Fig. 1: Examples from the four evaluation datasets (LIDC-IDRI, LNDb, MosMedData, and NSCLC-Radiomics). | npj Digital Medicine

Fig. 1: Examples from the four evaluation datasets (LIDC-IDRI, LNDb, MosMedData, and NSCLC-Radiomics).

From: CoreFormer high fidelity pulmonary nodule segmentation with structural core priors and geodesic implicit fields

Fig. 1

The significant visual differences in lesion appearance-ranging from small, well-defined nodules to large, complex tumors and diffuse infectious lesions-highlight the domain gap across the datasets, providing a robust benchmark for evaluating the model’s cross-domain generalization performance.

Back to article page