The goal of this observational study is to learn whether the artificial intelligence method can automatically identify and diagnose renal lesions using non-contrast CT or opportunistic screening.

This study first establishes an AI model capable of effectively detecting and diagnosing kidney lesions based on a multicenter retrospective cohort study. Then, the AI model is applied to a large-scale real-world retrospective and prospective population to validate and improve its effectiveness.

Inclusion Criteria:

1. Patients who underwent an abdominal CT examination.
2. Patients with renal lesions were managed according to standard clinical pathways, which included follow-up, biopsy, or surgery.
3. Malignant lesions were pathologically confirmed; benign lesions were confirmed by either pathological diagnosis or imaging follow-up.
4. No prior treatment had been received for the renal disease.

Exclusion Criteria:

1. Patients refuse to undergo recommended follow-up, biopsy, or surgery, which precluded definitive diagnosis of the renal lesion.
2. Absence of complete pathological confirmation for lesions suspected to be malignant.
3. Patients have received any form of prior treatment for the renal lesion.
4. Poor image quality that hampered diagnostic evaluation.

Patients who underwent an abdominal CT examination.