Scholarlink

Parkinson’s Disorder (PD) is a brain disorder that has two main characteristics and they are degeneration and progressiveness. PD has no cure. Capturing early symptoms using appropriate modalities will yield a better diagnosis. Bradykinesia (slowness of movement) and rest tremor are the prominent early symptoms of PD. The Spiral Pencil Sketch Test (SPST) (image) aids in capturing bradykinesia and rest tremor symptoms. Integrating SPST images data with Artificial Intelligence (AI) provides accurate and robust PD diagnosis. The proposed system presents a novel PD diagnosis framework based on the state-of-the-art AI model, the Densely Connected Convolution Network (DenseNet 201) with Image Net weights integrated into the data for the differentiation of Class 1 (PD subjects) from class 0 (healthy subjects). The testing of the proposed system is done using retrieved SPST image dataset. The proposed system is subjected to different experiments to obtain the best results for PD diagnosis. The highest average accuracy of 93% is obtained as the results of the proposed methodology.

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

Spiral Drawings Test Based Early Diagnosis of Parkinson Disorder Using Tuned Densenet201 Deep Transfer Learning Model

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • C. D. Anisha,
  • N. Arulanand

摘要

Parkinson’s Disorder (PD) is a brain disorder that has two main characteristics and they are degeneration and progressiveness. PD has no cure. Capturing early symptoms using appropriate modalities will yield a better diagnosis. Bradykinesia (slowness of movement) and rest tremor are the prominent early symptoms of PD. The Spiral Pencil Sketch Test (SPST) (image) aids in capturing bradykinesia and rest tremor symptoms. Integrating SPST images data with Artificial Intelligence (AI) provides accurate and robust PD diagnosis. The proposed system presents a novel PD diagnosis framework based on the state-of-the-art AI model, the Densely Connected Convolution Network (DenseNet 201) with Image Net weights integrated into the data for the differentiation of Class 1 (PD subjects) from class 0 (healthy subjects). The testing of the proposed system is done using retrieved SPST image dataset. The proposed system is subjected to different experiments to obtain the best results for PD diagnosis. The highest average accuracy of 93% is obtained as the results of the proposed methodology.