Deependra Rastogi

Mr. Deependra Rastogi

orcid

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

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

Representative Publications

View All Publications
20 February 2025 in Life
Verified
Journal Article
Open access
PDF file
DRDeependra Rastogi
Deependra Rastogi
PJPrashant Johri
Prashant Johri
Massimo DonelliMassimo Donelli
Massimo Donelli
Lalit KumarLalit Kumar
Lalit Kumar
SBShantanu Bindewari
Shantanu Bindewari
SKSunil Kumar Khatri
Sunil Kumar Khatri
Brain tumor diagnosis is a complex task due to the intricate anatomy of the brain and the heterogeneity of tumors. While magnetic resonance imaging (MRI) is commonly used for brain imaging, accurately detecting brain tumors remains challenging. This study aims to enhance brain tumor classification via deep transfer learning architectures using fine-tuned transfer learning, an advanced approach within artificial intelligence. Deep learning methods facilitate the analysis of high-dimensional MRI data, automating the feature extraction process crucial for precise diagnoses. In this research, several transfer learning models, including InceptionResNetV2, VGG19, Xception, and MobileNetV2, were employed to improve the accuracy of tumor detection. The dataset, sourced from Kaggle, contains tumor and non-tumor images. To mitigate class imbalance, image augmentation techniques were applied. The models were pre-trained on extensive datasets and fine-tuned to recognize specific features in MRI brain images, allowing for improved classification of tumor versus non-tumor images. The experimental results show that the Xception model outperformed other architectures, achieving an accuracy of 96.11%. This result underscores its capability in high-precision brain tumor detection. The study concludes that fine-tuned deep transfer learning architectures, particularly Xception, significantly improve the accuracy and efficiency of brain tumor diagnosis. These findings demonstrate the potential of using advanced AI models to support clinical decision making, leading to more reliable diagnoses and improved patient outcomes.

1 October 2024 in Journal of Advanced Research in Applied Sciences and Engineering Technology
Verified
Journal Article
Open access
PDF file
TMTanisha Madan
Tanisha Madan
SSShrddha Sagar
Shrddha Sagar
DVDeepali Virmani
Deepali Virmani
DRDeependra Rastogi
Deependra Rastogi
Tien Anh TranTien Anh Tran
Tien Anh Tran
Air pollution is a major concern nowadays that needs immediate action. Various acts have already been initiated by the government in controlling it. The harmful gases present in the air are Nitrogen dioxide, Sulphur dioxide, Ozone, and Carbon monoxide that is causing air pollution and are becoming the major cause of harmful diseases. It does not affect human beings only but the entire environment. According to World Health Organization, air pollution is killing seven million people around the world. Planting trees and avoiding using plastics is the possible solution to control air pollution. The air quality index is the measure through which we categorize the air pollution of various cities. In this paper, we will discuss the bagging and boosting method, the extra trees method of machine learning in predicting air quality index, and its comparison with Decision trees. The dataset is collected from the Open Weather Application Program Interface. In this, the prediction of the Air Quality Index is done based on previous data. Time i.e., hourly prediction and space series prediction are done on the dataset. After implementing each algorithm confusion matrix is obtained and from this precision and recall are calculated for each category of the Air quality index ranging from one to five with one being good and five being very unhealthy. It has been observed that Gradient Boosting achieves the highest accuracy with 98.89%.

20 November 2025 in Frontiers in Artificial Intelligence
Verified
Journal Article
Open access
PDF file
DRDeependra Rastogi
Deependra Rastogi
PJPrashant Johri
Prashant Johri
MDMassimo Donelli
Massimo Donelli
TATarun Agarwal
Tarun Agarwal
STShrikant Tiwari
Shrikant Tiwari
PSPushpa Singh
Pushpa Singh
Introduction: Accurate and early detection of brain tumors is critical for effective treatment and improved patient outcomes, yet manual radiological analysis remains time-consuming, subjective, and error-prone. To address these challenges and improve clinical trust in AI systems, this study presents XAI-BT-EdgeNet, an explainable, edge-aware deep learning framework integrated with squeeze-and-excitation (SE) modules for brain tumor detection using MRI scans. Methods: The proposed architecture employs a dual-branch design that fuses high-level semantic features from InceptionV3 with low-level edge representations via an Edge Feature Block, while SE modules adaptively recalibrate feature importance to enhance diagnostic accuracy. To ensure transparency, the model incorporates four XAI techniques—LIME, Grad-CAM, Grad-CAM++, and Vanilla Saliency—which provide interpretable visual justifications for predictions. The framework was trained and evaluated on the Brain Tumor Dataset by Preet Viradiya, comprising 4,589 labeled MRI images divided into Brain Tumor (2,513) and Healthy (2,076) classes. Results: The model achieved 99.58% training accuracy, 99.71% validation accuracy, and 100.00% testing accuracy, alongside minimal loss values of 0.0103, 0.0051, and 0.0026, respectively. These results demonstrate the robustness and precision of the proposed framework in brain tumor classification. Discussion: This work includes the development of a dual-branch CNN architecture that combines semantic and edge features for enhanced classification, the integration of SE modules to highlight clinically significant regions, and the application of multi-method XAI to offer transparent, interpretable outputs for clinical applicability. Overall, XAI-BT-EdgeNet delivers a high-performing, interpretable solution that bridges the gap between deep learning and trustworthy clinical decision-making in brain tumor diagnosis.

26 April 2025 in Universal Research Reports
Verified
Journal Article
Open access
PDF file
MTManish Tripathi
Manish Tripathi
DRDeependra Rastogi
Deependra Rastogi
Predicting patient readmission risk is key to improving healthcare, cutting costs, and making hospital operations more efficient. As hospitals work to provide quality care with limited resources, accurately identifying which patients are likely to be readmitted has become a priority. This study explores how machine learning models—like logistic regression, decision trees, support vector machines, and random forests—can help. Using patient data such as demographics, medical history, treatment patterns, and hospital factors, the models identify the main drivers of readmission risk. By comparing their performance through metrics like accuracy, precision, recall, and ROC curve scores, the study highlights the best-performing approach, showing how machine learning can support early interventions, improve care, and reduce unnecessary readmissions.

31 March 2025 in Journal of Quantum Science and Technology
Verified
Journal Article
Open access
PDF file
JTJustin Rajakumar Maria Thason
Justin Rajakumar Maria Thason
DRDeependra Rastogi
Deependra Rastogi
In the ever-evolving landscape of software development, orchestration of complex release pipelines in the DevOps universe has been a critical element in ensuring seamless and effective delivery cycles. However, the increasing complexity of modern applications—characteristically characterized by high numbers of dependencies, decentralized design, and heterogenous integration tools—represents monumental challenges to effective orchestration of release pipelines. While previous studies have investigated areas such as Continuous Integration (CI) and Continuous Delivery (CD) in isolation, less focus has been put on end-to-end governance of dependencies, automation, and pipeline orchestration as a whole. The current work explains the research gap in the orchestration of intricate release pipelines with a specific emphasis on interdependency management of the different stages of the delivery pipeline. In addition, it places emphasis on the need for advanced automation strategies with dynamic changing capabilities as software environments and architectures continue to evolve. By placing emphasis on its concern with issues of ensuring consistency, reducing downtime, and reducing manual interventions, this research attempts to provide insights that are actionable towards improving continuous delivery workflows. This research examines especially best practices and strategies to automate manual operations, manage complicated interdependencies, and unify diverse DevOps tools to achieve optimum pipeline efficiency. The research provides new ways to manage multiple stages of releases and enable seamless collaboration among different workflows of different teams, which leads to rapid software deployment cycles and enhanced product quality. In all, the findings provide a holistic framework for organizations that desire to optimize their release management processes such that DevOps pipelines can be adaptive, scalable, and efficient against growing software complexity.

1 March 2025 in International Journal of Research in Modern Engineering & Emerging Technology
Verified
Journal Article
Closed access
UMUniversity of Missouri
University of Missouri
LJLakshman Kumar Jamili
Lakshman Kumar Jamili
SKSoham Sunil Kulkarni
Soham Sunil Kulkarni
UCUniversity of California
University of California
DRDeependra Rastogi
Deependra Rastogi
IUIILM University
IILM University
Early detection of breast cancer is critical for improving patient survival rates, and recent advances in artificial intelligence have paved the way for more accurate diagnostic tools. This study introduces an innovative AI-driven workflow that employs the ResNet50 convolutional neural network to analyze histopathological images for early breast cancer detection. The proposed method leverages transfer learning to extract robust features from complex image data, significantly reducing the need for extensive manual annotation. An extensive preprocessing pipeline, including image normalization and data augmentation, was implemented to enhance the model’s generalization capabilities. The workflow integrates segmentation and classification steps, enabling the rapid identification of malignant tissue patterns within histopathological slides. Comparative experiments on publicly available datasets demonstrated that our approach achieves high sensitivity and specificity, outperforming several traditional machine learning techniques. Furthermore, the system’s scalability and potential for integration into clinical environments suggest its viability as a decision-support tool for pathologists. By automating parts of the diagnostic process, the workflow not only reduces diagnostic time but also minimizes inter-observer variability, leading to more consistent outcomes. Limitations related to data diversity and potential integration challenges have been identified, providing clear directions for future research. Overall, this study contributes to the evolving field of medical imaging by offering a robust, efficient, and scalable framework for early breast cancer detection, thereby enhancing the prospects for timely and personalized treatment interventions.