Removal of intrinsic brain tumors is a delicate process, where a high degree of specificity is required to remove all of the tumor tissue without damaging healthy brain. The accuracy of this process can be greatly enhanced by intraoperative guidance. Optical biopsies using Raman spectroscopy are a minimally invasive and lower-cost alternative to current guidance methods. A miniature Raman probe for performing optical biopsies of human brain tissue is presented. The probe allows sampling inside a conventional stereotactic brain biopsy system: a needle of length 200 mm and inner diameter of 1.8 mm. By employing a miniature stand-off Raman design, the probe removes the need for any additional components to be inserted into the brain. Additionally, the probe achieves a very low internal silica background while maintaining good collection of Raman signal. To illustrate this, the probe is compared with a Raman probe that uses a pair of optical fibers for collection. The miniature stand-off Raman probe is shown to collect a comparable number of Raman scattered photons, but the Raman signal to background ratio is improved by a factor of five at Raman shifts below ∼500 cm−1. The probe’s suitability for use on tissue is demonstrated by discriminating between different types of healthy porcine brain tissue.
Removal of intrinsic brain tumours is a delicate process, where a high degree of specificity is required to remove all of the tumour tissue without damaging healthy brain. The accuracy of this process can be greatly enhanced by intraoperative guidance. Optical biopsies using Raman spectroscopy are a minimally invasive and lower cost alternative to current guidance methods. A miniature Raman probe for performing optical biopsies of human brain tissue is presented. The probe allows sampling inside a conventional stereotactic brain biopsy system: a needle of length 200mm and inner diameter of 1.8mm. The probe achieves a very low fluorescent background whilst maintaining good collection of Raman signal by employing a miniature stand-off Raman design. To illustrate this, the probe is compared with a Raman probe that uses a pair of optical fibres for collection. The miniature stand-off Raman probe is shown to collect a comparable number of Raman scattered photons, but the fluorescence caused by silica fibres in a Raman needle probe is reduced by a factor of two for Raman shifts under 500 cm-1, and by 30% at 600-700 cm-1. In addition, this design contains only medically approved materials at the distal end. The probe’s suitability for use on tissue is demonstrated by discriminating between different types of porcine brain tissue.
We evaluate the potential of a custom-built fiber-optic Raman probe, suitable for in vivo use, to differentiate between benign, metaplastic (Barrett's oesophagus), and neoplastic (dysplastic and malignant) oesophageal tissue ex vivo on short timescales. We measured 337 Raman spectra (λ ex =830 nm ; P ex =60 mW ; t=1 s ) using a confocal probe from fresh (298) and snap-frozen (39) oesophageal tissue collected during surgery or endoscopy from 28 patients. Spectra were correlated with histopathology and used to construct a multivariate classification model which was tested using leave one tissue site out cross-validation in order to evaluate the diagnostic accuracy of the probe system. The Raman probe system was able to differentiate, when tested with leave one site out cross-validation, between normal squamous oesophagus, Barrett's oesophagus and neoplasia with sensitivities of (838% to 6%) and specificities of (89% to 99%). Analysis of a two group model to differentiate Barrett's oesophagus and neoplasia demonstrated a sensitivity of 88% and a specificity of 87% for classification of neoplastic disease. This fiber-optic Raman system can provide rapid, objective, and accurate diagnosis of oesophageal pathology ex vivo. The confocal design of this probe enables superficial mucosal abnormalities (metaplasia and dysplasia) to be classified in clinically applicable timescales paving the way for an in vivo trial.
Multivariate classifiers (such as Linear Discriminant Analysis, Support Vector Machines etc) are known to be useful
tools for making diagnostic decisions based on spectroscopic data. However, robust techniques for assessing their
performance (e.g. by sensitivity and specificity) are vital if the application of these methods is to be successful in the
clinic. In this work the application of repeated cross-validation for estimating confidence intervals for sensitivity and
specificity of multivariate classifiers is presented. Furthermore, permutation testing is presented as a suitable technique
for estimating the probability of obtaining the observed sensitivity and specificity by chance. Both approaches are
demonstrated through their application to a Raman spectroscopic model of gastrointestinal cancer.
Raman spectroscopy is an inelastic scattering technique capable of probing the biochemical changes associated with
neoplastic progression in oesophageal tissue. Custom-built fibre-optic Raman probes could potentially provide
opportunities for in vivo endoscopic diagnosis of pre-cancerous oesophageal lesions and targeted early therapy.
However, prior to commencing a clinical trial convincing ex vivo work must demonstrate multi-operator, multi-centre
and multi-system reliability. We report spectral consistency between two operators who independently evaluated two
optically identical probes ex vivo. In addition, we demonstrate compatibility with high-definition white light endoscopes
and narrow band imaging systems highlighting the potential for future endoscopic multi-modality imaging in the
oesophagus.
Rapid Raman mapping has the potential to be used for automated histopathology diagnosis, providing an adjunct technique to histology diagnosis. The aim of this work is to evaluate the feasibility of automated and objective pathology classification of Raman maps using linear discriminant analysis. Raman maps of esophageal tissue sections are acquired. Principal component (PC)-fed linear discriminant analysis (LDA) is carried out using subsets of the Raman map data (6483 spectra). An overall (validated) training classification model performance of 97.7% (sensitivity 95.0 to 100% and specificity 98.6 to 100%) is obtained. The remainder of the map spectra (131,672 spectra) are projected onto the classification model resulting in Raman images, demonstrating good correlation with contiguous hematoxylin and eosin (HE) sections. Initial results suggest that LDA has the potential to automate pathology diagnosis of esophageal Raman images, but since the classification of test spectra is forced into existing training groups, further work is required to optimize the training model. A small pixel size is advantageous for developing the training datasets using mapping data, despite lengthy mapping times, due to additional morphological information gained, and could facilitate differentiation of further tissue groups, such as the basal cells/lamina propria, in the future, but larger pixels sizes (and faster mapping) may be more feasible for clinical application.
The use of Raman spectroscopy in the detection and classification of malignancy within lymph nodes of the head and neck has been evaluated. Currently histopathology is considered the diagnostic gold standard. A consensus (majority) opinion from three expert histopathologists has been obtained and spectral diagnostic models developed by correlation with their opinions.
Raman spectra have been measured at 830nm from 103 lymph nodes collected from patients undergoing surgery for a suspicious node. The pathologies covered reactive lymph nodes, primaries from Hodgkin's and non-Hodgkin's lymphomas and metastases from squamous cell carcinomas and adenocarcinomas. Spectral diagnostic models were
constructed using PCA-fed-LDA and tested using leave-one-specimen-out cross validation. Models were constructed to distinguish between reactive and malignant nodes as well as a four group model to distinguish between the benign, metastatic and primary conditions. They achieved 89% and 84% correct prediction by node versus the gold standard, majority histopathology.
Ultra-low spatial resolution Raman (ULSRR) mapping using fibre probes has been performed on mammalian and human
tissues. This will provide an understanding of the potential for in vivo surveillance of the lining of organs using such a
technique and for identifying abnormal tissues such as residual tumours within a surgical field.
The aim of the study was to create Raman probe map images of excised oesophageal specimens following radical and
palliative oesophagectomy procedures. A reproducible mapping grid was placed over the excised tissue surface and
Raman mapping at 830nm performed at regular intervals to provide images of 200 pixels over the region of interest.
Principal component analysis was used to create pseudocolour score images of both porcine phantoms and a human
resected oesophagus.
A principal component fed linear discriminant (LD) classification model of 72 biopsy samples from 35 patients was
created using a novel single fibre Raman probe. A subset of the training dataset was used to populate a matrix of 200
pixels to simulate a Raman probe map. Spectra from the simulated map were then projected onto the LD model and a
pseudocolour LD pathology map created.
Delineation of clinically significant pathology groups was demonstrated therefore this study has shown the feasibility of
in vivo ULSRR for margin assessment using a Raman probe.
Raman spectroscopy has proved to be a highly sensitive tool for differentiating between normal, cancerous or pre-cancerous
tissues. To date, histological application of Raman mapping has been limited due to lengthy mapping times.
StreamlineTM Raman imaging is a novel mapping technique that has reduced total mapping times to a level that is
becoming clinically practicable. Raman Streamline mapping was carried out on a 20μm frozen section of an oesophageal
biopsy. A contiguous 7μm section was stained with haematoxylin and eosin (H&E) with histpathology analysed by a
pathologist. The step size and acquisition times were varied and the resulting spectra, principal component score maps
and loads were compared. The signal to noise for the raw spectra and a relative 'signal to noise' of the principal
component loads were determined. The Streamline mapping technique was also compared to traditional point Raman
mapping. The principal component loads were similar despite varying the acquisition time and number of spectra, with
the fifth load used for comparison of the noise levels. Gross biochemical information was extracted showing good
correlation with the H&E section even for short overall mapping times as low as 30-90 minutes for a biopsy ~2mm in
diameter (0.5s acquisition time per 25.3μm Raman pixel). Streamline mapping was of the order of 3-7 times faster than
traditional point mapping with the greatest improvement made for high resolution maps. Further optimization of the
system is still possible which will reduce this mapping time further making implementation in a clinical environment a
future possibility.
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.