Application of MRI-based tumor heterogeneity analysis for identification and pathologic staging of breast phyllodes tumors DOI Creative Commons
Liang Yue, Qingyu Li, Jiahao Li

и другие.

Magnetic Resonance Imaging, Год журнала: 2025, Номер 117, С. 110325 - 110325

Опубликована: Янв. 7, 2025

To explore the application value of MRI-based imaging histology and deep learning model in identification classification breast phyllodes tumors. Seventy-seven patients diagnosed as tumors fibroadenomas by pathological examination were retrospectively analyzed, traditional radiomics features, subregion features extracted from MRI images, respectively. The screened modeled using variance selection method, statistical test, random forest importance ranking Spearman correlation analysis, least absolute shrinkage operator (LASSO). efficacy each was assessed subject operating characteristic (ROC) curve, DeLong test used to assess differences AUC values different models, clinical benefit decision curve (DCA), predictive accuracy calibration (CCA). Among constructed models for tumors, fusion (AUC: 0.97) had best diagnostic highest benefit. 0.81) better compared with 0.70). De-Long there is a difference between model, training group. distinguish breast, TDT_CIDL 0.974) TDT_CI combination statistically remaining five Traditional based on sequences can help differentiate benign junctional fibroadenomas, provide personalized treatment patients.

Язык: Английский

Application of MRI-based tumor heterogeneity analysis for identification and pathologic staging of breast phyllodes tumors DOI Creative Commons
Liang Yue, Qingyu Li, Jiahao Li

и другие.

Magnetic Resonance Imaging, Год журнала: 2025, Номер 117, С. 110325 - 110325

Опубликована: Янв. 7, 2025

To explore the application value of MRI-based imaging histology and deep learning model in identification classification breast phyllodes tumors. Seventy-seven patients diagnosed as tumors fibroadenomas by pathological examination were retrospectively analyzed, traditional radiomics features, subregion features extracted from MRI images, respectively. The screened modeled using variance selection method, statistical test, random forest importance ranking Spearman correlation analysis, least absolute shrinkage operator (LASSO). efficacy each was assessed subject operating characteristic (ROC) curve, DeLong test used to assess differences AUC values different models, clinical benefit decision curve (DCA), predictive accuracy calibration (CCA). Among constructed models for tumors, fusion (AUC: 0.97) had best diagnostic highest benefit. 0.81) better compared with 0.70). De-Long there is a difference between model, training group. distinguish breast, TDT_CIDL 0.974) TDT_CI combination statistically remaining five Traditional based on sequences can help differentiate benign junctional fibroadenomas, provide personalized treatment patients.

Язык: Английский

Процитировано

0