Accurate lymph node (LN) characterization is crucial for cancer diagnosis and staging. Elastography (E) may improve Endoscopic Ultrasound (EUS) accuracy, while fractal geometry and colorimetric analysis also show promise in oncologic imaging. This study evaluates the diagnostic performance of EUS-E, fractal analysis, and colorimetric analysis in distinguishing malignant from benign LNs.

In this prospective single-center study (NCT04671784), LNs were evaluated using EUS-E (strain ratio, SR), fractal dimension, and automated colorimetric analysis (mean red, green, blue, RGB, values and channel ratios). Optimal cut-offs were defined by histology, and AUC and accuracy were compared across methods. Final diagnoses were confirmed by cytology, histology, or ≥24-month follow-up.

From September 2019 to January 2023, 100 patients with enlarged mediastinal or abdominal LNs (50 malignant) were prospectively enrolled. Quantitative E (SR cut-off: 8.7) achieved an AUC of 92% [95% CI, 80-100], with 100% sensitivity and 80% specificity. Colorimetric analysis showed higher accuracy, with an AUC of 94% [95% CI, 81-100], 100% sensitivity, and 87% specificity (p < 0.0001). Combining both E and colorimetric analysis yielded an AUC of 94.5% [95%CI, 83-100], with 100% sensitivity and 93.3% specificity (p < 0.0001). The addiction of fractal analysis did not improve the accuracy.

Our findings demonstrate that EUS-E is highly effective in differentiating benign from malignant LNs. Incorporating colorimetric analysis of elastographic images further improves diagnostic accuracy, while fractal analysis does not contribute significantly. Nevertheless, each modality offers unique and potentially complementary insights into LN pathology.