Reconstruction of initial pressure from limited view photoacoustic images using deep learning

Waibel, Dominik J. E.; Gröhl, Janek; Isensee, Fabian; Kirchner, Thomas; Maier‐Hein, Klaus H.; Maier-Hein, Lena

doi:10.1117/12.2288353

Cited by 34 publications

(17 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We chose a constant reduced scattering coefficient of 15 cm −1 in both background and vessel structures. Additional details on the simulation parameters can also be found in our previous work [53]. The raw time-series data was noised after k-space simulation with an additive Gaussian noise model of recorded noise of our system [54].…”

mentioning

confidence: 99%

Confidence Estimation for Machine Learning-Based Quantitative Photoacoustics

et al. 2018

Self Cite

View full text Add to dashboard Cite

In medical applications, the accuracy and robustness of imaging methods are of crucial importance to ensure optimal patient care. While photoacoustic imaging (PAI) is an emerging modality with promising clinical applicability, state-of-the-art approaches to quantitative photoacoustic imaging (qPAI), which aim to solve the ill-posed inverse problem of recovering optical absorption from the measurements obtained, currently cannot comply with these high standards. This can be attributed to the fact that existing methods often rely on several simplifying a priori assumptions of the underlying physical tissue properties or cannot deal with realistic noise levels. In this manuscript, we address this issue with a new method for estimating an indicator of the uncertainty of an estimated optical property. Specifically, our method uses a deep learning model to compute error estimates for optical parameter estimations of a qPAI algorithm. Functional tissue parameters, such as blood oxygen saturation, are usually derived by averaging over entire signal intensity-based regions of interest (ROIs). Therefore, we propose to reduce the systematic error of the ROI samples by additionally discarding those pixels for which our method estimates a high error and thus a low confidence. In silico experiments show an improvement in the accuracy of optical absorption quantification when applying our method to refine the ROI, and it might thus become a valuable tool for increasing the robustness of qPAI methods.

show abstract

mentioning

confidence: 99%

Confidence Estimation for Machine Learning-Based Quantitative Photoacoustics

et al. 2018

Self Cite

View full text Add to dashboard Cite

show abstract

“…In comparison to other approaches, U-Net-based networks generally performed better than other architectures, e.g., compared to a simple three-layer CNN, 98 VGG, 101 and compared to applying U-Net directly to the measurement data g, 104 especially with respect to robustness. It is interesting that Antholzer et al 99 compare their results to a classic l 1 -regularization approach for compressed sensing and report that when the system matrix is randomly sampled, and hence undersampling artifacts change as well, the classical variational approach clearly outperforms the network-based postprocessing approach.…”

Section: Postprocessingmentioning

confidence: 94%

“…This implies that large multiscale networks are needed to transform the signal into the sought-after PAT image effectively. In early studies by Waibel et al 104 and Gröhl et al, 126 it was shown that using an asymmetric U-Net to reconstruct the PA image directly from raw sensor data is feasible in a limitedview setting. In comparison to a postprocessing approach using a U-Net, it was competitive in terms of mean reconstruction error, but exhibited a higher variance in reconstruction error.…”

Section: Convolutional Approachesmentioning

confidence: 99%

Deep learning in photoacoustic tomography: current approaches and future directions

2020

View full text Add to dashboard Cite

Biomedical photoacoustic tomography, which can provide high-resolution 3D soft tissue images based on optical absorption, has advanced to the stage at which translation from the laboratory to clinical settings is becoming possible. The need for rapid image formation and the practical restrictions on data acquisition that arise from the constraints of a clinical workflow are presenting new image reconstruction challenges. There are many classical approaches to image reconstruction, but ameliorating the effects of incomplete or imperfect data through the incorporation of accurate priors is challenging and leads to slow algorithms. Recently, the application of deep learning (DL), or deep neural networks, to this problem has received a great deal of attention. We review the literature on learned image reconstruction, summarizing the current trends and explain how these approaches fit within, and to some extent have arisen from, a framework that encompasses classical reconstruction methods. In particular, it shows how these techniques can be understood from a Bayesian perspective, providing useful insights. We also provide a concise tutorial demonstration of three prototypical approaches to learned image reconstruction. The code and data sets for these demonstrations are available to researchers. It is anticipated that it is in in vivo applications-where data may be sparse, fast imaging critical, and priors difficult to construct by hand-that DL will have the most impact. With this in mind, we conclude with some indications of possible future research directions. © The Authors. Published by SPIE under a Creative Commons Attribution 4.0 Unported License. Distribution or reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI.

show abstract

“…A total of 15580 sets of data are obtained, each of which contains a pair of low-quality images and ground truth, of which 80% are used for training, 10% for verification and the rest for testing. The conventional FBP [7] and deep learning-based (U-Net [21] ) methods are used as comparison methods.…”

Section: Experimental Configurementioning

confidence: 99%

Limited-angle cherenkov-excited luminescence scanned tomography reconstruction based on spatial attention module

Geng

Zhang

et al. 2023

Sixteenth International Conference on Photonics and Imaging in Biology and Medicine (PIBM 2023)

View full text Add to dashboard Cite

Cherenkov-excited luminescence scanned tomography (CELST) is a new emerging imaging modality, which uses the Cherenkov light to excite fluorophores for tomographic imaging. In order to improve the imaging depth and spatial resolution, a rotational CELST was developed to scan the imaging object to produce sinogram data, and a filtered back projection (FBP) was used to recover the distribution of fluorophores. However, the images reconstructed by FBP are usually corrupted by artifacts due to measurements from limited angles. To reduce the artifacts, we propose a deep learning-based reconstruction algorithm (SAM-Unet), which is based on a fully convolutional deep neural network with U-Net structure, and a spatial attention module was added between the encoder and the decoder. The image features extracted by the spatial attention module are transferred to the decoder through a skip connection structure. The effectiveness of the proposed SAM-Unet is verified by numerical experiments, and the results show that the SAM-Unet can improve the mean square error (MSE) (97.5%), peak signal-to-noise ratio (PSNR) (81.9%) and structure similarity index measure (SSIM) (63.4%) compared with the FBP algorithm. Compared with the deep learning method U-Net, the MSE improved 39.8%, the PSNR improved 8.0% and SSIM improved 2.6%.

show abstract

Reconstruction of initial pressure from limited view photoacoustic images using deep learning

Cited by 34 publications

References 14 publications

Confidence Estimation for Machine Learning-Based Quantitative Photoacoustics

Confidence Estimation for Machine Learning-Based Quantitative Photoacoustics

Deep learning in photoacoustic tomography: current approaches and future directions

Limited-angle cherenkov-excited luminescence scanned tomography reconstruction based on spatial attention module

Contact Info

Product

Resources

About