Quantitative ultrasound (QUS) enhances imaging contrast for cancer detection by extracting tissue microstructural information from raw radiofrequency (RF) backscatter, complementing conventional B-mode imaging. While QUS has shown promise in identifying malignancies in the breast, thyroid, prostate, and lymph nodes, clinical adoption remains limited due to challenges such as restricted access to raw RF data, offline processing, and high computational demands. In this study, we modified a diagnostic ultrasound system with custom software enabling near real-time QUS analysis in a clinical research environment. Two QUS techniques were implemented: (1) the reference phantom method to estimate the backscatter coefficient (BSC) from RF spectra, and (2) statistical analysis of envelope-detected signals. A clinician scanned lymph nodes in two healthy volunteers, automatically storing RF data in DICOM format and generating QUS parametric maps at the bedside within seconds using our software. Additionally, lymph nodes from 24 patients undergoing needle biopsy were analyzed offline. A machine learning model was trained to classify lymph nodes as benign or malignant based on extracted QUS features. This work demonstrates the potential for real-time, scanner-integrated QUS to support clinical decision-making in cancer diagnostics.
Huda et al. (Wed,) studied this question.