SignificanceMueller matrix imaging (MMI) is a comprehensive form of polarization imaging useful for assessing structural changes. However, there is limited literature on the polarimetric properties of brain specimens, especially with multispectral analysis.AimWe aim to employ multispectral MMI for an exhaustive polarimetric analysis of brain structures, providing a reference dataset for future studies and enhancing the understanding of brain anatomy for clinicians and researchers.ApproachA multispectral wide-field MMI system was used to measure six fresh lamb brain specimens. Multiple decomposition methods (forward polar, symmetric, and differential) and polarization invariants (indices of polarimetric purity and anisotropy coefficients) have been calculated to obtain a complete polarimetric description of the samples. A total of 16 labels based on major brain structures, including grey matter (GM) and white matter (WM), were identified. K-nearest neighbors classification was used to distinguish between GM and WM and validate the feasibility of MMI for WM identification.ResultsAs the wavelength increases, both depolarization and retardance increase, suggesting enhanced tissue penetration into deeper layers. Moreover, utilizing multiple wavelengths allowed us to track dynamic shifts in the optical axis of retardance within the brain tissue, providing insights into morphological changes in WM beneath the cortical surface. The use of multispectral data for classification outperformed all results obtained with single-wavelength data and provided over 95% accuracy for the test dataset.ConclusionsThe consistency of these observations highlights the potential of multispectral wide-field MMI as a non-invasive and effective technique for investigating the brain’s architecture.
This study analyses intraoperative multispectral images taken from 47 brain tumour surgeries to investigate the diagnostic and surgical guidance potential of MSI. The research enrolled patients with various tumour types and introduces a hybrid model, uniting a transformer-coupled convolutional neural network (CNN), tailored for multispectral brain image segmentation. Leveraging MSI, the model was preliminarily assessed on ten meningioma and thirty-three glioma cases, each categorized into seven distinct classes. The model demonstrated a promising overall accuracies of 88.14% for meningioma and 85.64% for glioma. These initial results highlight the potential of the proposed hybrid architecture in multispectral brain image segmentation, laying the foundation for future research to optimize the model's performance with a larger patient cohort.
This study aims to integrate real-time hyperspectral (HS) imaging with a surgical microscope to assist neurosurgeons in differentiating between healthy and pathological tissue during procedures. Using the LEICA M525 microscope’s optical ports, we register HS data and RGB, in an efforts to improve margin delineation and surgical outcomes. The CUBERT ULTRIS SR5 camera with 51 bands and 15 Hz is employed, and critical calibration steps are outlined for clinical application. Experimental validation is conducted on ex-vivo animal tissue using reflectance spectroscopy. We present the preliminary validation results of the performance comparison between the designed hyperspectral imaging microscope prototype and diffuse reflectance spectroscopy conducted on animal tissue.
In this study, we applied Multispectral Mueller Matrix Imaging (MMI) at six distinct wavelengths in the visible range to analyze brain structures using lamb cerebral samples. The imaging of several brain sections revealed that white matter (WM) exhibits pronounced depolarization and retardance when contrasted with grey matter (GM), a phenomenon likely attributed to the elevated scattering and anisotropic nature of WM. More precisely, with an increase in wavelength, both depolarization and retardance also increase, suggesting additional penetration into deeper tissue layers. Employing various wavelengths enabled us to trace the shifts in the optical axis of retardance within the brain tissue, offering insights into the morphological changes in WM and GM below the cortical surface. The consistency observed in our results highlights the promise of Multispectral Wide-Field MMI as a non-intrusive, efficacious modality for probing brain architecture.
This research aims to deal with intraoperative multispectral images taken from brain tumour surgeries to investigate the diagnostic and guidance potential of MSI. These images were registered by feature-based (SIFT, PFN), intensity-based (LK) and machine learning (RANSAC-Flow) methods and classified via a CNN and Transformer model using anatomical labels. Based on the results from some initial training, MSI could achieve 95% overall accuracy. After labelling and registration are completed, a brain surgery dataset can be built to support intraoperative decision making.
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.