AUTHOR=Liang Mengdi , Zhang Cai , Xia Tiansong , Chen Rui , Wang Xinyang , Weng Miaomiao , Xie Hui , Chen Lin , Liu Xiaoan , Wang Shui
TITLE=Ultrasound radiomics features predicting the dosimetry for focused ultrasound surgery of benign breast tumor: A retrospective study
JOURNAL=Frontiers in Genetics
VOLUME=13
YEAR=2022
URL=https://www.frontiersin.org/journals/genetics/articles/10.3389/fgene.2022.969409
DOI=10.3389/fgene.2022.969409
ISSN=1664-8021
ABSTRACT=Purpose: To investigate the correlation between pre-ablation ultrasound radiomics features and the sonication energy for focused ultrasound surgery (FUS) of benign breast tumors.
Method: 53 benign breast tumors of 28 patients treated by ultrasound-guided focused ultrasound surgery (USgFUS) were included in this study. The sonication energy per unit volume of each tumor was calculated. Three-quarter point was chosen as the cut-off to divide the 53 included tumors into high sonication energy (HSE, n = 14) and low sonication energy (LSE, n = 39) groups. For each tumor, the region of interest (ROI) of both the tumor itself (tROI) and the near field tissue (nfROI) were delineated and analyzed separately using ImageJ software. Pearson correlation coefficient and multiple linear regression analysis were used for radiomics feature selection. To explore the diagnostic performance of different ultrasound radiomics features, a receiver operating characteristic (ROC) curve analysis was performed.
Results: In total of 68 radiomics features were extracted from pre-ablation ultrasound images of each tumor. Of all radiomics features, BX in tROI (p < 0.001), BX (p = 0.001) and Circ (p = 0.019) in nfROI were independently predictive features of sonication energy per unit volume. The ROC curves showed that the area under the curve (AUC) values of BX in tROI, BX, and Circ in nfROI were 0.797, 0.787 and 0.822, respectively.
Conclusion: This study provided three radiomics features of pre-ablation ultrasound image as predictors of sonication dose for FUS in benign breast tumors. Further clinical trials are needed to confirm the predictive effect of these radiomics features.