Photoacoustic-Based Intra- and Peritumoral Radiomics Nomogram for the Preoperative Prediction of Expression of Ki-67 in Breast Malignancy.

Objective: This study investigated the preoperative predictive efficiency of radiomics derived from photoacoustic (PA) imaging, integrated with the clinical features of Ki-67 expression in malignant breast cancer (BC), with a focus on both intratumoral and peritumoral regions.

Methods: This study involved 359 patients, divided into a training set (n = 251) and a testing set (n = 108). Radiomic features were extracted from intratumoral and peritumoral regions using PA imaging. Multivariate logistic regression was employed to identify significant clinical factors. LASSO regression was used to select the features extracted from the training set. The selected radiomics features were combined with clinical features to develop a radiomics nomogram. The predictive efficiency of the model was assessed using the area under the receiver operating characteristic curve (AUC), and its clinical utility and accuracy were evaluated through decision curve analysis and calibration curves, respectively.

Results: The developed nomogram combined 6 mm peritumoral data with intratumoral and clinical features and showed excellent diagnostic performance, achieving an AUC of 0.899 in the testing set. They both showed good calibrations. The outperformed models based solely on clinical features or other radiomics methods, with the 6 mm surrounding tumor area proving most effective in identifying Ki-67 status in BC patients.

Conclusions: Integrating PA radiomics with clinical features offers a robust preoperative tool for predicting Ki-67 status in BC, optimizing the delineation of peritumoral regions for enhanced diagnostic precision. The model's strong performance supports its potential as a non-invasive adjunct to traditional biopsy methods, aiding in the personalized management of BC treatment.