Research Square (Research Square), Год журнала: 2025, Номер unknown
Опубликована: Апрель 3, 2025
Язык: Английский
Research Square (Research Square), Год журнала: 2025, Номер unknown
Опубликована: Апрель 3, 2025
Язык: Английский
Journal of biomimetics, biomaterials and biomedical engineering, Год журнала: 2025, Номер 67, С. 9 - 20
Опубликована: Янв. 15, 2025
In this paper, a modified Bi-directional Convolutional Long Short-Term Memory U-Net (BCDU-Net) neural network is presented, which aims at enhancing medical image segmentation for lung cancer diagnosis. By integrating BConvLSTM in the decoding path and densely connected convolutional layers encoding path, proposed model achieves greater stability precision segmenting CT images. The addition of Batch Normalization (BN) after up-convolutional accelerates convergence speed by six times. A notable feature BCDU-Net its adaptability to different imaging modalities, enabling it generalize across diverse data sources reducing overfitting, limitation seen many existing models. This allows model’s ability be integrated into various clinical environments, ensuring consistent reliable results equipment. Another key contribution enhanced interpretability model, critical improvement when compared with traditional accurately segments complex anatomical structures, such as left right lungs, precisely identifies tumors near central bronchi or trachea airways, crucial treatment planning. was tested on Cancer Imaging Archive (TCIA) dataset from 2017 Lung Segmentation Challenge, achieving Dice Similarity Coefficients (DSC) 97.97 97.73 lung. Overall, demonstrates superior accuracy holds promise broader applications beyond
Язык: Английский
Процитировано
0Research Square (Research Square), Год журнала: 2025, Номер unknown
Опубликована: Апрель 3, 2025
Язык: Английский
Процитировано
0