Enhancing 3D Radio-Frequency Data in Quantitative Acoustic Microscopy Using Quantum-Driven Prior at 250-MHz and 500-MHz.
Academic Article
Overview
abstract
Quantitative acoustic microscopy (QAM) uses ultra-high-frequency ultrasound (>200-MHz) to create two-dimensional maps of acoustic and mechanical properties of tissue at microscopic resolutions (<8μm). Despite significant advancements in QAM, the spatial resolution of current systems, operating at 250-MHz and 500-MHz, may remain insufficient for certain biomedical applications. However, developing a QAM system with finer resolution by using higher-frequency transducers is costly, necessitates skilled operators, and these systems are more sensitive to the outside environment (e.g., vibrations, temperature). This study extends a resolution enhancement framework by proposing a generalized 3D-approach for processing QAM radio-frequency data. The framework utilizes a quantum-based adaptive-denoiser, DeQuIP, implemented as a regularization-prior (RED-prior) to enhance QAM-maps. Key contributions include a temporal hyperparameter optimization, accelerated algorithm integration, and application of quantum-interaction theory. DeQuIP employs quantum wave-functions, derived from the acquired data, as adaptive transformations that function as a RED-prior. This enables the framework to generate a temporally tailored regularization functional, allowing accurate modeling of complex physical phenomena in ultrasound propagation and providing a significant advantage over traditional regularizations in QAM imaging. The effectiveness of the proposed framework in enhancing resolution is demonstrated through both qualitative and quantitative analyses of experimental 2D parameter maps obtained from 250-MHz and 500-MHz QAM systems, alongside comparisons with a standard framework. Our framework demonstrates superior performance in recovering fine and subtle details, enhancing the spatial resolution of QAM-maps by 38.2-39.5%, surpassing the state-of-the-art framework, which achieved only 13.4-26.1% improvement, and shows notable visual improvements in spatial details when compared to histology images.
