Table 1. Lesion segmentation criteria and steps.
Number | Lesion segmentation criteria and steps |
---|---|
1 | The HE images were automatically transformed into negative images by ITK-SNAP software (version 3.6; www.itksnap.org) to segment the lesions |
2 | Lesion contours (ROIs) were separately delineated on HE, LE and DES images in the CC and MLO views if the lesions were visible on each image. If not, contours were delineated on either HE, LE or DES images depending on which provided the best visualization of the lesion. Then, these contours were mapped onto the other images, ensuring 6 ROIs for each lesion |
3 | For nonmass lesions such as microcalcifications, asymmetries, or architectural distortions on LE images, closed loops were delineated along the edge of the lesions |
4 | For patients with multiple lesions, only the largest lesions with histopathological results were delineated |
HE, high-energy; LE, low-energy; DES, dual-energy subtraction; CC, craniocaudal; MLO, mediolateral oblique; ROI, region-of-interest.