Scholarlink

<p>One of the life-threatening parasitic diseases named Visceral Leishmaniasis (VL) needs early and precise diagnosis to lessen severe morbidity and mortality. Before the use of computer based diagnosis, conventional methods of diagnostics including microscopic examination and serological assays, suffer from low sensitivity, time-intensive procedures, and human dependency that might lead to misdiagnosis. Most accurate prediction rate of true positive detection through Microscopic Cartography of Blood Cells is offered by Dense Vision Transformers (DenseViT). Due to their potential of capturing intricate hematological patterns that indicates the presence of VL by optimal feature selection and convergence tuning. In this paper, we introduced DViT-GKO-VL framework in which DenseViT with the Gravity Kinematic Approach (GKA), having the deep insight for peripheral blood microscopy in VL detection. DenseViT efficiently extracts multi-scale spatial representations from blood smear images, ensuring robust feature learning. After the extraction of full feature space, GKA is employed which is inspired by gravitational kinematics for obtaining optimized feature space by directing the model's focus towards diagnostically relevant cellular structures, mitigating overfitting and computational redundancy. The DViT-GKO-VL framework is evaluated on peripheral blood smear datasets using metrics such as accuracy, precision, recall, and F1-score to validate its efficacy and highlights higher detection rates and reduced false positives. This study underscores the transformative potential of microscopic cartography, deep vision transformers, and swarm-inspired optimization, revolutionizing automated blood pathology with precision and efficiency.&#xa0;While validated on VL, the scalability of this framework suggests its adaptability to other medical domains, including neurodegenerative disorders such as Alzheimer’s disease, where early detection depends on recognizing subtle patterns in imaging and electrophysiological data.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

DViT-GKO: Dense Vision Transformer with Gravity Kinematic Optimization for Automated Disease Diagnosis – Application to Visceral Leishmaniasis and Future Scope in Neurodegenerative Disorders

  • 期刊论文

  • 发表日期：

    2025年9月6日

查看全文
  • Bandaru A. Chakravarthi,
  • Gandla Shivakanth

摘要

One of the life-threatening parasitic diseases named Visceral Leishmaniasis (VL) needs early and precise diagnosis to lessen severe morbidity and mortality. Before the use of computer based diagnosis, conventional methods of diagnostics including microscopic examination and serological assays, suffer from low sensitivity, time-intensive procedures, and human dependency that might lead to misdiagnosis. Most accurate prediction rate of true positive detection through Microscopic Cartography of Blood Cells is offered by Dense Vision Transformers (DenseViT). Due to their potential of capturing intricate hematological patterns that indicates the presence of VL by optimal feature selection and convergence tuning. In this paper, we introduced DViT-GKO-VL framework in which DenseViT with the Gravity Kinematic Approach (GKA), having the deep insight for peripheral blood microscopy in VL detection. DenseViT efficiently extracts multi-scale spatial representations from blood smear images, ensuring robust feature learning. After the extraction of full feature space, GKA is employed which is inspired by gravitational kinematics for obtaining optimized feature space by directing the model's focus towards diagnostically relevant cellular structures, mitigating overfitting and computational redundancy. The DViT-GKO-VL framework is evaluated on peripheral blood smear datasets using metrics such as accuracy, precision, recall, and F1-score to validate its efficacy and highlights higher detection rates and reduced false positives. This study underscores the transformative potential of microscopic cartography, deep vision transformers, and swarm-inspired optimization, revolutionizing automated blood pathology with precision and efficiency. While validated on VL, the scalability of this framework suggests its adaptability to other medical domains, including neurodegenerative disorders such as Alzheimer’s disease, where early detection depends on recognizing subtle patterns in imaging and electrophysiological data.