Combined CT morphological and histogram features improve differentiation of PIMA from PINMA compared to separate models.
A retrospective cohort review analyzed data from 58 patients with solitary pulmonary invasive mucinous adenocarcinoma (PIMA) and 105 patients with pulmonary invasive non-mucinous adenocarcinoma (PINMA). The setting was not reported. The primary objective was to assess the ability of different computational models to differentiate between these two lung cancer subtypes.
The study compared three prediction strategies: a separate CT morphological model, a separate CT histogram-based model, and a combined model integrating both feature types. The primary outcome measured was the area under the curve (AUC) for each approach. Secondary outcomes included calibration and clinical applicability.
Results indicated that the separate CT morphological model achieved an AUC of 0.754. The separate CT histogram-based model performed better with an AUC of 0.820. The combined prediction model demonstrated the highest performance with an AUC of 0.845. No adverse events, discontinuations, or safety data were reported as the intervention involved computational analysis rather than a pharmacologic or procedural therapy.
Key limitations include the retrospective nature of the data and the relatively small sample size of 163 patients total. The study design precludes definitive conclusions on generalizability. While the authors note the nomogram provides a practical tool for non-invasive diagnosis, the evidence remains observational. Clinicians should interpret these findings as preliminary regarding the superiority of combined features over separate models in routine practice.
View Original Abstract ↓
Muscle ultrasonography shows high diagnostic accuracy for fasciculation detection in ALS
A Simple Scan Could Speed Up an ALS Diagnosis
Related in Radiology & Imaging
From Other Specialties
