Transforming Breast Cancer İmage Classification With Vision Transformers and LSTM İntegration

AI-powered computer-aided diagnostic tools allow radiologists to make a second, more accurate cancer diagnosis, advancing healthcare analytics. This study compares ultrasonography and histology for breast lesion diagnosis. Convolutional neural networks (CNNs) have successfully extracted information from visual input using their convolution kernels. Bidirectional Encoder Representations from Transformers is an NLP standard. In addition to segmentation and classification, this instrument encodes features. Vision transformers (ViT) are optimized for transfer learning when pre-trained with BERT using image transformers (BEiT) for feature encoding. Data categorization is done using an RNN-LSTM with encoded characteristics. High precision is attained by both methods. The dataset was 99.2% correct for breast histopathology photos.