TT-PADM: A Time-Driven Transformer Diffusion Model for Robust Sparse-View and Limited-View Photoacoustic Tomography.

Photoacoustic tomography (PAT) is an effective optical biomedical imaging method which is characterized with noninonizing and noninvasive, presenting good soft tissue contrast with excellent spatial resolution. To build a multi-dimensional breast PAT image, more ultrasound sensors are needed, which brings difficulties to data acquisition. The time complexity for multi-dimensional breast PAT image reconstruction also rises tremendously. Compressive sensing (CS) theory breaks the restriction of Nyquist sampling theorem and is capable to rebuild signals with fewer measurements. In this contribution, we propose an effective optimization method for multi-dimensional breast PAT, which combines the theory of CS and an unevenly, adaptively distributing data acquisition algorithm. With this method, the quality of our reconstructed breast PAT images are better than those using existing multi-dimensional breast PAT system. To build breast PAT images with the same quality, the required number of ultrasound transducers is decreased by using our proposed method. We have verified our method on simulation data and achieved expected results in both two dimensional and three dimensional PAT image reconstruction. In the future, our method can be applied to various aspects of biomedical PAT imaging such as early stage tumor detection and in vivo imaging monitoring.

Learning spatially variant degradation for unsupervised blind photoacoustic tomography image restoration

Generating an image of acceptable quality will take several minutes in circular scanning geometry-based Photoacoustic tomographic (PAT) imaging systems. Although, the imaging speed can be improved by employing multiple single-element ultrasound transducers (UST) and faster scanning. The low signal-to-noise ratio at higher and the artifacts arising from sparse signal acquisition hamper the imaging speed. Thus, there exists a need to improve the speed of the PAT imaging system without compromising the image quality. To improve the frame rate of the PAT system, we propose a convolutional neural network (CNN) based deep learning architecture for reconstructing the artifact-free PAT images from the fast-scanning data. The proposed model is trained with the simulated dataset and its performance was evaluated using experimental phantom and in-vivo imaging. The efficiency to improve the frame rate was evaluated on both the single-UST and multi-UST PAT systems. Our results suggest that the proposed deep learning architecture improves the frame rate by six-fold in a single UST PAT system and by two-fold in a multi-UST PAT system. The fastest frame rate of ~ 3Hz was achieved without compromising the quality of the PAT image.

Photoacoustic tomography (PAT) is a noninvasive vascular imaging modality that uses near-infrared pulse laser beams and ultrasound (US) to visualize vessels. We previously demonstrated the utility of PAT for visualizing anterolateral thigh (ALT) perforators in a clinical study of 10 thighs in 5 healthy adults. Evaluation of the correlation between PAT and US findings showed that PAT had comparable diagnostic potential but was superior in visualizing subcutaneous microvessels; however, there was no comparison with intraoperative findings. In this study, we used a newly developed technique to transfer a PAT image to a body-attachable transparent sheet to compare PAT and intraoperative findings. Eight patients were recruited in this prospective study. Patient age ranged from 32 to 79 years (average 60). Seven ALT flaps were applied in head and neck reconstruction. One flap was elevated in chest wall reconstruction. Each PAT scan of an 18 cm × 13.5 cm region took approximately 5 min. Acquired data were processed three-dimensionally using a novel imaging software program. Perforator vessel data from PAT imaging were traced and corrected for projection onto medical film sheets. The correlation between the perforator stem portions predicted by PAT and the intraoperative findings at the level of the fascia-penetrating points was evaluated, and distal branching patterns were analyzed. PAT imaging showed 16 perforators in 8 thighs. Intraoperative surgical findings revealed that all the perforator penetrating points at the deep fascia level matched the PAT findings within 10 mm. None of the eight ALT flaps demonstrated postoperative complications. The perforator complexes were classified as type I in three cases (19%), type II in eight cases (50%), and type III in five cases (31%). PAT imaging matched the intraoperative findings within 10 mm. Preoperative vascular evaluation allows for the creation of a vascular map for facilitating ALT flap surgeries.

Basal cell carcinoma (BCC) is the most common form of eyelid skin cancer, and the standard technique for removal is excision with a predetermined marginal and postoperative histological analysis. Mohs surgery is an alternative technique that allows examination of the tumour margins by staged resection and simultaneous histopathological examination (Beaulieu et al. 2018). It is considered the most efficient method for the complete removal of high risk and complicated BCC. However, the procedure is time consuming and expensive, which limits its practice to specific cases. We here present a case where photoacoustic imaging (PA), an emerging non-invasive biomedical imaging modality, was used to detect that a pentagonal excision of an eyelid BCC was non-radical, which was later confirmed by histological examination. Photoacoustic imaging (PA) combines the advantages of optical spectroscopy and ultrasound and can provide high-resolution 3D images of the molecular composition of tissue. However, hitherto mainly animal studies have been performed, and only a few studies have been carried out on human skin tumours (Attia et al. 2017; Chuah et al. 2018), mostly melanomas. No study has yet described PA imaging of a BCC on the face or the eyelids, or used it to determine whether an excision is complete. An 87-year-old Caucasian male presented at the Department of Ophthalmology at Skåne University Hospital, Sweden, with a suspicious skin-coloured lesion on the left lower eyelid, of at least 6 months’ duration. Clinically, a thin nodular BCC was suspected, and excision was planned. The lesion measured approximately 6 mm in diameter, and was located centrally, a few millimetres below the lash line. The tumour was surgically removed using a pentagonal excision procedure under local anesthesia, aiming at a surgical margin of 3–4 mm. The lesion was then placed in a plastic container filled with buffered saline solution, for ex vivo PA imaging using a Vevo LAZR-X imaging system (FUJIFILM VisualSonics Inc., Toronto, ON, Canada). Photoacoustic images were acquired using 59 wavelengths in the range 680–970 nm, and three-dimensional images were obtained by scanning the tumour using a linear stepping motor. Spectral unmixing was employed for the 3D scan to produce a coloured PA image, mapping the spatial distribution of tumour cells by means of their distinct optical absorption spectrum. After PA scanning, the lesion was examined histologically using standard haematoxylin and eosin staining. For methodological details see, Sheikh et al. (2018). The results of the PA examination showed a clear difference in the spectral signature from the BCC, defined as the middle of the visible tumour, and healthy tissue, defined as the tip of the pentagonal excision (Fig. 1A). Multiwavelength 3D scanning and spectral unmixing provided a map of the overall lesion architecture, and clearly revealed the tumour cell distribution (Fig. 1B). The tumour signal extended all the way out to the medial marginal, meaning that according to PA examination, the excision was non-radical. Subsequent histopathological examination revealed intermediate/infiltrating BCC, confirming that the excision was not complete in its medial portion (Fig. 1C). These results indicate that PA imaging could be used to delineate tumours in the periorbital region to achieve a more precise excision with better clearance, reducing the need for repeated excision. Further development is required to create a technique suitable for presurgical definition and delineation of the tumour borders. In the future, PA could also potentially be used for rapid examination of freshly excised unfixed tissue during surgery, to determine whether the excision is radical, or if further excision is necessary, saving time compared to conventional microscopic histology as in Mohs micrographic surgery.

In this paper we present a new high-contrast photoacoustic tomography (PAT) imaging system using a 4f acoustic lens, a 64-element linear transducer array and peak-hold technology. This PAT imaging system has been developed to obtain three-dimensional (3D) PAT images of experimental samples. By utilizing a 4f acoustic lens, the photoacoustic (PA) signals generated from the sample are directly imaged on the imaging plane and collected by the 64-element linear transducer array, which changes them into the corresponding electronic signals. Then we can get one-dimensional (1D) images from the electronic signals using a peak detection-and-hold circuit. After vertical scanning with a stepping motor on the imaging plane, a 2D PA image of the sample is successfully obtained. Combined with the time-resolved technique, we can then get 3D PAT images. The results show that the reconstructed images agree well with the original samples.

In circular scanning geometry based Photoacoustic tomographic (PAT) imaging systems, the axial resolution is spatially invariant and is limited by the bandwidth of the detector. However, the tangential resolution is spatially varying and is dependent on the aperture size of the detector. Conventionally, large aperture size detectors are the preferred choice for detection element in circular view PAT imaging systems but it hampers the tangential resolution. Although several techniques have been proposed to improve the tangential resolution, they are also hindered by their inherent limitations. Herein, we propose a deep learning architecture to counter the spatially variant tangential resolution in circular scanning PAT.We used a (U-Net) based CNN architecture to improve the tangential resolution of the acquired PAT images. Our results suggest the proposed deep learning architecture improves the tangential resolution of the acquired by five folds, without compromising the image quality.

Signal-domain speed-of-sound correction for ring-array-based photoacoustic tomography

This work describes the design, development and added value of breast-supporting cups to immobilize and position the pendant breast in photoacoustic tomographic imaging. We explain the considerations behind the choice of the material, the shape and sizes of a cup-shaped construct for supporting the breast in water in an imaging tank during full-breast imaging. We provide details of the fabrication, and other processing and testing procedures used. Various experiments were conducted to demonstrate the added value of using these cups. We show that breast movement during a measurement time of four minutes is reduced from maximum 2 mm to 0.1 mm by the use of cups. Further, the presence of the cup, centered in the aperture leading to the imaging tank, ensures that the breast can be reproducibly positioned at the center of the field-of-view of the detection aperture in the tank. Finally, since an accurate delineation of the water-tissue boundary can now be made, the use of the cup enables accurate application of a two-speed of sound model for reconstruction. All in all, we demonstrate that the use of cups to support the breast provides clear enhancement in contrast and resolution of breast images in photoacoustic imaging.

The combined effect of the transducer size and the width of the incident exciting light pulse in photoacoustic tomography (PAT) imaging has been studied. The transducers have been apodized with the Gaussian function and its full width at half-maximum (FWHM) has been varied to see its effect on blurring in PAT imaging. The Wiener filter has been used to deconvolve the PA signal with the input light pulse in order to reduce the blurring effect caused by the broad light pulse. Finally, the PA image was reconstructed using the time reversal interpolation algorithm. The two dimensional simulations were conducted for disc and vasculature phantoms using k-Wave toolbox. The results generated using finite transducers were compared with that of point detectors. The Pearson correlation coefficient value, for a Gaussian excitation pulse of width $1.0 \mu \mathbf{s}$ and for a vasculature phantom, increased from 0.17 to 0.82 when FWHM of a 6 mm transducer was 1.2 mm whereas it increased from 0.30 to 0.69 for the same transducer with 12.9 mm as FWHM. In general, it has been found that the larger diameter transducers have more blurring effect. Moreover, the images produced for small FWHM values are having less blur than those formed for larger FWHM values.

We built a photoacoustic tomographic (PAT) imaging system by scanning a single detector (φ 3.5 mm) made of piezoelectric copolymer poly(vinylidene difluoride-trifluoroethylene), P(VDF-TrFE), which had been fabricated for diagnostic photoacoustic measurement of cartilage tissues in our group. The PAT images of a phantom were obtained at two excitation wavelength of 687.5 nm and 795 nm. The phantom was made of agar including a black hair and agarose gels dissolving indocyanine green (ICG) and methylene blue (MB). Laser pulses (685-900 nm) were generated from a Ti:Sappire tunable laser to excite ICG and MB molecules. The PAT image at 687.5 nm shows signals due to all absorption sources. This is good agreement with dimension of the phantom. The PAT image at 795 nm shows a strong signal due to the ICG-dyed gel and almost no signal due to the MB-dyed gel. This result indicated that absorption sources were extracted by excitation wavelength according to their absorption spectra. The signal/noise ratio of the PAT images were compared between the P(VDF-TrFE) transducer in our group and a PZT transducer (Parametrics V309, 5 MHz, φ 12.7 mm) which is commercially available. The P(VDF-TrFE) transducer was more sensitive by 9 times (120 times per area) than the PZT transducer. By using this imaging system with a P(VDF-TrFE) transducer which is highly sensitive in a wide frequency range, we will achieve frequency analysis of the PAT images to associate photoacoustic waveforms with physical properties of sample tissues.

The spatial resolution of photoacoustic tomography (PAT) can be characterized by the point spread function (PSF) of the imaging system. Due to the tomographic detection geometry, the PAT image degradation model could be generally described by using spatially variant PSFs. Deconvolution of the PAT image with these PSFs could restore image resolution and recover object details. Previous PAT image restoration algorithms assume that the degraded images can be restored by either a single uniform PSF, or some blind estimation of the spatially variant PSFs. In this work, we propose a PAT image restoration method to improve image quality and resolution based on experimentally measured spatially variant PSFs. Using photoacoustic absorbing microspheres, we design a rigorous PSF measurement procedure, and successfully acquire a dense set of spatially variant PSFs for a commercial cross-sectional PAT system. A pixel-wise PSF map is further obtained by employing a multi-Gaussian-based fitting and interpolation algorithm. To perform image restoration, an optimization-based iterative restoration model with two kinds of regularizations is proposed. We perform phantom and in vivo mice imaging experiments to verify the proposed method, and the results show significant image quality and resolution improvement.

Photoacoustic tomography (PAT) is a promising imaging technique that utilizes the detection of light-induced acoustic waves for both morphological and functional biomedical imaging. However, producing high-quality images using PAT is still challenging and requires further research. Besides improving image reconstruction, which turns the raw photoacoustic signal into a PAT image, an alternative way to address this issue is through image post-processing, which can enhance and optimize the reconstructed PAT image. Image post-processing methods have rapidly emerged in PAT and are proven to be essential in improving image quality in recent research. In this review, we investigate the need for image post-processing in PAT imaging. We conduct a thorough literature review on the latest PAT image post-processing articles, including both general and PAT-specific post-processing techniques. In contrast to previous reviews, our analysis focuses specifically on advanced image post-processing rather than image reconstruction methods. By highlighting their potential applications, we hope to encourage further research and development in PAT image post-processing technology.

Although many algorithms are available for full-view photoacoustic tomography (PAT), no exact and stable algorithm for limited-view PAT has been proposed. In this paper the deconvolution reconstruction (DR) algorithm is proposed for both full-view and limited-view PAT. In the DR algorithm, first a new function is constructed from detected photoacoustic signals and approximately simplified, and then the tissue's electromagnetic absorption is derived from this function on the basis of Fourier-based deconvolution. Computer simulations are carried out to compare the DR algorithm with two popular PAT algorithms, the time-domain reconstruction (TDR) and the filtered back projection (FBP). Although the error of the DR algorithm increases with the size of the detected object, it is shown that the DR algorithm has good precision and strong robustness to noise in the full-view PAT, nearly equivalent to the TDR and FBP. Yet the DR algorithm is more than ten times faster in computation speed. In the limited-view PAT, the DR is superior to the TDR and FBP in terms of both accuracy and robustness to noise.

Background: Photoacoustic tomography (PAT) is an evolving technique that is capable of obtaining real-time, high-resolution images of tissues. The purpose of this research was to evaluate morphological and functional information of the rabbit airway using PAT. Methods: PAT experiments were performed with nine New Zealand white rabbits. Here we measured the cross-sectional tracheal wall thickness and cross-sectional area of airway in vivo and in vitro. PAT airway images and the corresponding histological findings were compared. Results: PAT images showed the continuous variation of tracheal lumen size at different phases of the respiratory cycle. There were no significant differences between tracheal wall thickness by PAT imaging in vivo and the corresponding pathological tissue (p>0.05). No significant differences were found between cross-sectional area of airway wall in vivo and its corresponding fresh specimens by PAT imaging (p<0.05), while differences were found between cross-sectional area of airway wall in vivo and its corresponding formalin-fixed specimen by PAT imaging (p<0.01).