This training strategy enabled us to enhance the bidirectional scanning images acquired in real imaging scenarios. Rather than using simulated or image-processed data, we trained and validated our DNN by using paired in-vivo OR-PAM images obtained via unidirectional and bidirectional scanning. Here, we present a scanning framework based on deep neural networks (DNNs) to halve the imaging acquisition time by correcting the misaligned OR-PAM images obtained via bidirectional scanning.

Specifically, an interesting application of DL techniques lies in improving the imaging system performance, such as quality and speed 26, 27, 37, 38. Consequently, the PA images are generally reconstructed using only unidirectionally collected data, while the data obtained from the other path are is discarded at the cost of doubling the image acquisition time.ĭeep learning (DL), which is a rapidly developing field, has addressed the challenges associated with enhancement, translation, segmentation, classification, and medical decision-making in biomedical imaging tools such as PAI, computed tomography (CT), ultrasound imaging, magnetic resonance angiography (MRI), and optical microscopy 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37. Owing to this instability, the time interval between scanning points are non-uniform, resulting in a misalignment between the bidirectional raster scanning paths. However, high-speed water-immersible scanners exhibit unstable raster scanning performance owing to heat and vibration. This high scanning speed allows us to adapt a highly repeated pulsed laser system operating at several hundred kHz in OR-PAM, significantly shortening the imaging acquisition time. In particular, water-immersible MEMS and galvanometer scanners with tunable scanning ranges have been extensively used in various studies, accelerating the B-scan imaging speed up to hundreds of Hz 11, 12, 13, 14, 15, 22, 23. Over the past few years, OR-PAM has utilized various high-speed scanning tools such as a voice-coil stage, microelectromechanical system (MEMS) scanners, polygonal mirror scanners, and galvanometer scanners to enhance the effective B-scan rate under point-by-point raster scan 2, 8, 9, 10, 11, 12, 13, 14, 15, 21, 22, 23, 24. OR-PAM has made significant contributions to the exploration of molecular, anatomical, and functional information for in vivo animal and human imaging in a variety of research fields, including biology, oncology, neurology, ophthalmology, dermatology, and pathology 2, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20. Optical-resolution photoacoustic microscopy (OR-PAM) is a type of microscopic PAI that uses a tightly focused optical beam to detect emitted ultrasound in the optical ballistic or quasi-ballistic regime to achieve a high lateral resolution and signal-to-noise ratio (SNR) with a point-by-point raster scan 2. The pressure rise results in the generation of ultrasonic waves called photoacoustic (PA) waves. When excited by an optical pulse, biomolecules experience an increase in temperature due to optical absorption, and hence a pressure rise due to thermo-elastic expansion.

Photoacoustic imaging (PAI) is a hybrid biomedical imaging technique that detects ultrasonic waves induced by the excitation of short optical pulses 1, 2, 3, 4, 5, 6, 7. Our DNN-assisted raster scanning framework can further potentially be applied to other raster scanning-based biomedical imaging tools, such as optical coherence tomography, ultrasound microscopy, and confocal microscopy. The proposed method doubles the imaging speed compared to that of conventional methods by aligning nonlinear mismatched cross-sectional B-scan photoacoustic images during bidirectional raster scanning. Here, we demonstrate a scanning framework based on a deep neural network (DNN) to correct misaligned PAM images acquired via bidirectional raster scanning.

Therefore, only one unidirectionally acquired image is typically used consequently, the imaging speed is reduced by half. However, when using high-speed water-immersible scanners, the two consecutively acquired bidirectional PAM images are misaligned with each other because of unstable performance, which causes a non-uniform time interval between scanning points. Simultaneous point-by-point raster scanning of optical and acoustic beams has been widely adapted to high-speed photoacoustic microscopy (PAM) using a water-immersible microelectromechanical system or galvanometer scanner.

Raster scan waveforms digilent