Intelligent Swin Transformer With DeepLAB v3 Hybrid Network for Brain Tumor Classification

The categorization of brain tumors is a key determinant in medical diagnosis particularly in the utilization of MRI scans in the initial diagnosis and adequate classification of different forms of tumors. The hybrid deep learning model, in its turn, is proposed in this research and is founded on integrating the high-resolution Swin Transformer (HR-Swin) and DeepLabv3 to classify brain tumors and improve the quality of tumor detection and segmentation. The HR-Swin Transformer model is appropriate in the sense that it can be used to delineate the local and global contextual information that is intrinsically difficult in the medical images, due to the heterogeneity of tumors and complex shapes. MRI data that is labeled and has four classes of brain tumors; glioma, meningioma, pituitary tumor and no tumor is used to train the model. The model performances in terms of accuracy classifications of 93 and specific tumor delineation with high precision values of 93 as symbolized by the macro and weighted averages of the values of the precision, recall and F1-score values. Based on the results of the confusion matrix, it can be concluded that the model is of good performance particularly in the distinction of the different pituitary tumors and gliomas, although it remains dependable in identifying any kind of tumor. The research has a significance to the development of the differentiated and powerful diagnostic tools of the clinical utilization of the classification of brain tumor in the medical imaging system.

Ҳали таржима қилинмаган