Photoacoustic (PA) computed tomography (PACT) combines the superiorities of both optical imaging and ultrasound imaging, which is based on the generation of acoustic signal (PA wave) by short pulsed laser light. However, in many applications, the sensor array can only partially enclose the target, resulting in limited-view setup and image reconstruction with severe artifacts. Standard reconstruction always induces artifacts using limited-view signals, which are influenced by limited angle coverage of transducers, finite bandwidth, and uncertain heterogeneous biological tissue. To address these issues, in this paper, a deep learning based approach is proposed to compensate the missing data due to limited view (e.g. only 90 degrees coverage). To acquire the missing data from the other 270 degrees’ view, we trained an end-to-end network to recover the limited-view PA data, which have been delayed to form a pressure map in region of interest.
KEYWORDS: Data acquisition, Acquisition tracking and pointing, Photoacoustic tomography, Signal detection, Biomedical optics, Ultrasonography, Transducers, Photoacoustic spectroscopy, New and emerging technologies, Image restoration
Photoacoustic tomography (PAT) is an emerging technology for biomedical imaging that combines the superiorities of high optical contrast and acoustic penetration. In the PAT system, more photoacoustic (PA) signals are preferred to be detected to reconstruct PA image with higher fidelity. However, more PA signals’ detection leads to more time consumption for single-channel scanning based PAT system, or higher cost of data acquisition (DAQ) module for array-based PAT system. To address this issue, we proposed a real-time PAT system only using single DAQ channel, and a deep learning method for PA signal recover and image reconstruction. We superimpose 30 channels’ signals together, shrinking to 4 channels (120/30=4). Furthermore, a four-to-one delay-line module is designed to combine this 4 channels’ data into one channel DAQ. In order to reconstruct the image from four superimposed 30-channels’ PA signals, we train a dedicated deep learning model to reconstruct final PA image.
Photoacoustic Tomography (PAT) is a novel biomedical imaging modality, whose reconstructed image is based on the target responds of the instantaneous pulse laser excitation. The target emits photoacoustic (PA) signal upon the laser impulse excitation, which contains the information of light absorption of the target. To reconstruct a high quality image of PAT, recovering the full bandwidth of the PA signal is necessary. While practical transducers cannot cover the full bandwidth of the PA signal, and has low sensitivity for ultralow frequency PA signal that induce the PA image distortion. In this study, a broadband PA signal enhancement method with morphological frequency convolution is proposed to recover PA signal. From the original reconstructed PA image frequency analyses, we get the convolution kernels and process the PA signal. The recovered full-bandwidth signal reconstructed the PA image more realistic and with less distortion.
Photoacoustic tomography (PAT) is a non-invasive imaging technique which provided high lateral resolution and axial resolution. Conventional linear photoacoustic (PA) imaging has been widely applied to state-of-art PA system, which excited by short pulse laser. Recently, nonlinear photoacoustic effect has been excavated and utilized, which indicates the different character with linear PA by dual-pulse laser. In this paper, we report a dual-contrast photoacoustic sensing by quasi-CW nonlinear PA effect. Using the high-repetition pulsed laser, and two different contrast imaging were extracted. Unless conventional contrast that using short laser pulse by detected tissue light absorption, we extracted another contrast that indicated the rising of temperature. The quasi-CW nonlinear PA effect could be illustrated by a simple mathematical derivation. The dual-contrast PA imaging have been demonstrated by experiment in vitro and ex vivo imaging. The results indicate different heat absorbency of different materials, which could be utilized to distinguish materials and intensify imaging contrast. Compared with conventional PA imaging, the proposed method has more scope of works and potential applications.
As an emerging hybrid imaging modality, photoacoustic imaging has attracted intensive research interest in recent years in various applications, such as breast cancer detection, brain imaging, and intravascular imaging, which provides functional and molecular information. In a typical photoacoustic imaging system, laser intensity fluctuation needs to be monitored by a photodiode (PD) that can provide proper normalization for photoacoustic signals. Conventionally, at least two data acquisition channels are necessary to receive both photoacoustic signal and photodiode signal. In this paper, we propose a simple and efficient method to receive both photoacoustic and photodiode signals using single data acquisition channel, which gives lower system cost and faster system speed. After connecting the photodiode output and ultrasound transducer for hybrid signal acquisition, sharing channel can be achieved in two ways: Use a direct separation algorithm when received a PA signal with desirable signal-to-noise. Filters are exploited to keep low SNR PA signal immune from the photodiode signal and separate them from the receiving hybrid signal. 2D PA images based on the separated PA and PD signals will be illustrated to demonstrate their performance and efficiency. This method will be valuable especially when designing a PAT imaging system with multi-channel ultrasound array and data-acquisition card.
Photoacoustic (PA) imaging has attracted increasing research interest in recent years due to its unique merit of combining light and sound. Enabling deep tissue imaging with high ultrasound spatial resolution and optical absorption contrast, PA imaging has been applied in various application scenarios including anatomical, functional and molecular imaging. However, the bulky and expensive laser source is one of the key bottlenecks that needs to address for further compact system development. Photoacoustic imaging system based on low-cost laser diode is one of the promising solutions. In this paper, we report a custom-made fingertip laser diode system enabling both pulsed and continuous modulation modes with the shortest pulse width of 30 ns, driving current of 10 A, and single modulation frequency of 3 MHz, which is suitable for both time and narrow-band frequency domain PA imaging. The experiments for generating PA signals were performed with more than 70 millivolts signals amplitude. By sweeping the pulse width, it is observed that the amplitude of PA signals is increasing due to higher laser energy. To the best of our knowledge, this may be the most compact laser source used for photoacoustic applications for PA imaging. Owing to its super-compact size, the reported laser diode system could pave the pathway to low-cost photoacoustic sensing and imaging device, even wearable photoacoustic biomedical sensors.
Nowadays, breast cancer has increasingly threatened the health of human, especially females. However, breast cancer is still hard to detect in the early stage, and the diagnostic procedure can be time-consuming with abundant expertise needed. In this paper, the main research is the application of deep learning method in the diagnosis of photoacoustic breast cancer and the comparison of the performance of the traditional machine learning classification algorithm and deep learning method in the actual scenario of the photoacoustic imaging breast cancer diagnosis. The traditional supervised learning method firstly obtains the photoacoustic images of breast cancer through preprocessing, extracts the SIFT features, and uses K-means clustering to obtain the feature dictionary. The histogram of the feature dictionary was used as the final feature of the image. Support vector machine (SVM) was used to classify the final features, achieving an accuracy of 82.14%. In the deep learning method, AlexNet and GoogLeNet were used to perform the transfer learning, achieving 88.23%, 89.23%, and 91.18% accuracy, respectively. Finally, by comparing the AUC, sensitivity, and specificity of SVM with AlexNet and GoogLeNet, it can be concluded that the combination of deep learning and photoacoustic imaging obtain a profound and important impact on clinical applications.
Accurate diagnosis of malignancy tumor in early stage is great significance to achieve high curability, which could improve survival rate in this stage. Precise classification to differentiate malignancy of tumors is favourable to reduce cost in treatment when there is no obvious features in radiology diagnose in early phase. Photoacoustic tomography (PAT) is a burgeoning new imaging modality, which combines optical contrast and ultrasound penetrating in deep medium. However, it has not been fully exploited on the capability of PAT to discriminate tumor’s malignancy. In this paper, a multistatic classification approach in PAT is proposed, which could discriminate malignant/benign tumors based on its morphological feature in clinical diagnosis that tumors usually show different shape irregularity compared with healthy tissue. The multistatic photoacoustic waves were used to extract two different features to differentiate the two types of tumors with high accuracy (<90%) in three different scenarios using Support Vector Machines (SVM). In addition, two conventional PAT image reconstructing algorithms are also performed to reconstruct images as a comparative study, which unfortunately cannot differentiate their malignancy precisely because of limited detector bandwidth and severe acoustic distortion. We performed the feasibility study in this paper with both simulation and experimental results, which shows that the proposed multistatic photoacoustic classification method to distinguish between malignant and benign tumors works well, and could be easily applied for state-of-art array-based PAT system to ameliorate the diagnostic accuracy.
Photoacoustic (PA) tomography is an imaging technology that reconstructs the distribution of light absorption in tissue by photoacoustic signals. In recent years, PA tomography has been widely used in anatomical, functional and molecular imaging. However, one of the great challenges is that the efficiency of light to sound conversion is very low due to photoacoustic effect, resulting in low signal-to-noise ratio (SNR) of photoacoustic signal, especially for deep tissue imaging. Conventional approach to enhance the SNR of photoacoustic signal is data averaging, which is quite time-consuming. In the absence of signal fidelity and imaging speed, an algorithm of using empirical mode decomposition (EMD) and independent component analysis (ICA) de-noising in photoacoustic tomography is proposed. Firstly, the photoacoustic signal is decomposed into a series of intrinsic mode functions (IMFs) with EMD. Each IMF is equivalent to an independent signal. Then, some IMFs are selected to construct the virtual noise channel according to the correlation between IMF and original photoacoustic signal. Finally, the original photoacoustic signal and the virtual noise channel are regarded as the input data for ICA. ICA extracts useful photoacoustic signals from artificially constructed multidimensional data. The de-noised results are compared with that the wavelet de-noising method and bandpass filtering method. The enhancement of the SNR of the photoacoustic signal and the contrast of the reconstructed image have been well demonstrated. The proposed method provides the potential to develop real-time low-cost PA tomography system with low-power laser source and poor PA signal’s SNR.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.