BCSD-2024

Published: 12 September 2024| Version 1 | DOI: 10.17632/fc72nbt5y2.1
Contributors:
JAVARIA Amin,

Description

The proposed BCSD-2024 dataset comprises 10 patients of mammography. The left and right breast are manually marked with the assistance of trained radiologists. The affected region is marked manually, and a binary mask is created for each slice. The dataset's performance is assessed using the proposed spatial attention mechanism (SAM) for breast lesion segmentation. The SAM model, trained from scratch with the chosen hyperparameters, achieves an intersection over union (IoU) of 0.75.

Files

Steps to reproduce

NOTHING

Categories

Breast Cancer

Licence