Innv-29. Multispectral Analysis for Intra-Operative Characterization of Brain Tumours, Margins of Resection, and Eloquent Areas Activation – Preliminary Results

Abstract

BACKGROUND The identification of intrinsic brain tumours’ boundaries and their anatomical relationship with eloquent connectomes is one of the main challenges in neurosurgery. Several new intra-operative spectroscopic technologies have been recently investigated for this purpose. Multispectral imaging (MSI) analysis is based on analysis of reflected light, divided in specific wavelengths ranges across the electromagnetic spectrum. By doing so, it can identify wavelengths invisible to the naked eye, and make them visible by applying specific filters. METHODS We have used MSI to acquire intraoperative images during brain tumours surgeries. The MSI camera has been adapted to be connected to the intra-operative microscope and acquire a stack of images of different spectra to be analysed on a separate time. The focus of the images acquisition includes tumour structure, its margins, and eloquent connectomes during awake surgery or neurostimulation to analyse the possibility of different perfusion signal during activation. Image segmentation using K-mean clustering has been used to detect structural information and identify tumour’s area and tumour’s margins, while oxygen maps have been created using linear least squares regression to detect areas of metabolic activity. RESULTS Twenty-three patients have been enrolled in the current study at the time of this submission. We have analysed the intra-operative acquisition using intra-operative microscope picture, pre- and post-operative MRI scan, and neurophysiological findings as ground truth. Our preliminary results show that MSI is a potential additional technology to visually show tumour core, margins, and different degree of perfusions on the tissues analysed during tumour surgery. Although on a very limited subgroup of patients, MSI seems to be reliable in identify eloquent areas activation during awake surgery. These results are promising and warrant more data acquisition to be confirmed.

Lightfield hyperspectral imaging in neuro-oncology surgery: an IDEAL 0 and 1 study

“…This work has the potential to contribute toward the increasing literature relating to the specific and individual spectral signatures of brain tumors using HSI (Fabelo et al, 2018 , 2019 ; Anichini et al, 2022 ). This is crucial as brain tumor surgery still caries a 3% mortality (Williams et al, 2016 ) and a 3.4% average complication rate (De la Garza-Ramos et al, 2016 ), with an overall morbidity of up to 44.4% (Moiyadi and Shetty, 2012 ).…”

“…HSI has been shown to have multiple potential applications within the medical field (Spigulis, 2017 ; Leon et al, 2020 ; Torti et al, 2020 ) and the surgical field (Shapey et al, 2019 ). Specifically within the neurosurgical literature, its reported use spans from intra-operative tissue differentiation capabilities, particularly in glioma surgery (Fabelo et al, 2018 , 2019 ; Manni et al, 2020 ; Ebner et al, 2021 ; Anichini et al, 2022 ; Puustinen et al, 2023 ), to intra-operative visualization of cortical haemodynamic responses (Giannoni et al, 2018 ) including during epilepsy surgery (Pichette et al, 2016 ). There is also a growing body of work where HSI is used to try and quantify fluorescence (e.g., with 5-ALA) in brain tumor surgery (Valdes et al, 2019 ; Walke et al, 2023 ).…”

“…Despite evidence that these scanning approaches work in principle (Fabelo et al, 2018 , 2019 ), intra-operative HSI systems based on them have generally not translated well to the neuro-oncology surgical workflow outside of a research capacity. This is due to limitations such as image acquisition time and the spatial footprint of the system causing significant disruption in an already time constrained and confined neurosurgical theater environment (Shapey et al, 2019 ; Ebner et al, 2021 ), albeit some advances have been made for specific use cases (e.g., in epilepsy surgery) (Pichette et al, 2016 ; Anichini et al, 2022 ). Snapshot hyperspectral imaging systems capture hypercubes in one go and can allow for real-time imaging which makes them attractive for integration in surgical practice (Shapey et al, 2019 ).…”