Artificial Intelligence Identifies Small Renal Tumours with Radiologist Precision - EMJ

Artificial Intelligence Identifies Small Renal Tumours with Radiologist Precision

1 Mins
Radiology

A DEEP learning (DL) algorithm based on contrast-enhanced multiphase CT has performed well in identifying small (3 cm) benign renal lesions, according to new research. Accurate characterisation of small renal masses is vital for optimal patient care, and deep learning algorithms offer the potential for diagnostic accuracy and consistency to be improved. This study aimed to develop and validate a DL algorithm for distinguishing the masses using contrast-enhanced multiphase CT, comparing its performance with that of urological radiologists. 

Data were retrospectively collected from a single hospital (2009–2021, 1,063 lesions) and used to train the DL algorithm, with an 8:2 split for training and internal testing. External testing was conducted using data from five independent hospitals (2013–2021, 537 lesions), and a prospective test set was obtained from one hospital (2021–2022, 103 lesions). Performance was assessed using the area under the receiver operating characteristic curve (AUC), and compared with seven urological radiologists’ assessments using the DeLong test. 

A total of 1,703 patients (mean age: 56±12 years; 619 female) with single renal masses were evaluated. In external testing, the DL algorithm achieved an AUC of 0.80 (95% CI: 0.75–0.85), comparable to radiologists’ AUC of 0.84 (95% confidence interval [CI]: 0.78–0.88; P=0.61). For the prospective test set, the DL algorithm’s AUC was 0.87 (95% CI: 0.79–0.93), while radiologists scored 0.92 (95% CI: 0.86–0.96; P=0.70). For lesions <1 cm, the algorithm’s AUC was 0.74 (95% CI: 0.63–0.83), similar to that of radiologists’ (0.81 [95% CI: 0.68–0.92; P=0.78]). 

“[Our] developed multiphase CT-based deep learning (DL) algorithm for identifying small [equal to or less than 3 cm] and subcentimeter […] benign renal masses demonstrated comparable performance with that of urological radiologists,” the researchers said. 

The DL algorithm performed comparably to urological radiologists in identifying benign small renal masses using contrast-enhanced multiphase CT. Its efficacy in detecting lesions that are 1 cm or less suggests it as valuable tool for clinical practice, enhancing diagnostic accuracy and consistency. 

 

Reference 

Dai C et al. Deep learning assessment of small renal masses at contrast-enhanced multiphase CT. Radiology. 2024;311(2):e232178. 

 

Rate this content's potential impact on patient outcomes

Thank you!

Please share some more information on the rating you have given