Scholarlink

<p>To enhance patient outcomes and survival rates, early and precise detection of lung and colon diseases is critical. This research describes a novel deep-learning strategy for accurately detecting different diseases using medical imaging data. The suggested method seamlessly combines Convolutional Neural Networks (CNNs) with Recurrent Neural Networks (RNNs). CNNs thrive in classifying individual computed tomography (CT) scan slices, but RNNs excel at capturing the data’s inherent temporal correlations. The new CNN-RNN model outperformed state-of-the-art techniques consistently, raising the bar high for lung and colon disease detection. The following CNN architectures were used in the experiments: DenseNet201, EfficientNetB3, Inception v3, InceptionResNetV2, Xception and MobileNetV3Large. Bi-LSTM and Bi-GRU modules were also used for sequence modelling. It should be mentioned that the application of Xception, EfficientNetB3 and MobileNetV3Large architectures along with Bi-GRU and Bi-LSTM led to a fantastic 99.97% accuracy in the experiments on Lung, and Colon Cancer Histopathological Image Dataset (LC25000).</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

Hybrid Convolution-Recurrent Deep Learning Framework for High-Precision Lung and Colon Disease Detection

  • 期刊论文

  • 发表日期：

    2025年12月4日

查看全文
  • Anirudh Singh,
  • Satyam Kumar,
  • Jayesh Gangrade,
  • Shweta Gangrade,
  • Harish Kumar Shakya

摘要

To enhance patient outcomes and survival rates, early and precise detection of lung and colon diseases is critical. This research describes a novel deep-learning strategy for accurately detecting different diseases using medical imaging data. The suggested method seamlessly combines Convolutional Neural Networks (CNNs) with Recurrent Neural Networks (RNNs). CNNs thrive in classifying individual computed tomography (CT) scan slices, but RNNs excel at capturing the data’s inherent temporal correlations. The new CNN-RNN model outperformed state-of-the-art techniques consistently, raising the bar high for lung and colon disease detection. The following CNN architectures were used in the experiments: DenseNet201, EfficientNetB3, Inception v3, InceptionResNetV2, Xception and MobileNetV3Large. Bi-LSTM and Bi-GRU modules were also used for sequence modelling. It should be mentioned that the application of Xception, EfficientNetB3 and MobileNetV3Large architectures along with Bi-GRU and Bi-LSTM led to a fantastic 99.97% accuracy in the experiments on Lung, and Colon Cancer Histopathological Image Dataset (LC25000).