KEY TAKEAWAYS
- The study aimed to investigate the efficacy of a deep-learning radiomics nomogram based on ultrasound imaging in predicting the malignant risk of ovarian tumors and compare its diagnostic performance with the O-RADS.
- Researchers noticed that the US-based DLR_Nomogram accurately predicts ovarian tumor malignancy, showing efficacy comparable to O-RADS, suggesting its potential as a valuable diagnostic tool in clinical practice.
The timely identification and management of ovarian cancer (OC) is pivotal for patient prognosis.
Yangchun Du and the team aimed to develop and validate a deep learning radiomics nomogram (DLR_Nomogram) employing ultrasound (US) imaging to predict the malignant risk associated with ovarian tumors accurately. Moreover, they compared the diagnostic performance of the DLR_Nomogram with that of the ovarian-adnexal reporting and data system (O-RADS).
Researchers performed an inclusive analysis encompassing 2 research tasks. Patients were randomly divided into training and testing sets in an 8:2 ratio for both tasks. In task 1, the malignancy risk of 849 patients with ovarian tumors was assessed, while in task 2, the malignancy risk of 391 patients with O-RADS 4 and O-RADS 5 ovarian neoplasms was evaluated. Three models were developed and validated to predict the risk of malignancy in ovarian tumors.
The predicted outcomes of the models for each sample were merged to form a new feature set, which was utilized as an input for the logistic regression (LR) model to construct a combined model, visualized as the DLR_Nomogram. Subsequently, the diagnostic performance of these models was evaluated by the receiver operating characteristic curve (ROC).
The DLR_Nomogram exhibited superior predictive performance for assessing the malignant risk of ovarian tumors. In task 1, the nomogram achieved ROC curve (AUC) values of 0.985 and 0.928 for the training and testing sets, respectively, while in task 2, the highest recorded AUC values of 0.955 and 0.869 for the training and testing sets, respectively.
Additionally, the DLR_Nomogram demonstrated satisfactory fitting performance in Hosmer-Lemeshow testing for both tasks. Decision curve analysis further illustrated its greater net clinical benefits in predicting malignant ovarian tumors within specific threshold values.
The study concluded that the US-based DLR_Nomogram effectively predicts the malignant risk of ovarian tumors, demonstrating predictive efficacy comparable to O-RADS.
This study was supported by the Guangxi Key Research and Development Plan and the Guangxi Promotion of Appropriate Health Technologies Project.
Source: https://pubmed.ncbi.nlm.nih.gov/38594729/
Du Y, Xiao Y, Guo W, et al. (2024). “Development and validation of an ultrasound-based deep learning radiomics nomogram for predicting the malignant risk of ovarian tumours.” Biomed Eng Online. 2024 Apr 9;23(1):41. doi: 10.1186/s12938-024-01234-y. PMID: 38594729; PMCID: PMC11003110.
