Fig. 4
From: The impact of pre-processing techniques on deep learning breast image segmentation
Original image of a patient’s slice from the Duke-Breast-Cancer-MRI dataset, showing: (a) post-processed image using the best-performing model, DS with input size 512 x 512 x 64, orientation standardized and Z-Score as intensity normalization, with labels (breast in blue, fibroglandular tissue in yellow, and vessels in orange) and the corresponding model predictions; (b) post-processed image using the least effective model, DS with input size of 256 x 256 x 64, spacing standardization and HistNorm of intensity normalization, with the same ground truth annotations and the corresponding model predictions.
