Predictive performance of image processing techniques in the diagnosis of bone metastasis: A meta‑analysis of a confusion matrix

Bone metastasis, a common cause of disability, markedly affects prognosis and life expectancy, and is usually diagnosed through radiographic imaging. An early and accurate diagnosis of bone metastasis is critical in order to optimize therapeutic strategies and palliative care. Computer‑based image processing techniques have proved promising to improve diagnostic efficiency. The present systematic review and meta‑analysis aimed to evaluate the efficacy of the diagnostic performance metrics of computational image processing techniques in patients with lung cancer and bone metastasis. Following the Preferred Reporting Items for Systematic reviews and Meta‑analysis guidelines, a comprehensive literature search was conducted across different databases between January 2010 and December 2024. Studies assessing computational image‑processing techniques in patients with lung cancer and bone metastasis were included. Confusion matrix parameters were extracted and the bivariate Reitsma model was applied to estimate the pooled diagnostic performance. Heterogeneity and inconsistency (variance of components and I2 values), and risk of bias (quality assessment of diagnostic accuracy studies‑2 tool) were appraised for the included studies. Overall, 6 studies were included in the meta‑analysis. A high overall diagnostic accuracy (area under the summary receiver operating characteristic curve, 0.931), a pooled sensitivity of 0.86 and a specificity of 0.88 were achieved. Favorable positive and negative likelihood ratios (7.22 and 0.165, respectively) indicated the strong discriminatory ability of the model. Despite some heterogeneity, stable high negative predictive values confirmed the reliability of ruling out non‑metastasis when screening for metastasis. Computer‑based image processing techniques demonstrated excellent diagnostic accuracy for bone metastasis in patients with lung cancer. Further studies should focus on investigating a unique standardized diagnostic tool that can be applied in a clinical setting to improve patient management.