15 September 2021 in Biomedical Optics Express
Photoacoustic tomography (PAT) is an emerging biomedical imaging technology that can realize high contrast imaging with a penetration depth of the acoustic. Recently, deep learning (DL) methods have also been successfully applied to PAT for improving the image reconstruction quality. However, the current DL-based PAT methods are implemented by the supervised learning strategy, and the imaging performance is dependent on the available ground-truth data. To overcome the limitation, this work introduces a new image domain transformation method based on cyclic generative adversarial network (CycleGAN), termed as PA-GAN, which is used to remove artifacts in PAT images caused by the use of the limited-view measurement data in an unsupervised learning way. A series of data from phantom and in vivo experiments are used to evaluate the performance of the proposed PA-GAN. The experimental results show that PA-GAN provides a good performance in removing artifacts existing in photoacoustic tomographic images. In particular, when dealing with extremely sparse measurement data (e.g., 8 projections in circle phantom experiments), higher imaging performance is achieved by the proposed unsupervised PA-GAN, with an improvement of ∼14% in structural similarity (SSIM) and ∼66% in peak signal to noise ratio (PSNR), compared with the supervised-learning U-Net method. With an increasing number of projections (e.g., 128 projections), U-Net, especially FD U-Net, shows a slight improvement in artifact removal capability, in terms of SSIM and PSNR. Furthermore, the computational time obtained by PA-GAN and U-Net is similar (∼60 ms/frame), once the network is trained. More importantly, PA-GAN is more flexible than U-Net that allows the model to be effectively trained with unpaired data. As a result, PA-GAN makes it possible to implement PAT with higher flexibility without compromising imaging performance.
24 September 2018 in IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control
The present study was designed to investigate the associations among ultrasonic backscatter, bone densities and microstructure in bovine cancellous bone. Ultrasonic backscatter measurements were performed on 33 bovine cancellous bone specimens with a 2.25-MHz transducer. Ultrasonic apparent backscatter parameters (‘apparent’ means not compensating for ultrasonic attenuation and diffraction) were calculated with optimal signal of interests (SOI). The results showed that ultrasonic backscatter was significantly related to bone densities and microstructure (R2=0.17-0.88, p<0.05). After adjusting the correlations by bone mineral density (BMD), the bone apparent density (BAD) and some trabecular structural features still contributed significantly to the adjusted correlations, with moderate additional variance explained (ΔR2=9.7% at best). Multiple linear regressions revealed that both BAD and trabecular structure contributed significantly and independently to the prediction of ultrasound backscatter (adjusted R2=0.75-0.89, p<0.05), explaining an additional 14.0% of the variance at most, compared with that of BMD measurements alone. The results proved that ultrasonic backscatter was primarily determined by BAD, not BMD, but the combination of bone structure and densities could achieve encouragingly better performances (89% of the variance explained at best) in predicting backscatter properties. This study demonstrated that ultrasonic apparent backscatter might provide additional density and structural features unrelated to current BMD measurement. Therefore, we suggest that ultrasonic backscatter measurement could play a more important role in cancellous bone evaluation.
4 November 2019 in IEEE Access
Ultrasonic imaging provides a non-invasive way to diagnose brain disease. However, due to imaging degradation effects from the phase-aberration and reverberation, it is still challenging to achieve an accurate transcranial imaging. The objective of this work is to improve the quality of transcranial imaging. To this end, a ray theory based transcranial phase correction method was proposed to correct the phase abberation induced by cranial bones. With the pre-knowledge of the shape and longitudinal velocity of the cranial bones, the corrected phases are derived by solving eikonal equation (ray-theory). The Ideal Synthetic Aperture Focusing Technique (I-SAFT) is applied for signal acquisitions in simulations and in-vitro phantom experiment with one-element transmitting and all-element receiving method. Dynamic focusing is achieved at each imaging position with I-SAFT and the transcranial imaging distortion is modified with the proposed phase correction method. Simulations and experiments show that the imaging distortion of target circular phantoms was corrected, and the imaging quality is improved after the phase correction. With the proposed method, the average error of the central position of target phantoms decreases from 1.98 mm to 0.21 mm, the eccentricity of fitted ellipse averagely decreases from 0.63 to 0.19, and the average maximum luminance contrast of phantoms improves from 37.36dB to 42.41dB. It is illustrated that the proposed ray-theory based phase correction method might be useful for intracranial imaging.
15 May 2023 in Ultrasonics
27 July 2023 in Ultrasonics
1 February 2020 in Ultrasound in Medicine & Biology