TL;DR
This paper introduces 3D-PAULM, a novel, non-invasive imaging system that combines photoacoustic tomography (PACT) and ultrasound localization microscopy (ULM) to provide unprecedented, multi-modal functional imaging of the deep mouse brain. By integrating these techniques, 3D-PAULM can simultaneously map blood oxygenation, vessel density, and blood flow velocity, offering a comprehensive view of brain hemodynamics, particularly in the context of stroke.
What Problem Does This Paper Address?
Imaging the deep brain's intricate vascular network and its functional status (like blood flow and oxygenation) non-invasively is a significant challenge. Traditional methods like PACT excel at visualizing blood vessels and quantifying oxygen saturation (sO2) but struggle with direct blood flow measurement and can suffer from limited views, especially through the skull. Ultrasound (US) imaging can detect blood flow via Doppler effects, but its spatial resolution for deep tissues is often insufficient. While ultrasound localization microscopy (ULM) offers super-resolution for microvasculature, integrating it with deep-brain PACT and achieving high volumetric rates has been difficult. This paper aims to bridge this gap by developing a single, integrated system that overcomes these limitations, enabling comprehensive, non-invasive, 3D functional brain imaging.
Methods
The core innovation is the 3D-PAULM system, which integrates multiple imaging modalities into a single device. Key components and techniques include:
- Low-Frequency 4 MHz Transducer: A custom 2D hemispherical transducer array with a 4 MHz center frequency is used. This lower frequency is crucial for deep penetration through the intact scalp and skull, a significant advantage over higher-frequency systems.
- Integrated Modalities: The system supports:
- Multi-spectral PACT: Uses multiple laser wavelengths (700 nm, 750 nm, 1064 nm) to image blood vessels and quantify sO2.
- US B-mode Imaging: Provides anatomical context.
- US Power Doppler (PD): Detects blood flow.
- Ultrasound Localization Microscopy (ULM): Achieves super-resolution microvascular imaging by tracking gas-filled microbubbles.
- Synthetic Aperture US Imaging: Employed to improve image quality and achieve a high volumetric frame rate (215 Hz) for ULM, despite the transducer's large element pitch.
- Raster Scanning: To cover the entire mouse brain, the transducer array is raster scanned across multiple positions, allowing for an expanded field of view.
- ULM Processing Workflow: A multi-step process involving spatial-temporal singular value decomposition (SVD) for clutter removal, power Doppler image formation, radial symmetry-based microbubble localization, and 3D particle tracking to generate density and flow speed maps.
- Vessel Segmentation: PD images are used for vessel segmentation due to their better continuity, providing masks for extracting functional information (sO2, flow speed) from ULM and PACT data.
Datasets
Not discussed in the paper. The study utilizes a stroke mouse model for in vivo evaluation.
Key Insights and Ideas
- Low Frequency is Key for Deep Brain: The choice of a 4 MHz transducer is a critical design decision that enables deep brain imaging through the skull, a major hurdle for previous ULM and PACT systems.
- Synergy of Modalities: The true power lies in the co-registration and fusion of data from PACT (sO2, vessel structure) and ULM (high-resolution flow, vessel density). This allows for a more complete picture than any single modality alone.
- ULM for Diving Vessels: ULM successfully visualizes "diving" vessels that are often missed by PACT due to the limited-view problem, highlighting its complementary nature.
- Penumbra Identification: The system's ability to map both sO2 and blood flow allows for the potential identification of the ischemic penumbra – tissue at risk but potentially salvageable – by observing regions with reduced perfusion but maintained oxygenation.
Key Results
- Resolution: Achieved lateral and axial resolutions of 260 µm and 390 µm for PA, and 270 µm and 440 µm for US, respectively. ULM provided significantly higher resolution for microvessels (e.g., 68.1 µm vs. 269.1 µm in PD).
- Volumetric Frame Rate: US imaging achieved a volumetric frame rate of 215 Hz, while PACT operated at 10 Hz.
- Stroke Model Performance: In a stroke mouse model, 3D-PAULM successfully visualized stroke-induced ischemia, hypoxia, and reduced blood flow.
- PA images showed reduced signal intensity in the stroke hemisphere, indicative of decreased blood perfusion.
- sO2 maps revealed lower oxygenation in the stroke core.
- PD and ULM images showed reduced vessel density and flow speed in the stroke hemisphere.
- Quantitative Analysis: Statistical analysis of segmented vessels revealed significant differences between healthy and stroke hemispheres, including increased vessel length, decreased vessel radius, reduced sO2, and lower median flow speed in the stroke hemisphere.
Limitations and Future Work
- Limited View Problem in PACT: The paper acknowledges that PACT can still suffer from the "limited view problem," potentially missing some structures like diving vessels. Future work could involve rotating the transducer or using microbubbles as virtual point sources.
- ULM Frame Rate Optimization: While 215 Hz is good, further improvements in the US volumetric frame rate could enhance ULM accuracy, potentially by optimizing transmission patterns.
- Image Reconstruction: The current reconstruction uses a homogenous speed of sound. Future work could incorporate multi-SOS reconstruction by mapping skull thickness and speed of sound heterogeneity using US B-mode, which is expected to improve PD and ULM accuracy.
- Vessel Segmentation: Current vessel segmentation relies on PD images. Acquiring more volumetric frames for ULM could enable segmentation directly from the higher-resolution ULM data.
- Clinical Translation: The system's non-invasive nature and real-time capabilities make it promising for clinical applications, such as guiding thermal therapies or incorporating PA molecular imaging.
Summary updated