Predicting Benign and Malignant Subpleural Pulmonary Lesions With a Nomogram Model Using Clinical and B-Mode Ultrasound Parameters

Abstract

Objective

To develop and validate an individualized nomogram for distinguishing between benign and malignant subpleural pulmonary lesions (SPLs) using B-mode ultrasound imaging and clinical data.

Methods

A total of 425 patients with SPLs were enrolled and classified into two groups: 220 patients were diagnosed with malignant lesions, and 205 with benign lesions. Patients were randomly assigned to a development cohort (DC, n = 297) and a validation cohort (VC, n = 128) in a 7:3 ratio. Statistical analyses included rank-sum tests and chi-square tests. Boruta analysis was used to identify key features associated with malignant SPLs. The multivariable logistic regression model based on independent malignant SPL factors was developed and represented as a nomogram. The model's performance was assessed in terms of discrimination, calibration and clinical utility.

Results

Six variables were selected to construct the nomogram: age, pack-year of smoking, air bronchogram, the angle between the lesion border and the thoracic wall, posterior echo of the lesion and visceral pleural invasion. The area under the receiver operating characteristic curve for the model was 0.859 (95% CI: 0.816–0.901) in the DC and 0.862 (95% CI: 0.800–0.923) in the VC. Calibration curve analysis demonstrated that the nomogram closely aligned with the ideal curve, reflecting its good calibration. Furthermore, decision curve analysis, clinical impact curve (CIC) and net reduction curve (NRC) further confirmed the model's favorable clinical utility.

Conclusion

We have developed a nomogram that serves as an effective tool for assessing malignant SPLs. This model holds significant promise as a complementary diagnostic aid, particularly in primary healthcare settings and bedside examination.