"SegAugmentBreastImg" is a project focused on developing and optimizing segmentation and data augmentation techniques for medical images related to the breast sound region. The primary goal is to enhance the performance of deep learning models by providing tailored data augmentation methods and achieving more accurate segmentation results.
Install the libraries below to use the model.
PyTorch and Torchvision => https://pytorch.org/ (the latest and GPU version are recommended) Pytorch Image Model => https://pypi.org/project/timm/ Transformers => https://huggingface.co/docs/transformers/installation
python main.py --model=1 --dataset_name='KVASIR-SEG'
- Breast Ultrasound Dataset BUS-BRA ( 8/2023)
4 scanners during studies at the National Institute of Cancer in Brazil. 1875 anonymized images from 1064 female patients. Biopsy-proven tumors: 722 benign and 342 malignant cases. BI-RADS assessments, the tumors into categories 2 to 5. 5-fold and 10-fold cross-validation. PNG format files, metadata. Masks and BBOX information. *Wilfrido Gómez-Flores, Maria Julia Gregorio-Calas, and Wagner Coelho de Albuquerque Pereira, "BUS-BRA: A Breast Ultrasound Dataset for Assessing Computer-aided Diagnosis Systems," Medical Physics, vol. 51, no. 4, pp. 3110-3123, 2024. (DOI http://doi.org/10.1002/mp.16812).
Results