Computational Enhancement of Molecularly Targeted Contrast-Enhanced Ultrasound: Application to Human Breast Tumor Imaging
This work addresses the problem of accurate early cancer detection in human breast tumor imaging, representing an incremental advancement in computational methods for clinical ultrasound.
The researchers tackled the challenge of improving molecularly targeted contrast-enhanced ultrasound (mCEUS) for early breast cancer detection by developing computational enhancement techniques that distinguish bound contrast agents from other signals, resulting in over 10x improvement in contrast ratios between lesions and normal tissue in human patient exams and eliminating a false positive.
Molecularly targeted contrast enhanced ultrasound (mCEUS) is a clinically promising approach for early cancer detection through targeted imaging of VEGFR2 (KDR) receptors. We have developed computational enhancement techniques for mCEUS tailored to address the unique challenges of imaging contrast accumulation in humans. These techniques utilize dynamic analysis to distinguish molecularly bound contrast agent from other contrast-mode signal sources, enabling analysis of contrast agent accumulation to be performed during contrast bolus arrival when the signal due to molecular binding is strongest. Applied to the 18 human patient examinations of the first-in-human molecular ultrasound breast lesion study, computational enhancement improved the ability to differentiate between pathology-proven lesion and pathology-proven normal tissue in real-world human examination conditions that involved both patient and probe motion, with improvements in contrast ratio between lesion and normal tissue that in most cases exceed an order of magnitude (10x). Notably, computational enhancement eliminated a false positive result in which tissue leakage signal was misinterpreted by radiologists to be contrast agent accumulation.