Malignant pleural mesothelioma classification and survival prediction with CT imaging using ResNet
A ResNet‑3D‑18 deep‑learning model can accurately distinguish malignant pleural mesothelioma from metastatic pleural disease and the extracted imaging features also provide prognostic information for overall survival.
This study aims to achieve accurate differentiation of malignant pleural mesothelioma (MPM) from metastatic pleural disease (MPD) and to predict the overall survival of MPM using a ResNet‑3D‑18 deep learning model applied to pretreatment CT scans. The model demonstrated high discriminative performance (AUC 0.972 in training, 0.840 in independent test) and superior sensitivity compared with a clinical logistic regression benchmark. A random‑forest classifier built on deep‑learning features also predicted overall survival (AUC 0.829).
Demonstrates that non‑invasive CT‑based AI can improve diagnosis and prognostication of mesothelioma, potentially guiding earlier treatment decisions.
Evidence level: Állatkísérletes. Állatmodellben vizsgálták.
Related signals
- Moving Beyond Morphology: Toward a Morpho‑Molecular Classification of Pleural Mesothelioma
- Comparison of Survival by Multimodal Treatment Regimen Among Malignant Pleural Mesothelioma Patients in an Integrated Health System
- Systemic Treatments for Unresectable Pleural Mesothelioma: A Review of the Literature
- Prognostic impact of somatic mutations among patients with pleural and peritoneal mesothelioma
- Tumor Treating Fields (TTFields) for Pleural Mesothelioma