Multimodal MRI radiomics-based stacking ensemble learning model with automatic segmentation for prognostic prediction of HIFU ablation of uterine fibroids: a multicenter study

2024

MRI Model for Predicting HIFU Treatment Success in Uterine Fibroids

Sample size: 360 publication 10 minutes Evidence: high

Author Information

Author(s): Wen Bing, Li Chengwei, Cai Qiuyi, Shen Dan, Bu Xinyi, Zhou Fuqiang

Primary Institution: Yiyang Central Hospital, Yiyang, China

Hypothesis

To evaluate the effectiveness of an MRI radiomics stacking ensemble learning model for preoperative prediction of the prognosis of HIFU ablation of uterine fibroids.

Conclusion

The DL based automatic segmentation MRI radiomics stacking ensemble learning model demonstrated high accuracy in predicting the prognosis of HIFU ablation of uterine fibroids.

Supporting Evidence

The MLP model achieved an AUC of 0.858 in the internal test set.
The stacking ensemble model demonstrated an AUC of 0.897 in the internal test set.
Automated segmentation reduced the need for manual contouring, minimizing human error.
Integrating T2WI and CE-T1WI improved predictive performance compared to single sequences.
Feature selection reduced the initial 2,394 features to 36 relevant features for analysis.

Takeaway

This study created a smart computer program that uses MRI pictures to help doctors predict how well a treatment for uterine fibroids will work.

Methodology

This retrospective study analyzed data from 360 patients, using a stacking ensemble model that combined multiple machine learning algorithms to predict treatment outcomes based on MRI features.