Alexandre Goy

Learning approach to optical tomography

Optical tomography has been widely investigated for biomedical imaging applications. In recent years optical tomography has been combined with digital holography and has been employed to produce high-quality images of phase objects such as cells. In this paper we describe a method for imaging 3D phase objects in a tomographic configuration implemented by training an artificial neural network to reproduce the complex amplitude of the experimentally measured scattered light. The network is designed such that the voxel values of the refractive index of the 3D object are the variables that are adapted during the training process. We demonstrate the method experimentally by forming images of the 3D refractive index distribution of Hela cells.

Digital confocal microscopy through a multimode fiber.

Confocal laser-scanning microscopy is a well-known optical imaging method where a pinhole is used in the illumination and detection pathways of a normal microscope, in order to selectively excite and detect a particular focal volume. The advantage of this method is a significant increase in contrast, due to the rejection of background contributions to the signal. Here, we propose to apply this method in the context of multimode fiber endoscopy. Due to modal scrambling, it is not possible to use a physical pinhole to filter light signals that have travel through multimode fibers. Instead, we use a transmission matrix approach to characterize the propagation of light through the fiber, and we apply the filtering operation in the digital domain.

Digital confocal microscope

We demonstrate experimentally a scanning confocal microscopy technique based on digital holography. The method relies on digital holographic recording of the scanned spot. The data collected in this way contains all the necessary information to digitally produce three-dimensional images. Several methods to treat the data are presented. Examples of reflection and transmission images of epithelial cells and mouse brain tissue are shown.

Optical Tomographic Image Reconstruction Based on Beam Propagation and Sparse Regularization

Optical tomographic imaging requires an accurate forward model as well as regularization to mitigate missing-data artifacts and to suppress noise. Nonlinear forward models can provide more accurate interpretation of the measured data than their linear counterparts, but they generally result in computationally prohibitive reconstruction algorithms. Although sparsity-driven regularizers significantly improve the quality of reconstructed image, they further increase the computational burden of imaging. In this paper, we present a novel iterative imaging method for optical tomography that combines a nonlinear forward model based on the beam propagation method (BPM) with an edge-preserving three-dimensional (3-D) total variation (TV) regularizer. The central element of our approach is a time-reversal scheme, which allows for an efficient computation of the derivative of the transmitted wave-field with respect to the distribution of the refractive index. This time-reversal scheme together with our stochastic proximal-gradient algorithm makes it possible to optimize under a nonlinear forward model in a computationally tractable way, thus enabling a high-quality imaging of the refractive index throughout the object. We demonstrate the effectiveness of our method through several experiments on simulated and experimentally measured data.

Imaging in focusing Kerr media using reverse propagation [Invited]

We present imaging experiments in focusing Kerr media using digital holography and digital reverse propagation (DRP) of the wave. For moderate power, the nonlinear DRP algorithm can be used to improve the quality of images over the linear DRP. We discuss the limits of the method at high power, the role of small-scale filaments, and the problem of time-dependent self-phase modulation.

Multiple contrast metrics from the measurements of a digital confocal microscope

We describe various methods to process the data collected with a digital confocal microscope (DCM) in order to get more information than what we could get from a conventional confocal system. Different metrics can be extracted from the data collected with the DCM in order to produce images that reveal different features of the sample. The integrated phase of the scattered field allows for the three-dimensional reconstruction of the refractive index distribution. In a similar way, the integration of the field intensity yields the absorption coefficient distribution. The deflection of the digitally reconstructed focus reveals the sample-induced aberrations and the RMS width of the focus gives an indication on the local scattering coefficient. Finally, in addition to the conventional confocal metric, which consists in integrating the intensity within the pinhole, the DCM allows for the measurement of the phase within the pinhole. This metrics is close to the whole-field integrated phase and thus gives a qualitative image of the refractive index distribution.

Resolution enhancement in nonlinear photoacoustic imaging

Nonlinear processes can be exploited to gain access to more information than is possible in the linear regime. Nonlinearity modifies the spectra of the excitation signals through harmonic generation, frequency mixing, and spectral shifting, so that features originally outside the detector range can be detected. Here, we present an experimental study of resolution enhancement for photoacoustic imaging of thin metal layers immersed in water. In this case, there is a threshold in the excitation below which no acoustic signal is detected. Above threshold, the nonlinearity reduces the width of the active area of the excitation beam, resulting in a narrower absorption region and thus improved spatial resolution. This gain is limited only by noise, as the active area of the excitation can be arbitrarily reduced when the fluence becomes closer to the threshold. Here, we demonstrate a two-fold improvement in resolution and quantify the image quality as the excitation fluence goes through threshold.

Digital holographic confocal microscope

We demonstrate experimentally a scanning confocal microscopy technique based on digital holographic recording of the scanned spot. The data collected in this way contains all the information to produce three-dimensional images. Several methods to treat the data are presented, such as the dynamic placement of the pinhole. Examples of reflection and transmission images of epithelial cells and mouse brain tissue are shown. The computations can be performed in real time, the speed being limited only by the frame rate of the camera. This method enables a convenient implementation of confocal microscopy, especially in transmission as no de-scan device is required.

Versatile reconstruction framework for diffraction tomography with intensity measurements and multiple scattering

Taking benefit from recent advances in both phase retrieval and estimation of refractive indices from holographic measurements, we propose a unified framework to reconstruct them from intensity-only measurements. Our method relies on a generic and versatile formulation of the inverse problem and includes sparsity constraints. Its modularity enables the use of a variety of forward models, from simple linear ones to more sophisticated nonlinear ones, as well as various regularizers. We present reconstructions that deploy either the beam-propagation method or the iterative Lippmann-Schwinger model, combined with total-variation regularization. They suggest that our proposed (intensity-only) method can reach the same performance as reconstructions from holographic (complex) data. This is of particular interest from a practical point of view because it allows one to simplify the acquisition setup.

Imaging complex objects using learning tomography

Optical diffraction tomography (ODT) can be described using the scattering process through an inhomogeneous media. An inherent nonlinearity exists relating the scattering medium and the scattered field due to multiple scattering. Multiple scattering is often assumed to be negligible in weakly scattering media. This assumption becomes invalid as the sample gets more complex resulting in distorted image reconstructions. This issue becomes very critical when we image a complex sample. Multiple scattering can be simulated using the beam propagation method (BPM) as the forward model of ODT combined with an iterative reconstruction scheme. The iterative error reduction scheme and the multi-layer structure of BPM are similar to neural networks. Therefore we refer to our imaging method as learning tomography (LT). To fairly assess the performance of LT in imaging complex samples, we compared LT with the conventional iterative linear scheme using Mie theory which provides the ground truth. We also demonstrate the capacity of LT to image complex samples using experimental data of a biological cell.