Jiahong Li a, Zhiyuan Zheng a, Zilong Li a, Yiguang Wang a, Yubin Cao a, Qiegen Liu a,* ,Xianlin Song a,* Establishment of data-domain sharing database for generative artificial intelligence-based photoacoustic tomography

If data from this database are used, please cite the above papers:"Establishment of data-domain sharing database for generative artificial intelligence-based photoacoustic tomography"-https://www.spiedigitallibrary.org/conference-proceedings-of-spie/13248/132480U/Establishment-of-data-domain-sharing-database-for-generative-artificial-intelligence/10.1117/12.3033668.short Date : Dec-6-2024

Traditional reconstruction algorithms use acoustic inversion analytical methods to reconstruct, such as time domain back-projection, time reversal, delayed summation, etc. To obtain high-quality imaging, one can utilize multi-element ultrasound detectors, such as a full-ring detector, spherical detection array, ring detection array, or planar detection array. Of course, this will make the system more complex and more expensive. Therefore, we use a single-probe photoacoustic system, which uses a rotating stage to drive the probe to rotate to different angles to collect photoacoustic signals, and then uses the collected photoacoustic data to reconstruct the image using a back-projection algorithm.

System.

Our system consists of step control module, laser trigger module and data acquisition(DAQ) module. The stepping module consists of a vertical stepping motor and a horizontal rotating stage. The laser trigger module is composed of a laser. During acquisition, the laser pulse triggers DAQ to collect data. The rotating stage rotates every 2° and collects 180 times in one rotation. The sampling frequency of the DAQ is 500MHz and 50,000 points are sampled each time. The rotation radius of the ultrasonic probe is ~7cm. Finally, the collected photoacoustic signals are formed into a sinogram, and the back-projection algorithm is used to reconstruct the sample image.

DAQ sampling frequency：500MHz，

Sampling points：50000，

Rotation radius of rotary table~7cm，

360° imaging (scan every 2°, 180 times in total)

Results.

Fig. 2.Reconstruction results of 180 angles, (a) (d) (g) (j) are black tape samples, (b) (e) (h) (k) are the obtained sinograms, (c) (f) (i) (l) are the back-projection reconstruction results

Other Related Projects

• Accelerated model-based iterative reconstruction strategy for sparse-view photoacoustic tomography aided by multi-channel autoencoder priors
  [Paper] [Code]

• Generative model for sparse photoacoustic tomography artifact removal
  [Paper]

• Sparse-view reconstruction for photoacoustic tomography combining diffusion model with model-based iteration
  [Paper] [Code]

• Unsupervised disentanglement strategy for mitigating artifact in photoacoustic tomography under extremely sparse view
  [Paper] [Code]

• Score-based generative model-assisted information compensation for high-quality limited-view reconstruction in photoacoustic tomography
  [Paper] [Code]