As the technological hurdles are overcome and optogenetic techniques advance to have more control over neurons, therapies based on these approaches will begin to emerge in the clinic. Here, we consider the technical challenges surrounding the transition of this breakthrough technology from an investigative tool to a true therapeutic avenue. The emerging strategies and remaining tasks surrounding genetically encoded molecules which respond to light as well as the vehicles required to deliver them are discussed.The use of optogenetics in humans would represent a completely new paradigm in medicine and would be associated with unprecedented technical considerations. To be applied for stimulation of neurons in humans, an ideal optogenetic tool would need to be non-immunogenic, highly sensitive, and activatable with red light or near-infrared light (to maximize light penetration while minimizing photodamage). To enable sophisticated levels of neuronal control, the combined use of optogenetic actuators and indicators could enable closed-loop all-optical neuromodulation. Such systems would introduce additional challenges related to spectral orthogonality between actuator and indicator, the need for decision making computational algorithms and requirements for large gene cassettes. As in any gene therapy, the therapeutic efficiency of optogenetics will rely on vector delivery and expression in the appropriate cell type. Although viral vectors such as those based on AAVs are showing great potential in human trials, barriers to their general use remain, including immune responses, delivery/transport, and liver clearance. Limitations associated with the gene cassette size which can be packaged in currently approved vectors also need to be addressed.
Purpose : Conventional processing of OCT spectrometer data is comprised of isolation of the fringe signal, resampling from frequency to wave space, dispersion compensation, and Fourier transformation. Resulting A and B-scans are fraught with speckle noise, requiring further processing to become clinically usable. We propose using machine learning (ML) digital signal processing (DSP) as an alternative direct pathway to produce OCT B-scans.
Methods : We applied a 1D neural network autoencoder built with the Keras Python library running on top of TensorFlow to process raw spectrometer data into single A-scans. Two ML algorithms were trained using entirely unprocessed spectra; one to post-processed and the second to averaged A-scans. The training A-scans were a combination of spectra obtained from a layered phantom, and spectra from a retinal imagery (Bioptigen OCT). In total, 15 different image volumes were used. Four were reserved for testing and validation. The 11 training volumes provided 1.5 million spectra lines and their corresponding processed counterparts for training. From this, 750,000 were randomly selected on each epoch for training. Equal distribution across all volumes reduced the probability of overfitting.
Results : Assessed subjectively, when trained to the averaged dataset, ML-DSP produced B-scans directly from raw spectrometer data with less speckle noise (Figure 1: Phantom, Figure 2: Retina). A-scan to A-scan alignment within the assembled B-Scan is reduced in the phantom image, and poor in the retinal image.
Conclusions : We have demonstrated that ML-DSP is capable of processing raw spectrometer data directly to produce B-scans. Future work will address A-scan alignment by expanding to a 2D ML-DSP engine, processing entire B-scans instead of individual A-scans. Future ML engines may be enhanced to provide measurements, segmentation, angiography, Doppler, and even diagnosis directly from raw spectral data; bypassing numerous conventional processing steps. The successful implementation of this technique could be a substantial advance in OCT technology.
Significance: Although the clinical potential for Raman spectroscopy (RS) has been anticipated for decades, it has only recently been used in neurosurgery. Still, few devices have succeeded in making their way into the operating room. With recent technological advancements, however, vibrational sensing is poised to be a revolutionary tool for neurosurgeons.
Aim: We give a summary of neurosurgical workflows and key translational milestones of RS in clinical use and provide the optics and data science background required to implement such devices.
Approach: We performed an extensive review of the literature, with a specific emphasis on research that aims to build Raman systems suited for a neurosurgical setting.
Results: The main translatable interest in Raman sensing rests in its capacity to yield label-free molecular information from tissue intraoperatively. Systems that have proven usable in the clinical setting are ergonomic, have a short integration time, and can acquire high-quality signal even in suboptimal conditions. Moreover, because of the complex microenvironment of brain tissue, data analysis is now recognized as a critical step in achieving high performance Raman-based sensing.
Conclusions: The next generation of Raman-based devices are making their way into operating rooms and their clinical translation requires close collaboration between physicians, engineers, and data scientists.
Optogenetics has become an integral tool for studying and dissecting the neural circuitries of the brain using optical control. Recently, it has also begun to be used in the investigation of the spinal cord and peripheral nervous system. However, information on these regions’ optical properties is sparse. Moreover, there is a lack of data on the dependence of light propagation with respect to neural tissue organization and orientation. This information is important for effective simulations and optogenetic planning, particularly in the spinal cord where the myelinated axons are highly organized. To this end, we report experimental measurements for the scattering coefficient, validated with three different methods in both the longitudinal and radial directions of multiple mammalian spinal cords. In our analysis, we find that there is indeed a directional dependence of photon propagation when interacting with organized myelinated axons. Specifically, light propagating perpendicular to myelinated axons in the white matter of the spinal cord produced a measured reduced scattering coefficient (μs′) of 3.52 ± 0.1 mm − 1, and light that was propagated along the myelinated axons in the white matter produced a measured μs′ of 1.57 ± 0.03 mm − 1, across the various species considered. This 50% decrease in scattering power along the myelinated axons is observed with three different measurement strategies (integrating spheres, observed transmittance, and punch-through method). Furthermore, this directional dependence in scattering power and overall light attenuation did not occur in the gray matter regions where the myelin organization is nearly random. The acquired information will be integral in preparing future light-transport simulations and in overall optogenetic planning in both the spinal cord and the brain.
Retinal oximetry is a non‐invasive imaging technology that enables the measurement of oxygen saturation (StO2) in the eye fundus. The goal of this research was to validate a convolutional neural network (CNN) algorithm designed to calculate and improve the precision of StO2 measurements from diffuse reflectance spectra (DRS) taken on the optic nerve head (ONH). Zilia’s multi‐wavelength retinal oximetry device was used to acquire experimental spectra from the ONH which allowed us to simulate reasonable digital spectra replicates with known absorber concentrations. These spectra were used to train a novel machine learning algorithm based on CNN. The device was then used to acquire diffuse reflectance spectra on several ONH‐mimicking liquid optical phantoms (phantom eye) with dynamic oxygenation cycles between 0% ‐ 100% in order to validate the improvements of this CNN on experimental data. Measurements were made simultaneously with gold standard devices for comparison. The procedure was then repeated with several cataract‐simulating contact lenses integrated in the optical path to show the robustness of oximetry measurements. We found good agreement in StO2 measurements between the results obtained with the Zilia device using the CNN algorithm and the gold standard references in all phantom cases. Applying the various algorithms to data acquired on the validation phantom show the marked improvements in using the CNN on experimental data, validating its high potential for clinical use. We specifically show strong robustness in precision, even when cataract‐simulating lenses were used. We present further validation that the oximetry device and CNN algorithm produces reliable, precise measurements even under conditions where blood volume fractions vary, optical scattering changes, and cataract‐simulating contact lenses are included.
Retinal oximetry is a non-invasive technique to investigate the hemodynamics, vasculature and health of the eye. Current techniques for retinal oximetry have been plagued by quantitatively inconsistent measurements and this has greatly limited their adoption in clinical environments. To become clinically relevant oximetry measurements must become reliable and reproducible across studies and locations. To this end, we have developed a convolutional neural network algorithm for multi-wavelength oximetry, showing a greatly improved calculation performance in comparison to previously reported techniques. The algorithm is calibration free, performs sensing of the four main hemoglobin conformations with no prior knowledge of their characteristic absorption spectra and, due to the convolution-based calculation, is invariable to spectral shifting. We show, herein, the dramatic performance improvements in using this algorithm to deduce effective oxygenation (SO2), as well as the added functionality to accurately measure fractional oxygenation (). Furthermore, this report compares, for the first time, the relative performance of several previously reported multi-wavelength oximetry algorithms in the face of controlled spectral variations. The improved ability of the algorithm to accurately and independently measure hemoglobin concentrations offers a high potential tool for disease diagnosis and monitoring when applied to retinal spectroscopy.
Purpose : Retinal oximetry is a non-invasive imaging technology that enables the measurement of oxygen saturation (SO2) in the eye fundus. The goal of this research was to validate a convolutional neural network (CNN) algorithm designed to calculate and improve the precision of SO2 measurements from diffuse reflectance spectra (DRS) taken on the optic nerve head (ONH).
Methods : The ocular oximetry device developed by Zilia was used to acquire diffuse reflectance spectra (DRS) on several ONH-mimicking liquid optical phantoms (phantom eye). The oxygenation of the blood circulating in the phantom eyes was dynamically cycled from 100% to 0% oxygenation (using yeast and oxygen gas). SO2 measurements were made simultaneously with Zilia’s device and gold standard devices for comparison throughout the experiments. The phantom eyes were made to assess variations in blood volumes and scattering coefficients, corresponding to typical ranges observed for ONH optical properties. The procedure was then repeated with several cataract-simulating contact lenses integrated to the optical path to show robustness of oximetry measurements. Finally, we apply the algorithm to spectra acquired in vivo on subjects at baseline and hyperoxic conditions.
Results : We found good agreement in SO2 measurements between the results obtain with the Zilia device using the CNN algorithm and the gold standard references in all phantom eyes. We specifically show strong robustness in precision, even when all 3 of the experimental cataract-simulating lenses were used. This is significant since cataracts has traditionally plagued oximetry measurements. Lastly, we show that ocular oximetry measurements showed reliable increases in in vivo oxygenation, consistent with results from tissue oximetry, in test-subjects provided with 100% oxygen.
Conclusions : We present, here, further validation that the oximetry device and CNN algorithm developed to measure SO2 in the eye fundus produces reliable, precise measurements even under conditions where blood volume fractions vary, optical scattering changes, and cataract-simulating contact lenses are included. Furthermore, we show that the algorithm can be applied to in vivo measurements.
Objective: The clinical outcome of deep brain stimulation (DBS) surgery relies heavily on the implantation accuracy of a chronic stimulating electrode into a small target brain region. Most techniques that have been proposed to precisely target these deep brain regions were designed to map intracerebral electrode trajectory prior to chronic electrode placement, sometimes leading to positioning error of the final electrode. This study was designed to create a new intraoperative guidance tool for DBS neurosurgery that can improve target detection during the final implantation of the chronic electrode.
Methods: Taking advantage of diffuse reflectance spectroscopy, the authors developed a new surgical tool that senses proximal brain tissue through the tip of the chronic electrode by means of a novel stylet, which provides rigidity to DBS leads and houses fiber optics.
Results: As a proof of concept, the authors demonstrated the ability of their noninvasive optical guidance technique to precisely locate the border of the subthalamic nucleus during the implantation of commercially available DBS electrodes in anesthetized parkinsonian monkeys. Innovative optical recordings combined to standard microelectrode mapping and detailed postmortem brain examination allowed the authors to confirm the precision of optical target detection. They also show the optical technique’s ability to detect, in real time, upcoming blood vessels, reducing the risk of hemorrhage during the chronic lead implantation.
Conclusions: The authors present a new optical guidance technique that can detect target brain regions during DBS surgery from within the implanted electrode using a proof of concept in nonhuman primates. The technique discriminates tissue in real time, contributes no additional invasiveness to the procedure by being housed within the electrode, and can provide complementary information to microelectrode mapping during the implantation of the chronic electrode. The technique may also be a powerful tool for providing direct anatomical information in the case of direct implantations wherein microelectrode mapping is not performed.
Microscopy methods used to measure Förster resonance energy transfer (FRET) between fluorescently labeled proteins can provide information on protein interactions in cells. However, these methods are diffraction-limited, thus do not enable the resolution of the nanodomains in which such interactions occur in cells. To overcome this limitation, we assess FRET with an imaging system combining fluorescence lifetime imaging microscopy with stimulated emission depletion, termed fluorescence lifetime imaging nanoscopy (FLIN). The resulting FRET-FLIN approach utilizes immunolabeling of proteins in fixed cultured neurons. We demonstrate the capacity to discriminate nanoclusters of synaptic proteins exhibiting variable degrees of interactions with labeled binding partners inside dendritic spines of hippocampal neurons. This method enables the investigation of FRET within nanodomains of cells, approaching the scale of molecular signaling.
Coherent Raman fiber probes have not yet found their way into the clinic despite their immense potential for label-free sensing and imaging. This is mainly due to the traditional bulky laser systems required to create the high peak power laser pulses needed for coherent Raman, as well as the complications that arise from the propagation of this type of energy through silica. Specifically, a coherent anti-Stokes Raman scattering (CARS) probe that could select its integration volume at high resolution, away from the tip of the fiber, is particularly interesting in the case of electrode implantation neurosurgeries, wherein it is possible to place optical fibers on-board the chronic electrode and provide optical guidance during its implantation, through the semi-transparent tip. To this clinical end, we have created an all fiber CARS system, consisting of small, rapidly tunable, turn-key fiber-lasers, capable of creating high wavenumber CARS spectra on the order of tens-of-milliseconds. The use of traditional silica fibers is made possible by the use of the laser’s long pulse-widths (25 ps). The probe itself has an outer diameter of 250 μm allowing it to fit within commercially available metal tubes that can replace deep brain stimulation (DBS) stylets. Using this system, we identified brain tissue types in intact nonhuman primates’ brains and showed the ability to delineate white and gray matters with high resolution. Its advantages over spontaneous Raman stem from the orders of magnitude improvement in spatial resolution, its inherent translatability to three-dimensional (3-D) imaging, as well as the theoretical ability to remove parasitic Raman signal from probe encasements, such as a DBS electrode. The system is planned to have clinical implications in neurosurgical guidance as well as diseased tissue detection.
Primary afferents transduce environmental stimuli into electrical activity that is transmitted centrally to be decoded into corresponding sensations. However, it remains unknown how afferent populations encode different somatosensory inputs. To address this, we performed two-photon Ca2+ imaging from thousands of dorsal root ganglion (DRG) neurons in anesthetized mice while applying mechanical and thermal stimuli to hind paws. We found that approximately half of all neurons are polymodal and that heat and cold are encoded very differently. As temperature increases, more heating-sensitive neurons are activated, and most individual neurons respond more strongly, consistent with graded coding at population and single-neuron levels, respectively. In contrast, most cooling-sensitive neurons respond in an ungraded fashion, inconsistent with graded coding and suggesting combinatorial coding, based on which neurons are co-activated. Although individual neurons may respond to multiple stimuli, our results show that different stimuli activate distinct combinations of diversely tuned neurons, enabling rich population-level coding.
We introduce the feature issue on the Optics in the Life Sciences Congress held on April 2–5, 2017 in San Diego, CA. The Congress consisted of 5 topical symposia: (i) Bio-optics Design and Application; (ii) Novel Techniques in Microscopy; (iii) Optical Molecular Probes, Imaging and Drug Delivery; (iv) Optical Trapping Applications; and (v) Optics and the Brain. These separate symposia also held joint sessions of common interest. The following highlights some of the topics from the Congress.
Conventional two-photon microscopy (TPM) is capable of imaging neural dynamics with subcellular resolution, but it is limited to a field-of-view (FOV) diameter <1 mm. Although there has been recent progress in extending the FOV in TPM, a principled design approach for developing large FOV TPM (LF-TPM) with off-the-shelf components has yet to be established. Therefore, we present a design strategy that depends on analyzing the optical invariant of commercially available objectives, relay lenses, mirror scanners, and emission collection systems in isolation. Components are then selected to maximize the space-bandwidth product of the integrated microscope. In comparison with other LF-TPM systems, our strategy simplifies the sequence of design decisions and is applicable to extending the FOV in any microscope with an optical relay. The microscope we constructed with this design approach can image <1.7-μm lateral and <28-μm axial resolution over a 7-mm diameter FOV, which is a 100-fold increase in FOV compared with conventional TPM. As a demonstration of the potential that LF-TPM has on understanding the microarchitecture of the mouse brain across interhemispheric regions, we performed in vivo imaging of both the cerebral vasculature and microglia cell bodies over the mouse cortex.
Granule cells (GCs) in the olfactory bulb (OB) play an important role in odor information processing. Although they have been classified into various neurochemical subtypes, the functional roles of these subtypes remain unknown. We used in vivo two-photon Ca2+ imaging combined with cell-type-specific identification of GCs in the mouse OB to examine whether functionally distinct GC subtypes exist in the bulbar network. We showed that half of GCs express Ca2+/calmodulin-dependent protein kinase IIα (CaMKIIα+) and that these neurons are preferentially activated by olfactory stimulation. The higher activity of CaMKIIα+ neurons is due to the weaker inhibitory input that they receive compared to their CaMKIIα-immunonegative (CaMKIIα−) counterparts. In line with these functional data, immunohistochemical analyses showed that 75%–90% of GCs expressing the immediate early gene cFos are CaMKIIα+ in naive animals and in mice that have been exposed to a novel odor and go/no-go operant conditioning, or that have been subjected to long-term associative memory and spontaneous habituation/dishabituation odor discrimination tasks. On the other hand, a perceptual learning task resulted in increased activation of CaMKIIα− cells. Pharmacogenetic inhibition of CaMKIIα+ GCs revealed that this subtype is involved in habituation/dishabituation and go/no-go odor discrimination, but not in perceptual learning. In contrast, pharmacogenetic inhibition of GCs in a subtype-independent manner affected perceptual learning. Our results indicate that functionally distinct populations of GCs exist in the OB and that they play distinct roles during different odor tasks.
Mesodiencephalic dopamine neurons play central roles in the regulation of a wide range of brain functions, including voluntary movement and behavioral processes. These functions are served by distinct subtypes of mesodiencephalic dopamine neurons located in the substantia nigra pars compacta and the ventral tegmental area, which form the nigrostriatal, mesolimbic, and mesocortical pathways. Until now, mechanisms involved in dopaminergic circuit formation remained largely unknown. Here, we show that Lmx1a, Lmx1b, and Otx2 transcription factors control subtype-specific mesodiencephalic dopamine neurons and their appropriate axon innervation. Our results revealed that the expression of Plxnc1, an axon guidance receptor, is repressed by Lmx1a/b and enhanced by Otx2. We also found that Sema7a/Plxnc1 interactions are responsible for the segregation of nigrostriatal and mesolimbic dopaminergic pathways. These findings identify Lmx1a/b, Otx2, and Plxnc1 as determinants of dopaminergic circuit formation and should assist in engineering mesodiencephalic dopamine neurons capable of regenerating appropriate connections for cell therapy.
The complete list of publications can also be found on Google Scholar.