A neuroimaging technique during surgery can further complement preoperative fMRI and electrical brain stimulation (EBS) to optimize the localization of eloquent areas during glioma resection and could improve dramatically the surgical procedure and patient care. Hyperspectral optical imaging is a non-invasive technique that is able to monitor hemodynamic and metabolic responses during neurosurgery. This technique can be used to complement the cortical functional mapping during neurosurgical procedures. However, a robust quantification of biomarkers of brain functionality is required to assist neurosurgeons. It is also essential to calibrate acquisition devices with robust optical phantoms to test instrument reliability. In this work, we explore the possibility to use a combined liquid blood phantom with cytochrome contained yeast to evaluate the reliability of hyperspectral imaging to measure oxygenation and metabolic changes. We also used hyperspectral imaging for identifying motor and sensory areas of human patients during neurosurgery. The results showed that a blood phantom and commercial yeast can be used to validate the measurement of hemodynamic and metabolic changes. This homogeneous phantom provides an excellent means to verify the reliability of intraoperative optical setups before moving on to clinical application. We showed that a commercial hyperspectral camera combined with a white light illumination could be used for identifying functional brain areas using hemodynamic and metabolic biomarkers. We also observed significant changes of the oxidative state of cytochrome-c-oxidase in periarterial tissue which seems to give an indication of the metabolism of the tissue during cerebral activity.

We present the HyperProbe consortium: a five-years, multinational, EU-funded project started in October, 2022, that aims at developing innovative hyperspectral imaging technologies for clinical translation. HyperProbe works towards providing highly enhanced neuronavigation during glioma resection.

Optical imaging is a marker-free, contactless, and non-invasive technique that is able to monitor hemodynamic and metabolic brain response following neuronal activation during neurosurgery. However, a robust quantification is complicated to perform during neurosurgery due to the critical context of the operating room, which makes the calibration and adjustment of optical devices more complex. To overcome this issue, tissue-simulating objects that mimic the properties of biological tissues are required for the development of detection or diagnostic imaging systems. In this study, we developed a digital instrument simulator to optimize the development of a novel hyperspectral system for application in brain/cortex imaging. This digital phantom is based on white Monte Carlo simulations of the light propagation in tissues. The output of the Monte Carlo simulations are integrated with the key instrument parameters in order to produce realistic images. The results can be beneficial and useful within the framework of our EU-funded HyperProbe project, which aims at transforming neuronavigation during glioma resection using novel hyperspectral imaging technology.

Spectral unmixing designates techniques that allow to decompose measured spectra into linear or non-linear combination of spectra of all targets (endmembers). This technique was initially developed for satellite applications, but it is now also widely used in biomedical applications. However, several drawbacks limit the use of these techniques with standard optical devices like RGB cameras. The devices need to be calibrated and a a priori on the observed scene is often necessary. We propose a new method for estimating endmembers and their proportion automatically and without calibration of the acquisition device based on near separable non-negative matrix factorization. This method estimates the endmembers on spectra of absorbance changes presenting periodic events. This is very common in in vivo biomedical and medical optical imaging where hemodynamics dominate the absorbance fluctuations. We applied the method for identifying functional brain areas during neurosurgery using four different RGB cameras (an industrial camera, a smartphone and two surgical microscopes). Results obtained with the auto-calibration method were consistent with the intraoperative gold standards. Endmembers estimated with the auto-calibration method were similar to the calibrated endmembers used in the modified Beer-Lambert law. The similarity was particularly strong when both cardiac and respiratory periodic events were considered. This work can allow a widespread use of spectral imaging in the industrial or medical field.

Significance:Diffuse optical modalities such as broadband near-infrared spectroscopy (bNIRS) and hyperspectral imaging (HSI) represent a promising alternative for low-cost, non-invasive, and fast monitoring of living tissue. Particularly, the possibility of extracting the molecular composition of the tissue from the optical spectra deems the spectroscopy techniques as a unique diagnostic tool. Aim:No established method exists to streamline the inference of the biochemical composition from the optical spectrum for real-time applications such as surgical monitoring. We analyze a machine learning technique for inference of changes in the molecular composition of brain tissue. Approach:We propose modifications to the existing learnable methodology based on the Beer-Lambert law. We evaluate the method's applicability to linear and nonlinear formulations of this physical law. The approach is tested on data obtained from the bNIRS- and HSI-based monitoring of brain tissue. Results:The results demonstrate that the proposed method enables real-time molecular composition inference while maintaining the accuracy of traditional methods. Preliminary findings show that Beer-Lambert law-based spectral unmixing allows contrasting brain anatomy semantics such as the vessel tree and tumor area. Conclusion:We present a data-driven technique for inferring molecular composition change from diffuse spectroscopy of brain tissue, potentially enabling intra-operative monitoring.

Histopathological examination of surgical biopsies, such as in glioma and glioblastoma resection, is hindered in current clinical practice by the long times required for the laboratory analysis and pathological screening, typically taking several days or even weeks to be completed. We propose here a transportable, high-density, spectral-scanning based hyperspectral imaging setup, named HyperProbe1, that can provide in situ, fast biochemical analysis and mapping of fresh surgical tissue samples, right after excision, and without the need of fixing or staining. HyperProbe1 is based on spectral scanning via supercontinuum laser illumination filtered with acousto-optic tuneable filters. Such methodology allows the user to select any number and type of wavelength bands in the visible and near-infrared range between 510 and 900 nm (up to 79), and to reconstruct 3D hypercubes composed of high-resolution, widefield images of the surgical samples, where each pixel is associated with a complete spectrum. The system is applied on 11 fresh surgical biopsies of glioma from routine patients, including different grades of tumour classification. Quantitative analysis of the composition of the tissue is performed via fast spectral unmixing to reconstruct mapping of major biomarkers. We also provided a preliminary attempt to infer tumour classification based on differences of composition in the samples, suggesting the possibility to use lipid content and differential cytochrome-c-oxidase concentrations to distinguish between lower and higher grade gliomas. A proof-of-concept of the performances of HyperProbe1 for quantitative, biochemical mapping of surgical biopsies is demonstrated, paving the way for improving current post-surgical, histopathological practice via non-destructive, in situ streamlined screening of fresh tissue samples in a matter of minutes after excision.

Diffuse low-grade gliomas are infiltrative tumors whose margins are not distinguishable from the adjacent healthy brain parenchyma. The aim was to precisely examine the results provided by the intraoperative use of macroscopic fluorescence in diffuse low-grade gliomas and to describe the new fluorescence-based techniques capable of guiding the resection of low-grade gliomas. Only about 20% and 50% of low-grade gliomas are macroscopically fluorescent after 5-amino-levulinic acid (5-ALA) or fluorescein sodium intake, respectively. However, 5-ALA is helpful for detecting anaplastic foci, and thus choosing the best biopsy targets in diffuse gliomas. Spectroscopic detection of 5-ALA-induced fluorescence can detect very low and non-macroscopically visible concentrations of protoporphyrin IX, a 5-ALA metabolite, and, consequently, has excellent performances for the detection of low-grade gliomas. Moreover, these tumors have a specific spectroscopic signature with two fluorescence emission peaks, which is useful for distinguishing them not only from healthy brain but also from high-grade gliomas. Confocal laser endomicroscopy can generate intraoperative optic biopsies, but its sensitivity remains limited. In the future, the coupled measurement of autofluorescence and induced fluorescence, and the introduction of fluorescence detection technologies providing a wider field of view could result in the development of operator-friendly tools implementable in the operative routine.

Optical imaging is a non-invasive technique that is able to monitor hemodynamic and metabolic brain response following neuronal activation during neurosurgery. However, it still lacks robustness to be used as a clinical standard. In particular, the quantification of the biomarkers of brain functionality needs to be improved. The quantification relies on the modified Be er Lambert law, which needs a correct estimation of the optical mean path length of traveled photons. Monte Carlo simulations are used for estimating the optical path length, but it is time-consuming, especially when modeling a patient's brain cortex. In this study, we developed a neural network based on the UNET architecture for a pixel-wise and real-time estimation of optical mean path length. The neural network was trained with segmentation of brain cortex as input and mean path length data as target. This deep learning approach allows a real time estimation of the optical mean path length. The results can be beneficial and useful within the framework of our EU-funded HyperProbe project, which a ims at transforming neuronavigation during glioma resection using novel hyperspectral imaging technology.

Optical imaging is a non-invasive technique that is able to monitor hemodynamic and metabolic responses during neurosurgery. However, a robust quantification is complicated to perform. To overcome this issue, phantoms that mimic biological tissues are required for the development of imaging systems in order to reach a true standardization. In this work, we explore the possibility to use a combined liquid blood phantom with cytochrome contained yeast to evaluate the reliability of hyperspectral imaging to measure oxygenation and metabolic changes. This phantom can be used to verify the reliability of intraoperative optical setups before moving on to clinical application.

In recent years, hyperspectral imaging (HSI) has demonstrated its capacity to non-invasively differentiate tumors from healthy tissues and identify cancerous regions during neurosurgery. Indeed, the spectral information contained in the HS images allows to identify more chromophores, refining the information provided by the imaging system, and allowing to identify the unique signature of each tissue types more accurately. Our HyperProbe project aims at developing a novel HSI system optimized for neurosurgery. As part of this project, we are developing a digital instrument simulator (DIS), based on Monte-Carlo (MC) simulations of the light propagation in tissues, in order to optimize both the hardware and data processing pipeline of our novel instrument. This framework allows us (1) to test the effect on the accuracy of the measurement of several hardware parameters, like the numerical aperture or sensitivity of the detector; (2) to be used as numerical phantoms to test various data processing algorithms; and (3) to generate generic data to develop and train machine learning (ML) algorithms. To do so, our framework is based on a 2-step method. Firstly, MC simulations are run to produce an ideal dataset of the photon transport in tissue. Then, the raw output parameters of the simulations, such as the exit positions and directions of the photons, are processed to take into account the physical parameters of an instrument in order to produce realistic images and test various scenarios. We present here the initial development of this DIS.

Complementary technique to preoperative fMRI and electrical brain stimulation (EBS) for glioma resection could improve dramatically the surgical procedure and patient care. Intraoperative RGB optical imaging is a technique for localizing functional areas of the human cerebral cortex that can be used during neurosurgical procedures. However, it still lacks robustness to be used with neurosurgical microscopes as a clinical standard. In particular, a robust quantification of biomarkers of brain functionality is needed to assist neurosurgeons. We propose a methodology to evaluate and optimize intraoperative identification of brain functional areas by RGB imaging. This consist in a numerical 3D brain model based on Monte Carlo simulations to evaluate intraoperative optical setups for identifying functional brain areas. We also adapted fMRI Statistical Parametric Mapping technique to identify functional brain areas in RGB videos acquired for 12 patients. Simulation and experimental results were consistent and showed that the intraoperative identification of functional brain areas is possible with RGB imaging using deoxygenated hemoglobin contrast. Optical functional identifications were consistent with those provided by EBS and preoperative fMRI. We also demonstrated that a halogen lighting may be particularity adapted for functional optical imaging. We showed that an RGB camera combined with a quantitative modeling of brain hemodynamics biomarkers can evaluate in a robust way the functional areas during neurosurgery and serve as a tool of choice to complement EBS and fMRI.

Recent advancements in imaging technologies (MRI, PET, CT, among others) have significantly improved clinical localisation of lesions of the central nervous system (CNS) before surgery, making possible for neurosurgeons to plan and navigate away from functional brain locations when removing tumours, such as gliomas. However, neuronavigation in the surgical management of brain tumours remains a significant challenge, due to the inability to maintain accurate spatial information of pathological and healthy locations intraoperatively. To answer this challenge, the HyperProbe consortium have been put together, consisting of a team of engineers, physicists, data scientists and neurosurgeons, to develop an innovative, all-optical, intraoperative imaging system based on (i) hyperspectral imaging (HSI) for rapid, multiwavelength spectral acquisition, and (ii) artificial intelligence (AI) for image reconstruction, morpho-chemical characterisation and molecular fingerprint recognition. Our HyperProbe system will (1) map, monitor and quantify biomolecules of interest in cerebral physiology; (2) be handheld, cost-effective and user-friendly; (3) apply AI-based methods for the reconstruction of the hyperspectral images, the analysis of the spatio-spectral data and the development and quantification of novel biomarkers for identification of glioma and differentiation from functional brain tissue. HyperProbe will be validated and optimised with studies in optical phantoms, in vivo against gold standard modalities in neuronavigational imaging, and finally we will provide proof of principle of its performances during routine brain tumour surgery on patients. HyperProbe aims at providing functional and structural information on biomarkers of interest that is currently missing during neuro-oncological interventions.

Separable spectral unmixing designates techniques that allow to decompose spectra as a linear or non-linear combination of spectra of the targets (endmembers) collected. These techniques allow quantitative measurements but several drawbacks limit its use with standard optical devices like RGB cameras. We propose a new method for estimating endmembers and their proportion without calibration of the acquisition device with the analysis of periodic events in the signal. We evaluated the performances of the method for identifying functional brain areas during neurosurgery using RGB imaging. Results were consistent with clinical gold standards. This work can allow a widespread use of spectral imaging in the industrial or medical field.

RGB imaging is a non-invasive technique that is able to monitor hemodynamic brain responses following neuronal activation during neurosurgery. These cameras are often present in operating rooms, but a robust quantification is complicated to perform during neurosurgery. Liquid blood have been proposed, but it is not possible to model hemodynamic responses similar to those that occur in the brain. To overcome this issue, we propose a 3D brain model, including activated, non-activated grey matter and temporal hemodynamic fluctuations using Monte Carlo simulations. Several setups were modeled to evaluate their impact for identifying activated brain areas using statistical parametric mapping.

The optical imaging described here is a marker-free, contactless, and non-invasive technique that is able to monitor hemodynamic brain response following neuronal activation during neurosurgery. However, a robust quantification is complicated to perform during neurosurgery due the critical context of the operating room, which makes the calibration and adjustment of optical devices more complex. To overcome this issue, tissue-simulating objects that mimic the properties of biological tissues are required for the development of detection or diagnostic imaging systems. In this study, we evaluated the performance of quantification of chromophore concentration changes measured by experimental setups using two phantoms: a liquid and a numeric brain-simulating phantom. These phantoms mimicked an exposed cerebral cortex as well as the slow concentration changes that occur after neuronal stimulation and the periodic changes due to heartbeat.

RGB optical imaging is a marker-free, contactless, and non-invasive technique that is able to monitor hemodynamic brain response following neuronal activation using task-based and resting-state procedures. As opposed to functional task-based analyses, resting-state functional connectivity aims to identify the low frequency cortical hemodynamic fluctuations during patient rest that are linked to resting-state networks. Using intraoperative optical imaging, the main issues of using resting-state procedures come from the partial access to the brain cortex, whereas fMRI or fNIRS resting-state models used whole brain imaging. Task-based fMRI brain maps were compared to intraoperative optical functional brain maps by registering these maps to a preoperative anatomical MRI volume. The objective is to improve the patient care process before, during and after neurosurgery. With the task-based procedure, the RGB brain map showed a good correspondence with task-based fMRI (DICE = 0.75). With the resting-state procedure, the RGB brain map showed a good correspondence with task-based fMRI (seed correlation method: DICE = 0.58 and ICA method: DICE = 0.75).

We present the methodology for the intraoperative pixel-wise identification of activated cortical areas using RGB imaging. The results indicate that RGB imaging could be a useful complement to the electrical brain stimulation.