Sabura Banu Urundai Meeran
Saveetha Engineering College

Published : 4 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 4 Documents
Search

Osprey optimization algorithm for VGG16 hyperparameter optimization in breast cancer detection Sabura Banu Urundai Meeran; Nafeena Abdul Munaf; Vengadeshwaran Velu
Bulletin of Electrical Engineering and Informatics Vol 15, No 3: June 2026
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eei.v15i3.11181

Abstract

Globally, breast cancer is one of the reason for mortality among women and accurate automated diagnosis remains a critical research challenge. This research is used to improve breast cancer classification performance by optimizing deep learning (DL) model hyperparameters using a bio-inspired optimization technique. The osprey optimization algorithm (OOA) is applied to fine-tune the hyperparameters of the VGG16 convolutional neural network (CNN) for histopathological breast cancer image classification. The optimized model is evaluated using a curated dataset and compared with established DL architectures, including AlexNet, Xception, InceptionV3, and ResNet50. Performance is assessed using standard evaluation metrics such as accuracy, precision, recall, F1-score, specificity, AUC-ROC, Matthews correlation coefficient (MCC), log loss, and inference time. Experimental results indicate that the OOA-optimized VGG16 model achieves superior performance, with an accuracy of 97.7%, precision of 96.71%, recall of 97.79%, AUC-ROC of 99.92%, and MCC of 0.9449, while maintaining competitive computational efficiency. The results demonstrate that bio-inspired hyperparameter optimization significantly enhances classification reliability and diagnostic accuracy. In summary, integrating OOA optimization with the VGG16 architecture yields a dependable framework for breast cancer identification, making it a promising candidate for deployment in automated diagnostic support systems.
Comparative analysis of unidirectional and bidirectional RNNs for ECG arrhythmia detection using augmented MIT-BIH data Sabura Banu Urundai Meeran; Nafeena Abdul Munaf
Bulletin of Electrical Engineering and Informatics Vol 15, No 1: February 2026
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eei.v15i1.10893

Abstract

Accurate classification of electrocardiogram (ECG) signals is essential for early arrhythmia detection. This study compares the performance of unidirectional and bidirectional recurrent neural networks (RNN), specifically gated recurrent unit (GRU)-based architectures, for classifying ECG beats as normal or arrhythmic. ECG data were sourced from the MIT-BIH Arrhythmia Database using the WFDB toolkit. Each beat was segmented into a 128-sample window centered on the R-peak and labeled into two classes. To address severe class imbalance (6,279 normal vs. 43 arrhythmic beats), data augmentation techniques—jittering and scaling—were applied, resulting in a balanced dataset. Both models were trained under identical conditions, with evaluation based on accuracy, precision, recall, F1-score, and other statistical metrics. The unidirectional RNN achieved poor recall (9.0%) despite high precision, yielding an overall accuracy of 54.0%. In contrast, the bidirectional RNN significantly outperformed, achieving 98.17% accuracy, 98.39% precision, 97.92% recall, and a 98.16% F1-score. The results demonstrate that bidirectional temporal modeling provides substantial improvements in ECG classification, especially for detecting minority class arrhythmias. This study highlights the importance of both data augmentation and model architecture in developing effective deep learning solutions for real-time ECG analysis and clinical diagnostics.
Cross-modality brain image translation using CycleGAN for MRI to CT and CT to MRI conversion Sabura Banu Urundai Meeran; Nafeena Abdul Munaf
Bulletin of Electrical Engineering and Informatics Vol 15, No 4: August 2026
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eei.v15i4.10892

Abstract

Accurate cross-modality brain image translation between magnetic resonance imaging (MRI) and computed tomography (CT) can enhance diagnosis and treatment planning by combining complementary structural and tissue information. This study develops a cycle-consistent generative adversarial network (CycleGAN)-based framework for unpaired MRI-to-CT and CT-to-MRI conversion, eliminating the need for paired datasets. Publicly available brain imaging datasets were preprocessed with intensity normalization and resizing, and the model was trained using a U-Net generator for anatomical preservation and a PatchGAN discriminator for texture fidelity. Quantitative evaluation using structural similarity index measure (SSIM), peak signal-to-noise ratio (PSNR), and mean absolute error (MAE) demonstrated that the proposed method outperforms baseline generative adversarial network (GAN) architectures, while expert radiologist reviews confirmed high structural consistency. The approach effectively maintains anatomical integrity and visual realism across modalities. This work’s novelty lies in optimizing CycleGAN architecture and hyperparameters for brain imaging translation, achieving superior performance without paired scans. The findings indicate potential for integration into clinical workflows, particularly in multimodal diagnosis and radiotherapy planning, offering a pathway to reduce scan redundancy and improve patient care. Unlike prior unpaired image-translation studies, this work introduces a U-Net–enhanced CycleGAN with optimized loss weighting and architecture tailoring for brain imaging, achieving improved structural fidelity and computational efficiency.
Sailfish optimized MobileNet for robust recognition of American sign language alphabets and digits Sabura Banu Urundai Meeran; Selvaraj Kavitha; Balakrishnan Chinthamani; Balasubramanian Suresh Chander Kapali
Bulletin of Electrical Engineering and Informatics Vol 15, No 4: August 2026
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eei.v15i4.11043

Abstract

American sign language (ASL) is widely used for communication among the deaf and hard-of-hearing communities. This study aims to develop an optimized deep learning (DL) model for recognizing 36 static ASL gestures representing English alphabets (A–Z) and digits (0–9). A publicly available ASL dataset was used, and five convolutional neural network (CNN) architectures—AlexNet, GoogleNet, Inception V3, MobileNet, and ResNet50—were implemented as baselines. MobileNet was further optimized using the sailfish optimization (SFO) algorithm to fine-tune key hyperparameters and architectural settings. The models were evaluated using accuracy, macro precision, recall, F1-score, specificity, Cohen’s Kappa, Matthews correlation coefficient (MCC), balanced accuracy, Jaccard index, and error rate. The SFO-enhanced MobileNet achieved the highest performance, with 98.28% accuracy, 98.27% macro F1-score, 99.95% specificity, and a 1.72% error rate, outperforming all baselines across metrics. These results demonstrate that SFO optimization significantly improves MobileNet’s ability to classify ASL gestures accurately and efficiently. The proposed model’s high accuracy, robustness, and low inference time (22 ms) make it suitable for real-time sign language interpretation tools and assistive communication devices, supporting broader accessibility and inclusivity.