This retrospectively acquired data includes contrast enhanced CT imaging studies of 53 confirmed ACC patients between 2006 to 2018 with available clinical and pathological data, including Ki-67 index. Semi-automatic segmentation of the adrenal tumor was created using AMIRA, then manually refined by an experienced radiologist. The segmentations of each contrast-enhanced CT were done for the purpose of radiomic features extraction.

The participants in this dataset fulfilled these inclusion criteria:

(1) Pathologically proven Adrenocortical carcinoma,

(2) Underwent surgical resection of the tumour,

(3) the Ki-67 index was determined as part of the histopathological evaluation of the resected tissue,

(4) imaging data (pre-resection contrast-enhanced CT of the abdomen) were available. Voxel level segmentation of the adrenal lesion are included as well.

Data from patients whose Ki-67 was quantified in biopsied tissue samples rather than from resected whole tumor, were excluded from this study. This exclusion was based on previous studies concluding that Ki-67 quantification should be based on tissue samples collected from the whole tumour.

This dataset can serve as a training set for machine learning algorithms for various purposes. We used the radiomic features extracted to predict the Ki-67 index (through regression) without the need of surgical intervention; features also can be used for multiple purposes, for example, segmentation and classification of adrenal tumors. In addition, there is no prior public available library for adrenal lesions, and this should be important to the scientific community.