Scholarlink

<p>Posterior Cortical Atrophy (PCA) is a neurodegenerative disease of the brain that is identified by the regular degeneration of the visual processing process and other higher-order brain activities that mostly involve the posterior parts of the brain. PCA continues to be difficult to diagnose due to its similarity with other forms of neurodegenerative conditions, in addition to the absence of any specific diagnostic criteria. The existing imaging methods are effective at identifying cortical thinning and hypometabolism but not necessarily inclusive at the initial stages. The study presents a novel TOSS-PCA model for classifying PCA with two types of inputs: MRI images and EEG signals. The proposed TOSS-PCA model integrates with advanced pre-processing, feature extraction, selection, and classification techniques. Initially, MRI images undergo preprocessing with an adaptive trilateral filter to enhance image quality and simultaneously, EEG signals are denoised with discrete wavelet transform (DWT) to remove noise. The deep learning based NASNet is used to extract relevant features from the noise-free images, and the Bi-GRU is used to capture temporal dependencies of the denoised EEG signals. The Tyrannosaurus optimization (T-Rex) algorithm is used for selecting the most fine-grained characteristics from the retrieved features. The multilayer perceptron classifies the selected features into normal and abnormal categories by demonstrating the model’s efficacy in handling dual-input data. According to the result, the proposed TOSS-PCA model achieves a success rate of 99.30% for the PCA classification. The proposed TOSS-PCA model outperforms FPN, DA-Net, DMN and SS-OCTA in terms of overall accuracy by 0.57%, 6.3%, 1.17%, and 11.30% respectively.</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

TOSS-PCA: Tyrannosaurus Optimized Features Based Posterior Cortical Atrophy Classification

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月29日

查看全文
  • Sabitha Rajagopal,
  • Kalaiselvi Kuppuswamy,
  • Karthik Subburathinam,
  • Kavitha Mettupalayam Subramaniam

摘要

Posterior Cortical Atrophy (PCA) is a neurodegenerative disease of the brain that is identified by the regular degeneration of the visual processing process and other higher-order brain activities that mostly involve the posterior parts of the brain. PCA continues to be difficult to diagnose due to its similarity with other forms of neurodegenerative conditions, in addition to the absence of any specific diagnostic criteria. The existing imaging methods are effective at identifying cortical thinning and hypometabolism but not necessarily inclusive at the initial stages. The study presents a novel TOSS-PCA model for classifying PCA with two types of inputs: MRI images and EEG signals. The proposed TOSS-PCA model integrates with advanced pre-processing, feature extraction, selection, and classification techniques. Initially, MRI images undergo preprocessing with an adaptive trilateral filter to enhance image quality and simultaneously, EEG signals are denoised with discrete wavelet transform (DWT) to remove noise. The deep learning based NASNet is used to extract relevant features from the noise-free images, and the Bi-GRU is used to capture temporal dependencies of the denoised EEG signals. The Tyrannosaurus optimization (T-Rex) algorithm is used for selecting the most fine-grained characteristics from the retrieved features. The multilayer perceptron classifies the selected features into normal and abnormal categories by demonstrating the model’s efficacy in handling dual-input data. According to the result, the proposed TOSS-PCA model achieves a success rate of 99.30% for the PCA classification. The proposed TOSS-PCA model outperforms FPN, DA-Net, DMN and SS-OCTA in terms of overall accuracy by 0.57%, 6.3%, 1.17%, and 11.30% respectively.