A Transformer-Based Method for Semantic Segmentation of Mitosis in Breast Histopathology Images
Küçük Resim Yok
Tarih
2025
Yazarlar
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Institute of Electrical and Electronics Engineers Inc.
Erişim Hakkı
info:eu-repo/semantics/closedAccess
Özet
According to the World Health Organization's 2022 reports, approximately 2.3 million women globally have received a diagnosis of breast cancer. The detection of mitosis is critical for assessing tumor proliferation rates, which serve as vital biomarkers in breast cancer diagnosis. Consequently, numerous studies have concentrated on the development of deep learning models to automate the identification of mitotic cells in breast cancer tissues. Deep learning methodologies such as U-Net and Mask R-CNN have demonstrated effectiveness in enhancing mitosis detection within breast cancer histopathology images. In this study, SAM and SegFormer, which are based on transformer architecture and have recently shown superior performance, were used. The ICPR14 dataset, an open-access breast histopathology dataset, was employed for the training of SegFormer. Given that the ICPR14 dataset is originally an object detection dataset, it was converted into a segmentation dataset utilizing SAM. Six distinct experiments were conducted to ascertain the optimal performance for mitosis detection. The experiments analyzed the influence of varying encoder sizes on model performance. Evaluation metrics including precision, recall, F1 score, and Intersection over Union (IoU) were employed for the comparative analysis of experimental results. The findings revealed that the Segformer-B4 model, which utilized a channel size of 512 and the MiT-B4 encoder, attained the highest performance metrics. Specifically, the Segformer-B4 model achieved precision of 0.8994, recall of 0.8931, F1 score of 0.8962, and IoU of 0.9063. These results indicate that the SegFormer approach can effectively facilitate mitosis detection in breast histopathology images. © 2025 IEEE.
Açıklama
9th International Artificial Intelligence and Data Processing Symposium, IDAP 2025 -- 6 September 2025 through 7 September 2025 -- Malatya -- 215321
Anahtar Kelimeler
breast histopathology, mitosis detection, segFormer, semantic segmentation, transformer
Kaynak
9th International Artificial Intelligence and Data Processing Symposium, IDAP 2025
WoS Q Değeri
Scopus Q Değeri
N/A
