Functional Ultrasound Imaging (fUSI) from scratch
An in-depth technical breakdown and accompanying Python notebooks explain how functional ultrasound imaging (fUSI) captures high-resolution brain activity by tracking blood flow.
read it It provides a clear physics and signal-processing walkthrough of functional ultrasound imaging, paired with open-source Python notebooks for computational neuroscientists.
summary
Functional ultrasound imaging (fUSI) tracks rapid brain hemodynamics via speckle interference in backscattered ultrasound, offering significantly higher resolution than fMRI without requiring heavy magnets. The author explains the underlying physics, ultrafast plane-wave reconstruction, Power Doppler estimation, and clutter filtering used to isolate blood volume changes from motion artifacts. The piece also explores how these signals can be analyzed using GLMs and decoders, highlighting opportunities for better signal processing. To bridge the gap, the author provides open-source Python tutorials simulating ultrasound propagation and reconstruction.
key points
- fUSI achieves up to roughly 100 micrometer linear resolution by tracking subtle changes in ultrasound speckle caused by red blood cell movement.
- Power Doppler estimates measure local blood volume changes but are vulnerable to global motion, necessitating advanced clutter filtering or motion correction.
- Open-source Python notebooks are provided to simulate ultrasound propagation and perform delay-and-sum reconstruction outside of locked MATLAB toolkits.
who should read it
Computational neuroscientists, machine learning engineers, and researchers interested in brain-computer interfaces or novel neuroimaging modalities.
discussion warm
consensus
Commenters express enthusiasm for the technology and its accessibility, sharing related DIY ultrasound projects and noting the future potential of multimodal setups like trimodal EEG, fNIRS, and fUSI headsets.
pushback
A commenter points out the major limitation that the rigid skull impedes ultrasound penetration unless done intraoperatively or through an acoustically transparent implant.
Very interesting, great write up, and awesome blog in general!
I never got to invest serious time in this idea, but MMW: trimodal EEG+fNIRS+fUSI dry headsets are going to change everything. Each technique has its strengths, and there are already people hacking together trimodal prototypes as we speak.
It highlights the future potential of trimodal dry headsets combining EEG, fNIRS, and fUSI to overcome individual modality limitations.
Generated by an LLM from the article and thread. Check the sources before relying on it.