Scholarlink

<p>Alzheimer’s disease (AD) and Parkinson’s disease (PD) are debilitating neurological disorders that significantly impair the quality of life for affected individuals. Early diagnosis and intervention are critical for optimizing patient care and improving treatment outcomes. However, accurately predicting these diseases using medical imaging remains a challenging task. This research addresses the problem of predicting the presence of AD and PD through the analysis of magnetic resonance imaging (MRI) scans. We propose a multi-stage convolutional neural network (CNN) architecture based on the state-of-the-art EfficientNet model to extract and analyze key features from MRI images. To enhance predictive accuracy, we incorporate transfer learning with EfficientNet, allowing for more effective utilization of the available data. We train the model on a comprehensive dataset of MRI scans from AD and PD patients, using data augmentation, cross-validation, and an adaptive learning rate technique to ensure robust generalization. Experimental results show that our EfficientNet-based multi-stage CNN model performs better at distinguishing between healthy and disease-affected brain MRI scans. The EfficientNet B6 model, in particular, performed better than other EfficientNet variants and the Inception V3 model. It improved validation accuracy by 0.29% to 4.25% and reduced validation loss by 22.02% to 3.55% on the 3-class dataset. On the 4-class dataset, the EfficientNet B6 model again performed the best, improving validation accuracy by 1.13% to 6.52% and reducing validation loss by 19.9% to 3.79%. This research offers a promising tool for healthcare professionals and radiologists, helping with the early diagnosis and monitoring of AD and PD.</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

A multi-stage efficientnet based framework for Alzheimer’s and Parkinson’s diseases prediction on magnetic resonance imaging

  • 期刊论文

  • 发表日期：

    2025年1月14日

查看全文
  • K Aditya Shastry,
  • Shilpa Shastry,
  • Aravind Shastry,
  • Sharath M. Bhat

摘要

Alzheimer’s disease (AD) and Parkinson’s disease (PD) are debilitating neurological disorders that significantly impair the quality of life for affected individuals. Early diagnosis and intervention are critical for optimizing patient care and improving treatment outcomes. However, accurately predicting these diseases using medical imaging remains a challenging task. This research addresses the problem of predicting the presence of AD and PD through the analysis of magnetic resonance imaging (MRI) scans. We propose a multi-stage convolutional neural network (CNN) architecture based on the state-of-the-art EfficientNet model to extract and analyze key features from MRI images. To enhance predictive accuracy, we incorporate transfer learning with EfficientNet, allowing for more effective utilization of the available data. We train the model on a comprehensive dataset of MRI scans from AD and PD patients, using data augmentation, cross-validation, and an adaptive learning rate technique to ensure robust generalization. Experimental results show that our EfficientNet-based multi-stage CNN model performs better at distinguishing between healthy and disease-affected brain MRI scans. The EfficientNet B6 model, in particular, performed better than other EfficientNet variants and the Inception V3 model. It improved validation accuracy by 0.29% to 4.25% and reduced validation loss by 22.02% to 3.55% on the 3-class dataset. On the 4-class dataset, the EfficientNet B6 model again performed the best, improving validation accuracy by 1.13% to 6.52% and reducing validation loss by 19.9% to 3.79%. This research offers a promising tool for healthcare professionals and radiologists, helping with the early diagnosis and monitoring of AD and PD.