Enhanced AI Model Outperforms Traditional Techniques in Endometrial Cancer Diagnosis
A cutting-edge artificial intelligence (AI) model significantly improves the accuracy and efficiency of diagnosing high-risk endometrial cancer (EC) and predicting postoperative recurrence compared with traditional methods, according to a study published in Scientific Reports.
Researchers developed an enhanced version of the ResNet-101 deep learning model, incorporating spatial and channel attention mechanisms. “The specific goal is to evaluate the performance of the improved model in the risk identification of EC and the prediction of postoperative recurrence, and to compare it with traditional models, in hopes of providing a more effective diagnostic tool for clinical practice,” the study author explained.
The study, conducted from January 2021 to May 2024, involved 210 patients with EC who underwent pelvic MRI examinations. The improved model was tested against traditional ResNet-101 and other variations, consistently outperforming them in key metrics.
For high-risk EC diagnosis, the new model achieved an area under the curve (AUC) of 0.918, surpassing traditional ResNet-101 (AUC: 0.613), SA-ResNet-101 (AUC: 0.760), and CA-ResNet-101 (AUC: 0.758). Similarly, in predicting postoperative recurrence, the model demonstrated an AUC of 0.926 compared with 0.620 for traditional ResNet-101, showcasing its superior prognostic capabilities.
Key metrics, including accuracy (AC), precision (PR), recall (RE), and F1 scores, were significantly higher for the proposed model across all comparisons (P <.05). This advancement highlights the model's ability to differentiate between low-risk and high-risk EC and predict recurrence with high reliability.
The model's success lies in its enhanced feature representation ability, allowing it to focus more effectively on regions of interest within MRI images. This advancement could lead to earlier detection of high-risk cases and more personalized treatment strategies.
“In summary, the improved model can not only help doctors discover and diagnose high-risk EC patients earlier but also effectively predict the risk of postoperative recurrence, providing a scientific basis for the formulation of personalized treatment and follow-up plans, and helping to improve patients’ survival rates and QoL,” the study author concluded.
While the results are promising, the researchers emphasized the need for further validation in larger, diverse patient populations. Nevertheless, this study marks a significant step forward in applying artificial intelligence to gynecological cancer management, potentially improving diagnostic accuracy and patient care in the near future.
Reference
Qi X. Artificial intelligence-assisted magnetic resonance imaging technology in the differential diagnosis and prognosis prediction of endometrial cancer. Sci Rep. 2024;14(1):26878. doi:10.1038/s41598-024-78081-3
