What type of study is this?

This is a Quantitative Study study.

September 17, 2025

Investigation of Deep Learning Models Based on Multiparametric MRI to Diagnose Solid Small Renal Masses: A Multi-Center Study

Key Points

The deep learning model accurately distinguishes benign from malignant solid small renal masses.
An AUC of 0.955 was achieved on internal testing, validating the model's performance.
Data from 913 patients across three institutions were utilized to train and test the classification model.
Highlighting the potential to reduce unnecessary resections of benign tumors, the model addresses diagnostic challenges.

Abstract

Motivation: Deep learning (DL) technique could efficiently diagnose the benign and malignant of solid small renal mass (SRM). Goal(s): To develop and evaluate the performances of DL model based on MRI to differentiate benign from malignant SRM. Approach: A total of 913 SRM patients from three institutions was used including for DL classification model training and testing for five sequences (T1WI, T2WI, CP, NP, DP). The model was tested on internal and external test sets. Results: The AUC of internal and two external test set yield 0.955 (95% CI: 0.914, 0.996), 0.842 (95% CI: 0.736, 0.949), and 0.848 (95% CI: 0.731, 0.966). Impact: Due to overlapping imaging features, distinguishing benign from malignant SRM is challenging, leading to unnecessary resection of benign SRM. The DL model offers an efficient tool for the accurate classification of SRM.

Bookmark

Investigation of Deep Learning Models Based on Multiparametric MRI to Diagnose Solid Small Renal Masses: A Multi-Center Study

Key Points

Abstract

Cite This Study