We present two neural networks: one capable of processing a raw spectrum into an A-scan with the second-order nonlinearity removed and another for processing a raw spectrum into an A-scan with the third-order nonlinearity removed. An algorithm is also proposed to enable to use these networks in a sequence for removal of both nonlinearities. The presented approaches allow for either independent switching off of each order or the simultaneous removal of all orders, offering a tool for analysing the effects of each nonlinearity order individually or simply for performing all-depth, blind OCT data linearisation.
We present a neural network able to fully linearise an OCT image without any a priori knowledge about the spectrometer characteristics or the extent of dispersion in the interferometer and the object. Unlike the earlier solutions, this blind line-arisation is not biased towards a specific object, nor its dispersion characteristics, and in the future can be made independent of the light source parameters.
A neural network is proposed as a much better performing alternative to Fourier transformation. It processes raw OCT spectra into A-scans with twice better nominal axial resolution which remains intact at all depths even for an uncalibrated spectrometer and uncompensated chromatic dispersion.
The behavior of the signal elements in a quantum-mimic OCT is modelled to provide coefficients for simultaneous spectral calibration and dispersion compensation. The method allows to obtain the correction vectors based on a single spectrum and in an almost fully automatic way.
In Intensity Correlation Optical Coherence Tomography (ICA-OCT), an OCT spectrum is processed into a two-dimensional signal incorporating elements which do not correspond to the structure of the imaged object. These elements, called artefacts, display a very well-defined behaviour in the presence of uncompensated chromatic dispersion. More importantly, their behaviour reflects only the dispersion of the layer which the artefacts uniquely correspond to. We show preliminary results indicating that a neural network can interpret this layer-specific behaviour and output corresponding Group Velocity Dispersion values, thus creating a depth-resolved dispersion profile of the object.
Quantum Optical Coherence Tomography (Q-OCT) is a non-classical equivalent of Optical Coherence Tomog- raphy (OCT) able to provide an increased axial resolution and immunity to even orders of dispersion. The main drawback of Q-OCT is artefacts which are additional elements that clutter an A-scan and lead to a complete loss of structural information for multilayered objects. To retrieve the structure of the object, we propose to use machine learning in Fourier domain Q-OCT. We present preliminary results for a model based on VGG16 architecture and compare it to analytical algorithms. We show on computer-generated data that machine learning outperforms previously proposed analytical algorithms. The trained neural network requires much less input data and achieves much better results than the analytical algorithms. Finally, due to the way the training of the neural network is performed and the increased axial resolution of Q-OCT, machine learning provides super-resolution - it is able to distinguish two peaks which are otherwise unresolved in traditional OCT.
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.