Development and validation of an ultrasound‑based radiomics nomogram to predict lymph node status in patients with...

Area of Science:

• Oncology
• Medical Imaging
• Radiology

Background:

• Accurate preoperative prediction of lymph node status is crucial but challenging in ovarian cancer.
• Advances in medical imaging offer potential for improved diagnostic accuracy.

Purpose of the Study:

• To evaluate the feasibility of using ultrasound-based radiomics combined with clinical characteristics for predicting lymph node metastasis (LNM) in high-grade serous ovarian cancer (HGSOC).

Main Methods:

• Retrospective study of 401 HGSOC lesions from two institutions.
• Extraction of radiomics features from preoperative ultrasound images.
• Feature selection using F-value and recursive feature elimination (RFE) to identify 12 significant features.
• Construction of radiomics, clinical, and combined models, including a nomogram, for LNM prediction.
• Validation using 10-fold cross-validation and an external test cohort.

Main Results:

• The radiomics model achieved an AUC of 0.899 (training) and 0.855 (test).
• The combined model demonstrated superior performance with AUCs of 0.930 (training) and 0.881 (test).
• The combined model showed good calibration and discrimination, outperforming individual radiomics and clinical models.

Conclusions:

• A nomogram integrating radiomics score and clinical characteristics shows strong diagnostic performance for predicting LNM in HGSOC.
• This noninvasive method can aid in assessing lymph node status in HGSOC patients.