Classification of Metastatic Lymph Nodes In Vivo Using Quantitative Ultrasound at Clinical Frequencies.

Area of Science:

• Medical imaging
• Ultrasound technology
• Oncology diagnostics

Background:

• Quantitative ultrasound (QUS) assesses tissue microstructure via backscattered echo signals.
• Previous high-frequency QUS (30 MHz) detected metastases in excised lymph nodes (LNs).
• Limited data exists on QUS efficacy for in vivo LNs at clinical frequencies (2-10 MHz).

Purpose of the Study:

• To evaluate the effectiveness of quantitative ultrasound (QUS) for in vivo lymph node (LN) metastasis detection at clinical frequencies.
• To assess QUS parameters derived from backscatter coefficient (BSC) and RF signal statistics.
• To develop and validate machine learning classifiers for metastatic LN identification.

Main Methods:

• Collected radiofrequency (RF) echo data from in vivo LNs of 19 cancer patients using a 10-MHz linear probe.
• Applied QUS methods to estimate BSC-based parameters (scatterer diameter, acoustic concentration) and RF signal statistics.
• Trained linear discriminant analysis (LDA) and support vector machine (SVM) classifiers using QUS parameters.

Main Results:

• Two BSC-based parameters, scatterer diameter and acoustic concentration, were most effective in detecting metastatic LNs.
• Both LDA and SVM classifiers achieved areas under the receiver operating characteristic (AUROC) curve ≥0.94.
• Classifying LNs based on the highest cancer probability echo frame yielded 88% specificity and 100% sensitivity (AUROC = 0.99).

Conclusions:

• QUS applied at clinical frequencies shows significant promise for accurate in vivo metastatic LN identification.
• This technique can aid in cancer diagnosis and potentially reduce unnecessary biopsies and surgical interventions.
• The study provides encouraging evidence for the clinical utility of QUS in lymph node staging.