In recent years, it was demonstrated that discrimination between white matter and tumor-infiltrated white matter based on optical coherence tomography (OCT) data is possible with high accuracy. However, gray matter is also present during the tumor resection and shows similar optical properties to tumor infiltration, which aggravates the tumor classification using optical coherence tomography. A semantic segmentation approach based on a convolutional neural network was applied to the problem in order to classify healthy brain tissue from tumor infiltrated brain tissue. A dataset was created, which consisted of ex vivo OCT B-scans, which were acquired by a swept-source OCT system with a central wavelength of 1300 nm. Each OCT B-scan was indirectly annotated by transforming histological labels from a corresponding H&E section onto it. The labels differentiate between white matter, gray matter and tumor infiltration. The output of the network was modeled to a Dirichlet prior distribution, which enabled the capturing of a prediction uncertainty. This approach achieved an intersection over union score of 0.72 for healthy brain tissue and 0.69 for highly tumor infiltrated brain tissue, when only confident predictions were considered.
Small defects in dental crowns due to the manufacturing or physiological/non-physiological stress in patients’ mouth can result in failure of restorations and sometimes even lead to the extraction of the tooth itself. If those microcracks can reliably be detected, defective crowns can be sorted out or be repaired in case of fatigue damage. Currently there is no non-destructive method to visualize microcracks within Computer-Aided Design/Computer-Aided Manufacturing (CAD/CAM)-materials, neither for dentists nor manufacturers. Therefore, eight different monolithic CAD/CAM-materials were inspected by different Optical Coherence Tomography (OCT) systems before and after mouth-motion-simulation. All OCT devices were suitable to visualize microcracks in the crowns. However, differences regarding parameters and CAD/CAM-materials were detected.
The ill-defined tumor borders of glioblastoma multiforme pose a major challenge for the surgeon during tumor resection, since the goal of the tumor resection is the complete removal, while saving as much healthy brain tissue as possible. In recent years, optical coherence tomography (OCT) was successfully used to classify white matter from tumor infiltrated white matter by several research groups. Motivated by these results, a dataset was created, which consisted of sets of corresponding ex vivo OCT images, which were acquired by two OCT-systems with different properties (e.g. wavelength and resolution). Each image was annotated with semantic labels. The labels differentiate between white and gray matter and three different stages of tumor infiltration. The data from both systems not only allowed a comparison of the ability of a system to identify the different tissue types present during the tumor resection, but also enable a multimodal tissue analysis evaluating corresponding OCT images of the two systems simultaneously. A convolutional neural network with dirichlet prior was trained, which allowed to capture the uncertainty of a prediction. The approach increased the sensitivity of identifying tumor infiltration from 58 % to 78 % for data with a low prediction uncertainty compared to a previous monomodal approach.
The identification of ex vivo brain tumor tissue was investigated with two different optical coherence tomography systems exploiting two optical parameters. The optical parameters were calculated from semantically labelled OCT B-scans.
We demonstrate that the coherence roll-off and dynamic range of OCT systems using Fourier-domain mode-locked (FDML) lasers can be significantly improved by a high-finesse fiber Fabry-Perot tunable filter (FFP-TF). The newly developed high-finesse FFP-TFs have a finesse of more than 3000, a more than fivefold improvement over previous designs. We show that this results in reduced instantaneous laser linewidth and reduced noise for a 1310 nm FDML laser with 1.6 MHz sweep rate. Since in practice, OCT image range is limited by data acquisition bandwidth, we demonstrate OCT imaging over many centimeters by reducing the sweep range of the laser. In contrast to previous work, standard resampling using a pre-acquired signal (as in SD-OCT) with no k-clocking is sufficient for both small and large sweep range, significantly reducing the system complexity. Live 3D-OCT video rate imaging at 20 cm imaging range is demonstrated.
The separation of tumorous brain tissue and healthy brain tissue is still a big challenge in the field of neurosurgery, especially when it comes to the detection of different infiltration grades of glioblastoma multiforme at the tumor border. On the basis of a recently created labelled OCT dataset of ex vivo glioblastoma multiforme tumor samples the detection of brain tumor tissue and the identification of zones with varying degrees of infiltration of tumor cells was investigated. The identification was based on the optical properties, which were extracted by an exponential fit function. The results showed that a separation of tumorous tissue and healthy white matter based on these optical properties is possible. A support vector machine was trained on the optical properties to separate tumor from healthy white matter tissue, which achieved a sensitivity of 91% and a specificity of 76% on an independent training dataset.
Optical coherence tomography (OCT) has the potential to become an additional imaging modality for surgical guidance in the field of neurosurgery, especially when it comes to the detection of different infiltration grades of glioblastoma multiforme at the tumor border. Interpretation of the images, however, is still a big challenge. A method to create a labeled OCT dataset based on ex vivo brain samples is introduced. The tissue samples were embedded in an agarose mold giving them a distinctive shape before images were acquired with two OCT systems (spectral domain (SD) and swept source (SS) OCT) and histological sections were created and segmented by a neuropathologist. Based on the given shape, the corresponding OCT images for each histological image can be determined. The transfer of the labels from the histological images onto the OCT images was done with a non-affine image registration approach based on the tissue shape. It was demonstrated that finding OCT images of a tissue sample corresponding to segmented histological images without any color or laser marking is possible. It was also shown that the set labels can be transferred onto OCT images. The accuracy of method is 26 ± 11 pixel, which translates to 192 ± 75 μm for the SS-OCT and 94 ± 43 μm for the SD-OCT. The dataset consists of several hundred labeled OCT images, which can be used to train a classification algorithm.
Fourier domain mode locked (FDML) laser are fast swept light sources. Measuring the linewidth and coherence length of such light sources is not straightforward, but very important for a physical understanding of FDML lasers and their performance in optical coherence tomography (OCT). In order to characterize the dynamic (“instantaneous”) linewidth, we performed beat signal measurements between a stationary narrowband continuous wave laser and an FDML laser and detected the signals with a 63 GHz real time oscilloscope. The evaluation of the beat signals of consecutive FDML wavelength sweeps yields information about the phase evolution within one sweep and over several sweeps. These measurements suggest the existence of a distinct comb like mode structure of the FDML laser and help to determine the locking strength of individual modes (comb lines).
In Fourier domain mode locked (FDML) lasers, extremely precise and stable matching of the filter tuning period and light circulation time in the cavity is essential for ultra-low noise operation. During the operation of FDML lasers, the ultra-low noise mode can be lost due to temperature drifts of the already temperature stabilized cavity resulting in increased intensity noise. Until now, the filter frequency is continuously regulated to match the changing light circulation time. However, this causes the filter frequency to constantly change by a few mHz and leads to synchronization issues in cases where a fixed filter frequency is desired. We present an actively cavity length controlled FDML laser and a robust and high precision feedback loop algorithm for maintaining ultra-low noise operation. Instead of adapting the filter frequency, the cavity length is adjusted by a motorized free space beam path to match the fixed filter frequency. The closed-loop system achieves a stability of ~0.18 mHz at a sweep repetition rate of ~418 kHz which corresponds to a ratio of 4×10-10. We investigate the coherence properties during the active cavity length adjustments and observe no noise increase compared to fixed cavity length. The cavity length control is fully functional and for the first time, offers the possibility to operate an FDML laser in sweet spot mode at a fixed frequency or phase locked to an external clock. This opens new possibilities for system integration of FDML lasers.
Fourier domain mode locking (FDML) is a recently developed technique for lasers to generate ultra-rapid wavelength sweeps, equivalent to a train of extremely chirped pulses. FDML lasers are the light sources of choice for fastest megahertz optical coherence tomography (MHz-OCT). Measuring the coherence properties of FDML lasers is of particular importance for the image quality in OCT but it is also crucial to develop a better understanding of this unconventional mode locking mechanism. Usually, experiments to analyze the phase stability of FDML lasers use interferometers to generate interference of a single laser by delaying a part of the output to generate a beat signal. Here, for the first time, we present real time beat signal measurements between two independent FDML lasers over the entire sweep range of ~5 THz width for more than 80 roundtrips (~200 μs), evaluate their phase stability and explain the consequence for our understanding of the FDML mechanism. Beat signal measurements allow direct access to the phase difference between the FDML lasers and therefore the difference in timing of the circulating sweeps as well as their instantaneous frequency.
The aim of this work is the creation of segmented data set consisting of optical coherence tomography (OCT) scans, which were taken of brain tumor tissue with different tumor infiltration rates. In an ongoing clinical study more than 140 human brain samples with different infiltration grades were recorded ex vivo with two OCT systems, a spectral domain OCT system and a swept-source OCT system that uses a 1310 nm Fourier domain mode locked laser. The histological analysis of the recorded samples builds the ground truth for labeling the corresponding OCT B-Scans. The segmented data set gained from this process will be used to train a classification algorithm, taking into account structural and optical properties such as the attenuation coefficient. In the future the classification algorithm together with a microscope integrated OCT system will be used for the in vivo identification of brain tumors as a guidance tool for the surgeon to increase tumor resection efficiency.
Optical coherence tomography (OCT) applications like ultra-widefield and full eye-length imaging are of high interest for various diagnostic purposes. In swept-source OCT these techniques require a swept light source, which is coherent over the whole imaging depth. We present a zero roll-off 1060 nm Fourier Domain Mode Locked-Laser (FDML-Laser) for retinal OCT imaging at 1.7 MHz A-scan rate and first long-range imaging results with it. Several steps such as improved dispersion compensation and frequency regulation were performed and will be discussed. Besides virtually no loss in OCT signal over the maximum depth range of 4.6 mm and very good dynamic range was observed. Roll-off measurements show no decrease of the point-spread function (PSF), while maintaining a high dynamic range.
We implemented a real-time video-rate 4D-OCT system with virtual reality display. To achieve the required low latencies we optimized the dataflow path and the placement of the necessary synchronization points. Employing temporal reprojection enables to perform volume rendering at 1/3 of the display refresh rate, yet maintaining smooth updates to the HMD; thus we achieve display updates at 90Hz, volume rendering at 30Hz and C-scan acquisition at <15Hz. By mounting of a tracking accessory to the scanning head we can render the OCT volume in virtual space in the position of the actual imaging volume.
MHz-OCT systems based on FDML swept laser sources combined with the massive parallel processing capabilities of modern computer hardware enable volumetric imaging, processing and stereoscopic display at video rates. The increasing image quality and speed might enable new fields of application where the volumetric OCT completely replaces stereoscopic microscopes instead of being a mere supplement. Aside from the depth resolving capability, a particular advantage is the ability to display a whole image volume from arbitrary points of view without the need to move the actual microscope or to rotate the patient’s eye. Purely digital microscopy is already offered as alternative to traditional through-an-eyepiece surgical microscope. We explore the use of virtual reality to present digital OCT microscopy images to a trained surgeon, carrying out a series of surgical procedures ex-vivo on a porcine eye model.
Optical coherence tomography (OCT) is a non-invasive imaging technique which is currently investigated for intraoperative detection of residual tumor during resection of human gliomas. Three different OCT systems were used for imaging of human glioblastoma in vivo (830nm spectral domain (SD) OCT integrated into a surgical microscope) and ex vivo (940nm SD-OCT and 1310nm swept-source MHz-OCT using a Fourier domain mode locked (FDML) laser). Before clinical data acquisition, the systems were characterized using a three-dimensional point-spread function phantom. To distinguish tumor from healthy brain tissue later on, attenuation coefficients of each pixel in OCT depth profiles are calculated. First examples from a clinical study show that the pixel-resolved calculation of the attenuation coefficient provides a good image contrast and confirm that white matter shows a higher signal and more homogeneous signal structure than tumorous tissue.
We demonstrate that the 3.2 MHz a-scan rate and the improved coherence of our new low noise FDML laser enables live 3D-OCT with different spectral zooms and up to 10 cm of imaging range.
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.