We collected clinical data from 141 patients who completed partial nephrectomy at the First Affiliated Hospital of Xi'an Jiaotong University from September 2019 to August 2023, as well as their enhanced CT data and performed 3D reconstruction.All surgical operations are performed by senior chief urologists.The aim of this study was to predict surgical outcomes, and there were no interventions on surgical procedures or patients.Ethical approval from our institutional ethics committee was not required.

Robot-assisted partial nephrectomy was performed on all patients in this study, with either a transabdominal or retroperitoneal approach based on the mass's location. The renal artery was occluded with laparoscopic artery clamps, and the mass and part of the renal tissue were resected with scissors, creating a wedge resection 0.5–1 cm from the edge of the mass, ensuring the mass was resected intact. The collecting system was closed intermittently with 3–0 absorbable sutures, while the outer renal parenchyma was closed intermittently with 2–0 absorbable sutures.

Automatic measurement of SPARE score. A three-dimensional visualization system ( IPS system, Yorktal)[8]was applied to reconstruct the patient image data with automatic AI segmentation to obtain three-dimensional visualization images. The reconstructed 3D visualization images were imported into the virtual surgical planning software (Touch Viewer System, Yorktal) for automatic measurement of SPARE score.

SPARE scoring indexes: (1) the longest diameter of the mass; (2) Exophytic rate of the mass; (3) the location of the mass; (4) Whether the mass invades the renal sinus.

SPARE score measurement operation process:

The reconstructed three-dimensional model was imported into the virtual surgical planning software, where it can be rotated, scaled, and displayed with adjustable transparency. The software also allows for automatic measurement of both the long and short diameters, as well as simulation of cutting and other surgical operations. (1) Measurement of short and long diameters: Select and display the mass to be calculated in the 3D view, click on the short and long diameter measurement tool, the system automatically calculates and the results are displayed in the 3D view. (2) Mass convexity: Select and display the mass to be cut in the 3D view, and display the kidney where the mass is located at the same time, click on the surface cutting tool, and draw a closed curve along the edge of the renal parenchyma, the system will automatically generate a plane to divide the mass into two parts, and display the volume and the volume of the percentage of the mass. (3) Position of the mass: Select and display the kidney where the mass is located in the 3D view, and click on the long and short diameters measurement tool to measure the upper edge of the kidney. The longest distance between the upper and lower edges of the kidney will be measured to determine the location of the mass. (4) Whether the mass invades the renal sinus: Select and display the kidney and the mass in the 3D view to determine whether it invades the renal sinus.

All CT scans in this study were performed using high-definition imaging protocols with a slice thickness of 1 mm, conducted on a Philips CT scanner (Philips, Best, The Netherlands) to ensure high-resolution outputs suitable for 3D modeling. We adhere to strict internal quality control procedures, where both radiologists and bioengineers review the images to ensure they meet the necessary standards for clarity and accuracy Fig. 1.

Several cases were used to demonstrate how the SPARE score can be applied to 2D and 3D VMs for assessment, respectively

We collected basic information, demographic data(age, sex, body mass index, comorbidities classified according to Charlson’s comorbidity index [9] and American Society of Anesthesiologists score(ASA) [10]),perioperative-related markers, pathologic data ( the stage according to TNM classification [11] and histology and grading according to the WHO and International Society of Urological Pathology [12] (ISUP) classifications), and postoperative complication data.Clavien-Dindo classification [13] < grade III was considered a mild complication, and ≥ grade III was considered a severe complication. Tetrafecta outcomes were defined as thermal ischemia time (WIT) < 25 min, negative surgical margins, no major perioperative complications, and no reduction in postoperative renal function.SPARE score in 3DVM measured by the engineer and doublech-ecked by an experienced urologist and the 2D group were scored by two single-blinded urologists in previously reported methods [14,15,16].

Mean and standard deviation were used to describe continuous variables. Categorical variables were described using n(%).Student's t-test was used to compare means between the two groups. Spearman's correlation coefficient was used for correlation analysis between continuous variables, while Kendall's correlation coefficient was chosen for correlation analysis between continuous and dichotomous variables.Receiver-operating characteristic curve was used to evaluate the sensitivity and specificity of the SPARE score. The delong test was used to test whether there was a significant difference between two ROC curves. Statistical analysis was performed using SPSS software 25.0 and R 4.3.2.