was developed using a retrospective dataset of 1,595 panoramic radiographs and a YOLO-based architecture comprising modules for dental anatomy, dental treatment, and periapical diagnostics. Anatomical structures and dental treatments were analyzed using a two-stage pipeline, whereas periapical lesions were analyzed by detection-and-classification. YOLOv8m, YOLOv9m, YOLO11m, and YOLO26m were benchmarked. Detection was evaluated by precision, sensitivity, F1-score, AP, and mAP; segmentation by IoU and Dice; and processing time. Model differences were assessed using Friedman tests with Holm-adjusted Wilcoxon signed-rank post-hoc comparisons. YOLO26m achieved the highest overall performance. The largest mAP0@50:0.95 gains were observed for crowns (+ 21.3%), radicular cysts (+ 20.5%), root canal treatments (+ 17.3%), bridges (+ 14.3%), periodontal bone level (+ 14.3%), and periapical granulomas (+ 13.5%). Dental implants achieved precision, sensitivity, and F1-score of 0.99, 0.98, and 0.98, respectively. Radicular cysts achieved precision, sensitivity, and F1-score of 0.95, 0.93, and 0.94 versus 0.92, 0.90, and 0.91 for periapical granulomas. Significant differences were observed in 8 of 10 detection tasks and all segmentation tasks (all p < 0.001); periodontal bone level ( p = 0.193) and dental fillings ( p = 0.073) were nonsignificant. Dice coefficients reached 0.97 for dental anatomy and 0.96 for dental treatments. The workflow generated a segmentation overlay, dental chart, and structured radiological report in 15 ± 1.4 s per radiograph. Panognostics-Plus may improve panoramic radiograph analysis and clinical efficiency, although further validation is required.
- 5 November 2026
- Published by Springer Nature in Journal of Imaging Informatics in Medicine
- p. 1-17
Open AccessPanognostics-Plus: An Integrated Multi-Task AI Framework for Automated Panoramic Radiograph Interpretation with Dental Charting and Structured Reporting—Development and Technical Validation Study
Dental charting from panoramic radiographs is essential for clinical assessment, yet remains time-consuming. This study presents Panognostics-Plus, an automated cloud-based AI framework for panoramic radiograph analysis integrating dental charting and structured radiological reporting. Panognostics-Plus was developed using a retrospective dataset of 1,595 panoramic radiographs and a YOLO-based architecture comprising modules for dental anatomy, dental treatment, and periapical diagnostics. Anatomical structures and dental treatments were analyzed using a two-stage pipeline, whereas periapical lesions were analyzed by detection-and-classification. YOLOv8m, YOLOv9m, YOLO11m, and YOLO26m were benchmarked. Detection was evaluated by precision, sensitivity, F1-score, AP, and mAP; segmentation by IoU and Dice; and processing time. Model differences were assessed using Friedman tests with Holm-adjusted Wilcoxon signed-rank post-hoc comparisons. YOLO26m achieved the highest overall performance. The largest mAP0@50:0.95 gains were observed for crowns (+ 21.3%), radicular cysts (+ 20.5%), root canal treatments (+ 17.3%), bridges (+ 14.3%), periodontal bone level (+ 14.3%), and periapical granulomas (+ 13.5%). Dental implants achieved precision, sensitivity, and F1-score of 0.99, 0.98, and 0.98, respectively. Radicular cysts achieved precision, sensitivity, and F1-score of 0.95, 0.93, and 0.94 versus 0.92, 0.90, and 0.91 for periapical granulomas. Significant differences were observed in 8 of 10 detection tasks and all segmentation tasks (all p < 0.001); periodontal bone level ( p = 0.193) and dental fillings ( p = 0.073) were nonsignificant. Dice coefficients reached 0.97 for dental anatomy and 0.96 for dental treatments. The workflow generated a segmentation overlay, dental chart, and structured radiological report in 15 ± 1.4 s per radiograph. Panognostics-Plus may improve panoramic radiograph analysis and clinical efficiency, although further validation is required.
- Open Access
- 15 July 2026
- Published by Elsevier in SSRN Electronic Journal
AI-decision support for MRONJ diagnosis in multiple myeloma patients
The aim of this study was to validate and evaluate AI-assisted decision support for detecting medication-related osteonecrosis of the jaw (MRONJ) in multiple myeloma patients receiving antiresorptive drug (ARD) treatment. A retrospective dataset of 80 panoramic radiographs from adult subjects was included, comprising 40 multiple myeloma patients treated with ARD and clinically diagnosed with MRONJ, and 40 age- and sex-matched healthy controls. Standardised regions of interest were selected on each image and analysed using a pre-trained convolutional neural network adapted for binary classification of bone status (healthy vs diseased). Model performance was evaluated against clinical diagnosis of MRONJ based on clinical lesions using standard diagnostic metrics and receiver operating characteristic (ROC) analysis. The AI model correctly identified 35/40 diseased cases and all controls, resulting in no false positives and five false negatives. Diagnostic performance showed a sensitivity of 87.5%, specificity of 100%, and overall accuracy of 93.8%, with excellent agreement (κ = 0.875). ROC analysis confirmed strong discriminative ability (AUC = 0.95). These findings suggest that AI-assisted analysis of panoramic radiographs may enhance the detection of MRONJ-related bone changes and support clinical decision-making in patients with multiple myeloma receiving antiresorptive therapy.
- 15 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessAI-decision support for MRONJ diagnosis in multiple myeloma patients
The aim of this study was to validate and evaluate AI-assisted decision support for detecting medication-related osteonecrosis of the jaw (MRONJ) in multiple myeloma patients receiving antiresorptive drug (ARD) treatment. A retrospective dataset of 80 panoramic radiographs from adult subjects was included, comprising 40 multiple myeloma patients treated with ARD and clinically diagnosed with MRONJ, and 40 age- and sex-matched healthy controls. Standardised regions of interest were selected on each image and analysed using a pre-trained convolutional neural network adapted for binary classification of bone status (healthy vs diseased). Model performance was evaluated against clinical diagnosis of MRONJ based on clinical lesions using standard diagnostic metrics and receiver operating characteristic (ROC) analysis. The AI model correctly identified 35/40 diseased cases and all controls, resulting in no false positives and five false negatives. Diagnostic performance showed a sensitivity of 87.5%, specificity of 100%, and overall accuracy of 93.8%, with excellent agreement (κ = 0.875). ROC analysis confirmed strong discriminative ability (AUC = 0.95). These findings suggest that AI-assisted analysis of panoramic radiographs may enhance the detection of MRONJ-related bone changes and support clinical decision-making in patients with multiple myeloma receiving antiresorptive therapy.
- Open Access
- 13 July 2026
- Published by Springer Nature in Journal of Imaging Informatics in Medicine
JADE-Plus: A Multimodal Agentic Retrieval-Augmented Generation Large Language Framework for Diagnostic Support in Jawbone Lesions: Development and Technical Validation Study.
Diagnosing jawbone lesions in oral and maxillofacial radiology remains challenging due to overlapping radiological features and the need for integrated clinical reasoning. This study aimed to develop and validate JADE-Plus, a novel multimodal, agent-controlled retrieval-augmented generation (RAG) framework for diagnostic decision support in jawbone lesion assessment. JADE-Plus was implemented as a cloud-based, tablet-optimized system integrating a vision-language model (VLM) for panoramic radiograph analysis, a knowledge-grounded RAG module, and an agentic verification loop for diagnostic fusion and re-ranking. The system was evaluated using 40 representative jawbone lesion cases and compared with JADE, GPT-5.4, GPT-5.4 VLM, and ORAD. Performance was assessed using Top-1 and Top-3 accuracy, ablation and statistical analyses, intra-model stability, and response time. JADE-Plus achieved the highest diagnostic performance, with a Top-1 accuracy of 90% (36/40; 95% CI 76-97%) and a Top-3 accuracy of 100%, with no missed diagnoses. Cochran's Q test demonstrated significant differences among models for both Top-3 correctness (Q = 25.66, p < 0.001) and Top-1 correctness (Q = 27.55, p < 0.001). Post-hoc McNemar tests with the Benjamini-Hochberg correction showed that JADE-Plus significantly outperformed other models, particularly under the stricter Top-1 evaluation. Ablation analysis showed that the agentic verification stage improved Top-1 and Top-3 accuracy by 15 and 8 percentage points, respectively. JADE-Plus achieved the highest reproducibility (mean Jaccard similarity 0.97 ± 0.12) while maintaining a mean response time of 33 ± 1.5 s per case. JADE-Plus demonstrated superior diagnostic accuracy, stability, and reproducibility compared with baseline systems, supporting multimodal agentic RAG frameworks for jawbone lesion diagnosis.
- 13 July 2026
- Published by Springer Nature in Journal of Imaging Informatics in Medicine
Open AccessJADE-Plus: A Multimodal Agentic Retrieval-Augmented Generation Large Language Framework for Diagnostic Support in Jawbone Lesions: Development and Technical Validation Study.
Diagnosing jawbone lesions in oral and maxillofacial radiology remains challenging due to overlapping radiological features and the need for integrated clinical reasoning. This study aimed to develop and validate JADE-Plus, a novel multimodal, agent-controlled retrieval-augmented generation (RAG) framework for diagnostic decision support in jawbone lesion assessment. JADE-Plus was implemented as a cloud-based, tablet-optimized system integrating a vision-language model (VLM) for panoramic radiograph analysis, a knowledge-grounded RAG module, and an agentic verification loop for diagnostic fusion and re-ranking. The system was evaluated using 40 representative jawbone lesion cases and compared with JADE, GPT-5.4, GPT-5.4 VLM, and ORAD. Performance was assessed using Top-1 and Top-3 accuracy, ablation and statistical analyses, intra-model stability, and response time. JADE-Plus achieved the highest diagnostic performance, with a Top-1 accuracy of 90% (36/40; 95% CI 76-97%) and a Top-3 accuracy of 100%, with no missed diagnoses. Cochran's Q test demonstrated significant differences among models for both Top-3 correctness (Q = 25.66, p < 0.001) and Top-1 correctness (Q = 27.55, p < 0.001). Post-hoc McNemar tests with the Benjamini-Hochberg correction showed that JADE-Plus significantly outperformed other models, particularly under the stricter Top-1 evaluation. Ablation analysis showed that the agentic verification stage improved Top-1 and Top-3 accuracy by 15 and 8 percentage points, respectively. JADE-Plus achieved the highest reproducibility (mean Jaccard similarity 0.97 ± 0.12) while maintaining a mean response time of 33 ± 1.5 s per case. JADE-Plus demonstrated superior diagnostic accuracy, stability, and reproducibility compared with baseline systems, supporting multimodal agentic RAG frameworks for jawbone lesion diagnosis.
- Open Access
- 23 March 2026
- Published by Oxford University Press (OUP) in Dentomaxillofacial Radiology
- Vol. 55 (5) , 497-507
JADE: jawbone lesion diagnosis and decision supporting system
To develop and evaluate JADE, a proof-of-concept retrieval-augmented generation (RAG) diagnostic assistive system, designed to enhance large language model (LLM) reasoning for jawbone lesion assessment. This study examined whether RAG improves diagnostic accuracy and stability compared with standalone LLMs and ORAD, a supervised learning-based system. JADE was developed as a cloud-based RAG system integrating an expert-curated oral radiology database embedded using text-embedding-3-large and indexed in Qdrant Cloud. Structured clinical inputs were encoded as prioritized vector queries. Hybrid semantic and keyword-based retrieval appended relevant evidence to prompts for differential diagnosis generation across LLM backbones. Performance was evaluated in 25 validation cases and compared with GPT-5, Claude Sonnet 4.5, DeepSeek-R1, Gemini 2.5 Flash, their RAG configurations, and ORAD. Diagnostic accuracy was analysed using Cochran’s Q test, with post-hoc McNemar’s tests and Bonferroni correction. Intra-model stability and response time were assessed. RAG-GPT-5 achieved the highest diagnostic accuracy (20/25), followed by RAG-Claude Sonnet 4.5 (18/25), RAG-DeepSeek R1 (17/25), and RAG-Gemini 2.5 Flash (15/25). Standalone models achieved 9–13/25 correct diagnoses, ORAD achieved 17/25. No significant differences were observed among standalone models or RAG-based models and ORAD. A significant improvement was observed for GPT-5 when integrated with RAG (p = 0.002). RAG configurations showed higher intra-model stability, with RAG-GPT-5 achieving mean stability of 0.90 ± 0.11. Mean response times ranged from 3-10 seconds. JADE improved diagnostic accuracy and stability compared with standalone LLMs, underscoring value of RAG reasoning in jawbone lesion assessment and marking the first RAG application in dentomaxillofacial radiology.
- 23 March 2026
- Published by Oxford University Press (OUP) in Dentomaxillofacial Radiology
- Vol. 55 (5) , 497-507
Open AccessJADE: jawbone lesion diagnosis and decision supporting system
To develop and evaluate JADE, a proof-of-concept retrieval-augmented generation (RAG) diagnostic assistive system, designed to enhance large language model (LLM) reasoning for jawbone lesion assessment. This study examined whether RAG improves diagnostic accuracy and stability compared with standalone LLMs and ORAD, a supervised learning-based system. JADE was developed as a cloud-based RAG system integrating an expert-curated oral radiology database embedded using text-embedding-3-large and indexed in Qdrant Cloud. Structured clinical inputs were encoded as prioritized vector queries. Hybrid semantic and keyword-based retrieval appended relevant evidence to prompts for differential diagnosis generation across LLM backbones. Performance was evaluated in 25 validation cases and compared with GPT-5, Claude Sonnet 4.5, DeepSeek-R1, Gemini 2.5 Flash, their RAG configurations, and ORAD. Diagnostic accuracy was analysed using Cochran’s Q test, with post-hoc McNemar’s tests and Bonferroni correction. Intra-model stability and response time were assessed. RAG-GPT-5 achieved the highest diagnostic accuracy (20/25), followed by RAG-Claude Sonnet 4.5 (18/25), RAG-DeepSeek R1 (17/25), and RAG-Gemini 2.5 Flash (15/25). Standalone models achieved 9–13/25 correct diagnoses, ORAD achieved 17/25. No significant differences were observed among standalone models or RAG-based models and ORAD. A significant improvement was observed for GPT-5 when integrated with RAG (p = 0.002). RAG configurations showed higher intra-model stability, with RAG-GPT-5 achieving mean stability of 0.90 ± 0.11. Mean response times ranged from 3-10 seconds. JADE improved diagnostic accuracy and stability compared with standalone LLMs, underscoring value of RAG reasoning in jawbone lesion assessment and marking the first RAG application in dentomaxillofacial radiology.
- Open Access
JADE: Jawbone Lesion Diagnosis and Decision Supporting System
Objectives: To develop and evaluate JADE, a proof-of-concept retrieval-augmented generation (RAG) diagnostic assistive system was designed to enhance large language model (LLM) reasoning for the assessment of jawbone lesions. This study examined whether integrating structured retrieval with GPT-5 improves diagnostic accuracy and stability compared with standalone LLMs. Methods: JADE was developed as a cloud-based application integrating GPT-5 with a curated oral radiology and pathology database using a hybrid semantic-keyword retrieval strategy. Clinical and radiographic characteristics were imported as a structured query to guide retrieval and support diagnostic reasoning. Performance was compared with standalone GPT-5, Claude Sonnet 4.5, DeepSeek-R1, and Gemini 2.5 Flash across 25 cases. Accuracy was analysed using Cochran’s Q test with post-hoc McNemar’s tests and Bonferroni correction. Intra-model stability was measured using the majority agreement ratio, and response time was recorded to assess real-time usability. Results: JADE showed the highest diagnostic performance, correctly identifying 20 out of 25 cases and outperforming all standalone LLMs. Significant differences were observed across models (Cochran’s Q = 33.2, df = 4, p < 0.001), with post-hoc analyses confirming that JADE significantly outperformed GPT-5, Gemini 2.5 Flash, and Claude Sonnet 4.5 ( p < 0.01). JADE also exhibited the greatest run-to-run stability (mean MAR = 0.90 ± 0.18). The average prediction time of 6 ± 0.5 seconds supported its feasibility for real-time clinical use. Conclusions: JADE improved diagnostic accuracy and stability over standalone LLMs, underscoring the value of RAG reasoning in jawbone lesion assessment and its potential for real-time clinical use.
Open AccessJADE: Jawbone Lesion Diagnosis and Decision Supporting System
Objectives: To develop and evaluate JADE, a proof-of-concept retrieval-augmented generation (RAG) diagnostic assistive system was designed to enhance large language model (LLM) reasoning for the assessment of jawbone lesions. This study examined whether integrating structured retrieval with GPT-5 improves diagnostic accuracy and stability compared with standalone LLMs. Methods: JADE was developed as a cloud-based application integrating GPT-5 with a curated oral radiology and pathology database using a hybrid semantic-keyword retrieval strategy. Clinical and radiographic characteristics were imported as a structured query to guide retrieval and support diagnostic reasoning. Performance was compared with standalone GPT-5, Claude Sonnet 4.5, DeepSeek-R1, and Gemini 2.5 Flash across 25 cases. Accuracy was analysed using Cochran’s Q test with post-hoc McNemar’s tests and Bonferroni correction. Intra-model stability was measured using the majority agreement ratio, and response time was recorded to assess real-time usability. Results: JADE showed the highest diagnostic performance, correctly identifying 20 out of 25 cases and outperforming all standalone LLMs. Significant differences were observed across models (Cochran’s Q = 33.2, df = 4, p < 0.001), with post-hoc analyses confirming that JADE significantly outperformed GPT-5, Gemini 2.5 Flash, and Claude Sonnet 4.5 ( p < 0.01). JADE also exhibited the greatest run-to-run stability (mean MAR = 0.90 ± 0.18). The average prediction time of 6 ± 0.5 seconds supported its feasibility for real-time clinical use. Conclusions: JADE improved diagnostic accuracy and stability over standalone LLMs, underscoring the value of RAG reasoning in jawbone lesion assessment and its potential for real-time clinical use.
- Open Access
- 28 August 2025
- Published by JMIR Publications Inc. in JMIR Preprints
Panognostics: A Novel Intelligent Tablet Application for Real-Time Automated Dental Charting and Structured Reporting on Panoramic Radiographs (Preprint)
BACKGROUND Panoramic radiographs are essential in dental diagnostics. They provide a broad overview of the oral and maxillofacial region. However, their interpretation is often limited by overlapping structures, patient positioning issues, and low spatial resolution. Moreover, reliance on manual analysis can be highly time-consuming, and subject to diagnostic inconsistency. These challenges highlight the need for an intelligent system that can automate interpretation and support consistency and accuracy in clinical decision-making. OBJECTIVE This study presents Panognostics, a novel cloud-based, tablet-compatible AI application developed to automate the interpretation of panoramic radiographs through detection, segmentation, charting, and structured reporting of key anatomical structures. The system is designed to enhance diagnostic accuracy, streamline reporting, and diagnostic support through an intuitive, real-time interface optimized for use in chairside clinical settings. METHODS An ethically approved dataset of 2,768 panoramic radiographs was used to develop and validate a two-stage YOLOv11-based pipeline for detecting and segmenting teeth, periodontal bone level, and the maxillary sinus floor. Performance was evaluated using precision, sensitivity, and mAP. A clinical inference module was developed to classify maxillary posterior teeth as either overlapping or in proximity to the sinus floor, tested on 40 panoramic radiographs (399 maxillary posterior teeth). Interpretations were visualized on an interactive dental chart and summarized in an automatically generated structured report. Application response time was recorded to assess real-time feasibility. RESULTS The application demonstrated high detection (mAP@0.5: 0.96–0.97) and segmentation performance, with precision and sensitivity above 0.94 for all dental and bony structures. Sinus segmentation reached 0.98 precision and 0.99 sensitivity, while tooth and periodontal bone level segmentation achieved mAP@0.5 of 0.92. Among 399 MT analysed, molars accounted for over 65% of sinus involvement, while premolars and canines were more often associated with sinus proximity. Panognostics’ average response time was under 16 seconds. CONCLUSIONS Panognostics offers accurate, fast, and interpretable AI-based detection and segmentation of teeth, periodontal bone level, and the maxillary sinus floor, enabling automated creation of an interactive dental chart and a standardized structured report. This tool allows efficient visualization and documentation of panoramic radiograph findings through a scalable, tablet-compatible platform. Optimized for both professional decision-support and patient communication, Panognostics integrates easily into daily dental workflow, underscoring the emphasizing the innovative potential of tablet-based AI solutions in modern oral healthcare.
- 28 August 2025
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessPanognostics: A Novel Intelligent Tablet Application for Real-Time Automated Dental Charting and Structured Reporting on Panoramic Radiographs (Preprint)
BACKGROUND Panoramic radiographs are essential in dental diagnostics. They provide a broad overview of the oral and maxillofacial region. However, their interpretation is often limited by overlapping structures, patient positioning issues, and low spatial resolution. Moreover, reliance on manual analysis can be highly time-consuming, and subject to diagnostic inconsistency. These challenges highlight the need for an intelligent system that can automate interpretation and support consistency and accuracy in clinical decision-making. OBJECTIVE This study presents Panognostics, a novel cloud-based, tablet-compatible AI application developed to automate the interpretation of panoramic radiographs through detection, segmentation, charting, and structured reporting of key anatomical structures. The system is designed to enhance diagnostic accuracy, streamline reporting, and diagnostic support through an intuitive, real-time interface optimized for use in chairside clinical settings. METHODS An ethically approved dataset of 2,768 panoramic radiographs was used to develop and validate a two-stage YOLOv11-based pipeline for detecting and segmenting teeth, periodontal bone level, and the maxillary sinus floor. Performance was evaluated using precision, sensitivity, and mAP. A clinical inference module was developed to classify maxillary posterior teeth as either overlapping or in proximity to the sinus floor, tested on 40 panoramic radiographs (399 maxillary posterior teeth). Interpretations were visualized on an interactive dental chart and summarized in an automatically generated structured report. Application response time was recorded to assess real-time feasibility. RESULTS The application demonstrated high detection (mAP@0.5: 0.96–0.97) and segmentation performance, with precision and sensitivity above 0.94 for all dental and bony structures. Sinus segmentation reached 0.98 precision and 0.99 sensitivity, while tooth and periodontal bone level segmentation achieved mAP@0.5 of 0.92. Among 399 MT analysed, molars accounted for over 65% of sinus involvement, while premolars and canines were more often associated with sinus proximity. Panognostics’ average response time was under 16 seconds. CONCLUSIONS Panognostics offers accurate, fast, and interpretable AI-based detection and segmentation of teeth, periodontal bone level, and the maxillary sinus floor, enabling automated creation of an interactive dental chart and a standardized structured report. This tool allows efficient visualization and documentation of panoramic radiograph findings through a scalable, tablet-compatible platform. Optimized for both professional decision-support and patient communication, Panognostics integrates easily into daily dental workflow, underscoring the emphasizing the innovative potential of tablet-based AI solutions in modern oral healthcare.
- Open Access
- 16 March 2025
- Published by Elsevier in Journal of Dentistry
- Vol. 156, 105688
Automated classification of panoramic radiographs with inflammatory periapical lesions using a CNN-LSTM architecture
- 16 March 2025
- Published by Elsevier in Journal of Dentistry
- Vol. 156, 105688
Open AccessAutomated classification of panoramic radiographs with inflammatory periapical lesions using a CNN-LSTM architecture
- Open Access
- 13 December 2024
- Published by Elsevier in Journal of Dentistry
- Vol. 153, 105525
Development and validation of an AI-driven tool to evaluate chewing function: a proof of concept
- 13 December 2024
- Published by Elsevier in Journal of Dentistry
- Vol. 153, 105525
Open AccessDevelopment and validation of an AI-driven tool to evaluate chewing function: a proof of concept
- 8 October 2023
- Published by Springer Nature
- p. 345-366
Brain Tumor Segmentation Based on Zernike Moments, Enhanced Ant Lion Optimization, and Convolutional Neural Network in MRI Images
- 8 October 2023
- Published by Springer Nature
- p. 345-366
Brain Tumor Segmentation Based on Zernike Moments, Enhanced Ant Lion Optimization, and Convolutional Neural Network in MRI Images
- Open Access
- 23 March 2022
- Published by Wiley in Journal of Healthcare Engineering
- Vol. 2022, 1-11
Investigation of Effectiveness of Shuffled Frog-Leaping Optimizer in Training a Convolution Neural Network
One of the leading algorithms and architectures in deep learning is Convolution Neural Network (CNN). It represents a unique method for image processing, object detection, and classification. CNN has shown to be an efficient approach in the machine learning and computer vision fields. CNN is composed of several filters accompanied by nonlinear functions and pooling layers. It enforces limitations on the weights and interconnections of the neural network to create a good structure for processing spatial and temporal distributed data. A CNN can restrain the numbering of free parameters of the network through its weight-sharing property. However, the training of CNNs is a challenging approach. Some optimization techniques have been recently employed to optimize CNN’s weight and biases such as Ant Colony Optimization, Genetic, Harmony Search, and Simulated Annealing. This paper employs the well-known nature-inspired algorithm called Shuffled Frog-Leaping Algorithm (SFLA) for training a classical CNN structure (LeNet-5), which has not been experienced before. The training method is investigated by employing four different datasets. To verify the study, the results are compared with some of the most famous evolutionary trainers: Whale Optimization Algorithm (WO), Bacteria Swarm Foraging Optimization (BFSO), and Ant Colony Optimization (ACO). The outcomes demonstrate that the SFL technique considerably improves the performance of the original LeNet-5 although using this algorithm slightly increases the training computation time. The results also demonstrate that the suggested algorithm presents high accuracy in classification and approximation in its mechanism.
- 23 March 2022
- Published by Wiley in Journal of Healthcare Engineering
- Vol. 2022, 1-11
Open AccessInvestigation of Effectiveness of Shuffled Frog-Leaping Optimizer in Training a Convolution Neural Network
One of the leading algorithms and architectures in deep learning is Convolution Neural Network (CNN). It represents a unique method for image processing, object detection, and classification. CNN has shown to be an efficient approach in the machine learning and computer vision fields. CNN is composed of several filters accompanied by nonlinear functions and pooling layers. It enforces limitations on the weights and interconnections of the neural network to create a good structure for processing spatial and temporal distributed data. A CNN can restrain the numbering of free parameters of the network through its weight-sharing property. However, the training of CNNs is a challenging approach. Some optimization techniques have been recently employed to optimize CNN’s weight and biases such as Ant Colony Optimization, Genetic, Harmony Search, and Simulated Annealing. This paper employs the well-known nature-inspired algorithm called Shuffled Frog-Leaping Algorithm (SFLA) for training a classical CNN structure (LeNet-5), which has not been experienced before. The training method is investigated by employing four different datasets. To verify the study, the results are compared with some of the most famous evolutionary trainers: Whale Optimization Algorithm (WO), Bacteria Swarm Foraging Optimization (BFSO), and Ant Colony Optimization (ACO). The outcomes demonstrate that the SFL technique considerably improves the performance of the original LeNet-5 although using this algorithm slightly increases the training computation time. The results also demonstrate that the suggested algorithm presents high accuracy in classification and approximation in its mechanism.