Photoacoustic digital brain and deep-learning-assisted image reconstruction

Image Enhancement Method for Photoacoustic Imaging of Deep Brain Tissue

“…PAI leverages this principle to measure the optical absorption of endogenous contrast agents within brain tissue [1]. In this study, a multi-wavelength algorithm employing wavelengths of 760 nm, 840 nm, and 930 nm was used to assess brain tissue hemodynamics, including oxygen concentration (sO 2 ), oxygenated hemoglobin concentration (HbO 2 ), deoxyhemoglobin concentration (HbR), and water concentration (H 2 O) [4,6]. As depicted in Figure 7, hemodynamic images were acquired from both the normal group (shown in Figure 7a-d) and the cerebral hemorrhage group (shown in Figure 7i-l) and subsequently enhanced to produce images for the enhanced control group (Figure 7e-h) and the cerebral hemorrhage group (Figure 7m-p).…”

“…Photoacoustic imaging (PAI) is an innovative biomedical imaging technique that leverages the photoacoustic effect to transform laser energy into acoustic energy through light absorption and subsequent thermal expansion [1][2][3]. PAI offers the dual benefits of high optical contrast and superior ultrasound penetration resolution [1,4,5]. Additionally, it allows for the quantitative determination of hemodynamic parameters such as hemoglobin, oxygen, and water concentration from multi-wavelength photoacoustic data.…”

Photoacoustic imaging for detection of blood clot with different sizes in the brain

“…The proposed method compensates for experimental signals in terms of their attenuation, time-of-flight, and signal broadening to achieve accurate correction. More recently, two separate studies have been published, both introducing methods that utilize a specific deep-neural network known as U-Net [54] for analyzing simulated PAI data [55][56][57]. In the work by Zhang et al [56], they proposed an image correction technique for simulated brain PAI using U-Net trained with simulated data [58].…”

“…More recently, two separate studies have been published, both introducing methods that utilize a specific deep-neural network known as U-Net [54] for analyzing simulated PAI data [55][56][57]. In the work by Zhang et al [56], they proposed an image correction technique for simulated brain PAI using U-Net trained with simulated data [58]. In the study conducted by Gao et al [57], they proposed a modified version of the U-Net.…”