1. Accurate Bladder Cancer Classification and Prognosis with a Hybrid Vision Transformer.
- Author
-
Alkhalidy, Roaa, AlShemmary, Ebtesam, and Zhentai Lu
- Subjects
TRANSFORMER models ,IMAGE recognition (Computer vision) ,CONVOLUTIONAL neural networks ,ARTIFICIAL intelligence ,BLADDER cancer - Abstract
Bladder cancer, a common and increasingly prevalent malignant neoplasm, has become a major threat to public health. The major significance enjoyed by this combination is that it emphasizes the need to comprehend and diagnose a condition at a very young age. A number of automatic diagnostic techniques employing Artificial Intelligence technology have been deployed in the area of bladder cancer and have generally been found to be effective but the actual percentage accuracy required for real-world diagnosis has not been achieved yet. The proposed hybrid model integrates Vision Transformers and Convolutional Neural Networks to enhance the performance of medical vision tasks, such as the classification and prediction of bladder cancer. Vision Transformers excel in capturing long-range relationships through self-attention mechanisms, while Convolutional Neural Networks are more effective in extracting local features using spatial convolution filters. This combination leverages the strengths of both architectures to achieve superior results in medical image analysis. The Endoscopy dataset, and Pathological dataset were used and the hybrid model provided better accuracy for diagnosing bladder cancer than the basic CNN model and transferred learning, VGG16, Inception-v3, MobileNetV2 of 99.93%, 93.67%, 97.8%, 95.5%, 95% respectively. The experiments show that these hybrid architectures are more effective than the conventional Convolutional Neural Networks for the Bladder cancer Image classification and establish the suitability of Vision Transformer in progressing towards higher performance with the Bladder cancer Image analysis. [ABSTRACT FROM AUTHOR]
- Published
- 2025
- Full Text
- View/download PDF