Suhas M.V

Mr. Suhas M.V

Electronics and Communication Engineering, Manipal Institute of Technology, Manipal Academy of Higher Education, Manipal, India
orcid

Research Focus & Metrics

Academic Career Length

11Years (2016-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
14 August 2023 in Brain Sciences
Verified
Journal Article
Open access
PDF file
ULUtkarsh Lal
Utkarsh Lal
SMSuhas M.V
Suhas M.V
A.Anitha .H
Anitha .H
Accurate sleep stage detection is crucial for diagnosing sleep disorders and tailoring treatment plans. Polysomnography (PSG) is considered the gold standard for sleep assessment since it captures a diverse set of physiological signals. While various studies have employed complex neural networks for sleep staging using PSG, our research emphasises the efficacy of a simpler and more efficient architecture. We aimed to integrate a diverse set of feature extraction measures with straightforward machine learning, potentially offering a more efficient avenue for sleep staging. We also aimed to conduct a comprehensive comparative analysis of feature extraction measures, including the power spectral density, Higuchi fractal dimension, singular value decomposition entropy, permutation entropy, and detrended fluctuation analysis, coupled with several machine-learning models, including XGBoost, Extra Trees, Random Forest, and LightGBM. Furthermore, data augmentation methods like the Synthetic Minority Oversampling Technique were also employed to rectify the inherent class imbalance in sleep data. The subsequent results highlighted that the XGBoost classifier, when used with a combination of all feature extraction measures as an ensemble, achieved the highest performance, with accuracies of 87%, 90%, 93%, 96%, and 97% and average F1-scores of 84.6%, 89%, 90.33%, 93.5%, and 93.5% for distinguishing between five-stage, four-stage, three-stage, and two distinct two-stage sleep configurations, respectively. This combined feature extraction technique represents a novel addition to the body of research since it achieves higher performance than many recently developed deep neural networks by utilising simpler machine-learning models.

1 May 2016
Verified
Conference Paper
Closed access
AMAnindita Mishra
Anindita Mishra
SMSuhas M.V
Suhas M.V
Bones form the supporting framework of the body. It has a hard outer layer made of compact (cortical) bone that covers a lighter spongy (trabecular) bone inside. Osteoblast (cell that lays down new bone) and osteoclast (cell that dissolves old bone) are the two types of cells present in the bone. Throughout our lifetime, new bone keeps replacing the dissolving old bone. An uncontrollable division of these cells along with fat cells and blood forming cells in the bone marrow could destroy surrounding body tissue causing bone cancer. This work presents a Computer Aided Diagnosis (CAD) system that helps radiologists in differentiating malignant and benign bone lesions in the spine on CT images. Firstly, the lesions are segmented using active contour models and then texture is analyzed through second order statistical measurements based on the Gray Level Co-occurrence Matrix (GLCM). We use features like autocorrelation, contrast, cluster shade, cluster prominence, energy, maximum probability, variance and difference variance to train and test the Random Forest. The aim of this paper is to discuss a technique that improves the sensitivity, specificity and accuracy of detecting the bone lesions.

18 October 2024 in IEEE Access
Verified
Journal Article
Open access
BSB. P. Swathi
B. P. Swathi
GMGeetha Maiya
Geetha Maiya
GAGirija Attigeri
Girija Attigeri
SMSuhas M.V
Suhas M.V
SHSrinithya Halaharvi
Srinithya Halaharvi
Question Answering (QA) systems are increasingly essential in educational institutions, enhancing both learning and administrative processes by providing quick and accurate answers to user queries. However, existing systems often struggle with accurately classifying and responding to diverse and context-dependent questions, especially when dealing with large knowledge graphs. Predicting the domain of a question can significantly narrow down the search space within a vast knowledge graph, improving the system’s efficiency and accuracy. This study addresses this gap by developing and evaluating domain prediction models. We compare the performance of various deep learning architectures, including Bi-GRU, Bi-LSTM, GRU, and LSTM. Our results demonstrate that the 1-layer Bi-GRU model outperforms the others, achieving the highest test accuracy of 82.13%. Additionally, by employing an ensemble technique that combines models with highest performance measures from each architecture, we further enhance overall performance, achieving an accuracy of 87.14%, which demonstrates improved predictive capability. This work is significant as it provides a robust solution for improving the accuracy and relevance of QA systems in educational settings, thereby enhancing user satisfaction and operational efficiency.

1 May 2016
Verified
Conference Paper
Closed access
RKRishav Kumar
Rishav Kumar
SMSuhas M.V
Suhas M.V
Skeletal metastasis has tendency to develop from any kind of primary tumor. In the spine, the vertebral body is the most common site of metastasis which then extends to pedicle. About 2/3rd of the malignant tumor cases are found to develop metastasis. This work presents a Computer Aided Diagnosis (CAD) system that helps radiologists in differentiating malignant and benign bone lesions in the spine on Computed Tomography (CT) images usingSupport Vector Machines(SVM). The CT images are segmented using Snakes or Active Contour Model to retrieve the Region of Interest(ROI). From the segmented images, Haralick features are calculated. These features are then passed to the SVM classifier. With the help of SVM model generated, the data are classified into benign and malignant nodules. The performances of different kernel functions are compared.

5 April 2024 in IEEE Access
Verified
Journal Article
Open access
SMSuhas M.V
Suhas M.V
SSSanjib Sinha
Sanjib Sinha
NMN. Mariyappa
N. Mariyappa
A.Anitha .H
Anitha .H
KKKarunakar A K
Karunakar A K
The significance of fiducial marker detection in neuroimaging cannot be overstated, as these markers serve as vital reference points for accurate spatial alignment during image registration. Our proposed model addresses challenges in consistent marker placement and variability during MRI scanning, ensuring reliable localization for subsequent analysis. This paper introduces an innovative approach to fiducial marker detection in T1-weighted MRI volumes, specifically targeting the Left Preauricular Point (LPA), Right Pre-auricular Point (RPA), and Nasion. The implementation employs a 3D Convolutional Neural Network (CNN) to achieve precise localization of these crucial anatomical landmarks. Operating on a high-performance system our algorithm demonstrated exceptional accuracy and sensitivity using MATLAB R2023a as the primary tool for development and evaluation. Rigorous experiments on a diverse dataset showcased the algorithm’s robust performance. For RPA detection, the model achieved 96.55% accuracy, emphasizing sensitivity (96.78% recall) and precision (96.35%). LPA detection demonstrated an impressive accuracy of 96.88%, with heightened sensitivity (96.95%) and precision (96.83%). The nasion detection process exhibited precise localization, with a Mean Square Error (MSE) of 0.3439 for 36 volume data. These results highlight the algorithm’s potential to enhance accuracy and efficiency in fiducial point detection for improved neuroimaging studies.