Dr. Anindya Apriliyanti Pravitasari  

Universitas Padjadjaran, Indonesia

Anindya Apriliyanti Pravitasari (Mem- ber, IEEE) was born in Tuban, Indonesia, in 1984. She received the B.S. (Sarjana) and M.Si. degrees in statistics and the Ph.D. degree in statistics science from Institut Teknologi Sepuluh Nopember Surabaya, Indonesia, in 2006, 2008, and 2020, respectively.,She currently holds the positions of a Lecturer and a Researcher with Universitas Padjadjaran, where her primary research focuses on data mining, statistical machine learning, Bayesian analysis, natural language processing, and computer vision. As part of her academic responsibilities, she is actively involved in teaching with the Department of Statistics, Universitas Padjadjaran. Additionally, she serves as the Director of the Research Center for Artificial Intelligence and Big Data. Her expertise and contributions to the field have earned her recognition, as she provides her referee services to reputable journals, including IEEE Access and European Radiology. Moreover, her dedication to research has been acknowledged with honors such as high-quality scientific articles in health and medicine from the Ministry of Research, Technology, and Higher Education, Indonesia. Her research findings were also selected as one of the Top 20 national posters in artificial intelligence by the ministry, contributing to the advancement of Indonesian artificial intelligence innovation. She is deeply engaged in extensive research in the fields of big data and AI, continuously pushing the boundaries of knowledge and making a significant impact in the academic and research communities.

Topic: Deep Detection Meets Spatial Clustering: A YOLO–ScNenomimo Hybrid for MRI Brain Tumor Localization and Segmentation

Abstract: Accurate brain tumor segmentation from magnetic resonance imaging (MRI) is essential for clinical diagnosis, treatment planning, and radiotherapy. Nevertheless, conventional deep learning models often require extensive annotated datasets and face difficulties in delineating tumors with irregular boundaries and intensity inhomogeneity. To address these challenges, this study proposes a hybrid YOLO–ScNenomimo framework that integrates the fast and precise tumor localization capability of YOLO with the spatially regularized unsupervised segmentation power of ScNenomimo. In this framework, YOLO is first employed to detect the region of interest (ROI) corresponding to the tumor area. Subsequently, Sc-Nenomimo performs fine-grained segmentation within the detected ROI by modeling voxel intensity distributions through a neo-normal mixture formulation while incorporating spatial neighborhood constraints to preserve structural continuity. This hybrid strategy reduces dependence on large-scale manual annotations, lowers computational complexity, and enhances both segmentation precision and spatial coherence. Experimental evaluations on publicly available MRI brain tumor datasets demonstrate that the proposed YOLO–ScNenomimo approach outperforms conventional clustering-based and CNN-based segmentation methods in terms of Dice coefficient, silhouette coefficient, and Fit Density Ratio. The results indicate that combining deep detection with spatially constrained probabilistic clustering provides a robust and interpretable semi-supervised framework for accurate brain tumor segmentation in medical image analysis.