Enhanced ISUP grade prediction in prostate cancer using multi-center radiomics data DOI Creative Commons

Yuying Liu,

Xueqing Han,

Haohui Chen

et al.

Abdominal Radiology, Journal Year: 2025, Volume and Issue: unknown

Published: March 6, 2025

To explore the predictive value of radiomics features extracted from anatomical ROIs in differentiating International Society Urological Pathology (ISUP) grading prostate cancer patients. This study included 1,500 patients a multi-center study. The peripheral zone (PZ) and central gland (CG, transition + zone) were segmented using deep learning algorithms defined as regions interest (ROI) this A total 12,918 image-based T2-weighted imaging (T2WI), apparent diffusion coefficient (ADC), diffusion-weighted (DWI) images these two ROIs. Synthetic minority over-sampling technique (SMOTE) algorithm was used to address class imbalance problem. Feature selection performed Pearson correlation analysis random forest regression. prediction model built classification algorithm. Kruskal-Wallis H test, ANOVA, Chi-Square Test for statistical analysis. 20 ISUP grading-related selected, including 10 PZ ROI CG ROI. On test set, combined exhibited better performance, with an AUC 0.928 (95% CI: 0.872, 0.966), compared alone (AUC: 0.838; 95% 0.722, 0.920) 0.904; 0.851, 0.945). demonstrates that radiomic based on sub-region can contribute enhanced grade prediction. combination GG provide more comprehensive information improved accuracy. Further validation strategy future will enhance its prospects improving decision-making clinical settings.

Language: Английский

Enhanced ISUP grade prediction in prostate cancer using multi-center radiomics data DOI Creative Commons

Yuying Liu,

Xueqing Han,

Haohui Chen

et al.

Abdominal Radiology, Journal Year: 2025, Volume and Issue: unknown

Published: March 6, 2025

To explore the predictive value of radiomics features extracted from anatomical ROIs in differentiating International Society Urological Pathology (ISUP) grading prostate cancer patients. This study included 1,500 patients a multi-center study. The peripheral zone (PZ) and central gland (CG, transition + zone) were segmented using deep learning algorithms defined as regions interest (ROI) this A total 12,918 image-based T2-weighted imaging (T2WI), apparent diffusion coefficient (ADC), diffusion-weighted (DWI) images these two ROIs. Synthetic minority over-sampling technique (SMOTE) algorithm was used to address class imbalance problem. Feature selection performed Pearson correlation analysis random forest regression. prediction model built classification algorithm. Kruskal-Wallis H test, ANOVA, Chi-Square Test for statistical analysis. 20 ISUP grading-related selected, including 10 PZ ROI CG ROI. On test set, combined exhibited better performance, with an AUC 0.928 (95% CI: 0.872, 0.966), compared alone (AUC: 0.838; 95% 0.722, 0.920) 0.904; 0.851, 0.945). demonstrates that radiomic based on sub-region can contribute enhanced grade prediction. combination GG provide more comprehensive information improved accuracy. Further validation strategy future will enhance its prospects improving decision-making clinical settings.

Language: Английский

Citations

0