The goal of this observational study is to develop a CNN-based machine module to predict postoperative fractional renal function in people who are proposed to undergo partial nephrectomy. The main question it aims to answer is:

• Does this machine learning model accurately predict renal function after partial nephrectomy?

This prospective study is conducted to predict postoperative fractional renal function using the perfusion deficit method from a preoperatively established renal arterial perfusion model for people who are proposed to undergo partial nephrectomy. In this study, this prediction method will be compared with the true missing values of renal units on nuclear renal function, eGFR, and CTA. This study aims to evaluate the feasibility of applying the CNN-based model in predicting postoperative renal function after partial nephrectomy and provide high-level clinical evidence for the preoperative integrated diagnostic and treatment process of renal tumors, especially in terms of the functional evaluation.

Inclusion Criteria:

• people with stage cT1 renal tumors confirmed by preoperative CT or MR
• people who are proposed to undergoing partial nephrectomy
• localized renal tumors without lymph node and distant metastases as defined by NCCN guidelines
• ECOG score of 0 or 1
• Life expectancy greater than 10 years

Exclusion Criteria:

• people with surgically unresectable lesions
• people with Abnormal preoperative renal function, eGFR(estimated by CKD-EPI)<90ml/min/1.73m2
• people who receive preoperative molecular targeted therapy, immunotherapy, chemotherapy
• people with any contraindications to surgery
• people who convert to radical nephrectomy during surgery
• people who receive molecular targeted therapy, immunotherapy or chemotherapy during the postoperative follow-up period
• people with serious systemic disease

people with localized renal tumors scheduled for partial nephrectomy