Editorial. The Raman effect on intraoperative diagnosis of central nervous system tumors

Howard ColmanDepartment of Neurosurgery and Huntsman Cancer Institute, University of Utah, Salt Lake City, Utah

Search for other papers by Howard Colman in
jns
Google Scholar
PubMed
Close
 MD, PhD
Free access

The Raman effect, also known as Raman scattering, was discovered by C. V. Raman in 1928 and subsequently led to a Nobel Prize in Physics in 1930. In contrast to the more commonly understood scattering of light reflecting off an object in the visual world, in which the photons retain their original frequency and energy, Raman scattering results in the emission of photons with a different frequency and energy than the incident light. These alterations depend on interactions with molecules of matter, resulting in the inelastic scattering of photons and causing both a change in direction and a change in energy of the reflected light. As a result, the output of Raman scattering can be used to evaluate the specific properties of the material that is interacting with the incident light. Interpretation of these data requires high-intensity light sources such as lasers and specialized photodetectors on a microscopic scale.

A growing body of literature indicates that techniques based on Raman scattering can be used for histopathological diagnosis. Prior studies have demonstrated that stimulated Raman scattering provides sufficient signal to noise to be successfully implemented for imaging of multiple tissues without additional staining.1 Subsequent studies utilized various applications of this approach for diagnosis and molecular classification of brain tumors from histopathological samples.25 These applications have included distinguishing active tumor from necrosis, distinguishing areas of infiltrating tumor from normal brain, and more recently intraoperative diagnosis.5 In the current issue of Neurosurgical Focus, Wadiura et al. present the results of a study that compared the diagnostic results of stimulated Raman histology (SRH) with those of conventional frozen section and H&E images from the same patients treated at a single institution.6

Interpretation of frozen sections during a neurosurgical procedure has been the standard approach for intraoperative diagnosis and informing intraoperative decision-making. However, the preparation and interpretation of frozen sections require multiple steps that take time. Ultimately, the sensitivity and specificity of frozen section diagnosis—even in the hands of an experienced neuropathologist—is suboptimal compared with that based on permanent sections and additional immunohistochemical and molecular information obtained later in the diagnostic process. Thus, any advances that improve the accuracy of real-time intraoperative diagnosis could have a significant impact on patient care.

In the current paper by Wadiura et al., the study population included 94 patients with suspected newly diagnosed or known recurrent CNS tumors. Samples were obtained and processed during neurosurgical resection or biopsy for SRH and subsequent digital images were produced. Additional intraoperative frozen section was also performed according to the neurosurgeon’s preference. Comparison of diagnostic classification with SRH versus conventional histology was performed retrospectively by aggregating the digital SRH images with digitized frozen section and H&E images from the same patient. These SRH and conventional histopathological images were reviewed by two neuropathologists for image quality and diagnosis. Individual tumors were independently scored on the basis of various parameters, including cellularity and diagnostic histopathologic characteristics, by each neuropathologist.

The series of 94 tumors consisted of different final diagnoses, including meningioma (32%), high-grade glioma (30%), and metastases (20%) as the most frequent histologies. The results demonstrated that intraoperative SRH image acquisition was feasible in all cases and that a histopathologic diagnosis was possible based on SRH images in 92 of 94 cases (98%). Comparison of the SRH diagnoses with the final H&E histopathologic diagnoses in the same cases demonstrated good concordance, with the same diagnosis obtained using the different approaches in 91 of 92 cases. In the 21% of cases where frozen section was performed in addition to SRH, the concordance with histopathologic diagnosis based on frozen section was 100%. Semiquantitative scoring of cellularity and histopathologic characteristics also showed reasonable correlations between methods.

The authors concluded that this study validated the potential use of SRH for intraoperative diagnosis by demonstrating the diagnostic accuracy and concordance of SRH compared with both frozen section and final H&E histology. In addition, the authors demonstrated the feasibility of image acquisition and quality in a high percentage of neurosurgical cases, and they provided at least preliminary evidence that quantification of cellularity and diagnostic histologic features may be similar between SRH and conventional histological techniques. All these findings support the idea that use of SRH for intraoperative diagnosis is feasible and may have advantages over frozen section, at least in some situations. However, there are some caveats for the widespread adoption of SRH for intraoperative diagnosis. First, although the authors prospectively acquired all samples for SRH and conventional histology, the actual comparison for diagnostic yield and accuracy was retrospective. In order to seriously consider replacing frozen section with SRH analyses, a prospective study is required in which the intraoperative diagnosis is determined in a blinded fashion using both approaches and then feasibility and accuracy of diagnosis compared with final histopathology is determined for SRH versus frozen section. Second, this approach involves the purchase of specialized equipment for imaging and/or interpretation, and all this technology needs to be immediately and rapidly available in order to be feasible for real-time intraoperative diagnosis. Lastly, the use of SRH for diagnostic purposes would undoubtedly require significant training to improve familiarity for neuropathologists used to conventional frozen section methods. Thus, the results obtained in this study by investigators with possible prior experience with this technology need to be demonstrated in a wider variety of centers that do not have that prior experience. The use of automated interpretation of SRH images may be a solution to this problem, with a prior multicenter study demonstrating the feasibility of acquiring SRH images and the noninferior diagnostic accuracy of a convoluted neural network–trained algorithm compared with that of neuropathologist-based interpretation.

Overall, the current study adds to a growing body of data showing that advanced techniques for acquisition and interpretation of microscopic images are likely to play an important role, both in combination with and sometimes instead of conventional histopathology. The role of digital pathology and artificial intelligence for determination of diagnosis, prognosis, treatment prediction, molecular diagnosis, and other applications has been highlighted in a number of recent high-profile reviews and journal editions.7,8

Disclosures

Dr. Colman in a consultant for Best Doctors/Teladoc, Orbus Therapeutics, Bristol Meyers Squibb, Regeneron, and Novocure.

References

  • 1

    Freudiger CW, Min W, Saar BG, et al. Label-free biomedical imaging with high sensitivity by stimulated Raman scattering microscopy. Science. 2008;322(5909):18571861.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 2

    Ji M, Orringer DA, Freudiger CW, et al. Rapid, label-free detection of brain tumors with stimulated Raman scattering microscopy. Sci Transl Med. 2013;5(201):201ra119.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 3

    Hollon T, Lewis S, Freudiger CW, et al. Improving the accuracy of brain tumor surgery via Raman-based technology. Neurosurg Focus. 2016;40(3):E9.

  • 4

    Reinecke D, von Spreckelsen N, Mawrin C, et al. Novel rapid intraoperative qualitative tumor detection by a residual convolutional neural network using label-free stimulated Raman scattering microscopy. Acta Neuropathol Commun. 2022;10(1):109.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 5

    Hollon TC, Pandian B, Adapa AR, et al. Near real-time intraoperative brain tumor diagnosis using stimulated Raman histology and deep neural networks. Nat Med. 2022;26(1):5258.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 6

    Wadiura LI, Kiesel B, Roetzer-Pejrimovsky T, et al. Toward digital histopathological assessment in surgery for central nervous system tumors using stimulated Raman histology. Neurosurg Focus. 2022;53(6):E12.

    • Search Google Scholar
    • Export Citation
  • 7

    Shmatko A, Laleh NG, Gerstung M, et al. Artificial intelligence in histopathology: enhancing cancer research and clinical oncology. Nat Cancer. 2022;3(9):10261038.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8

    Herrington CS, Poulsom R, Pillay N, et al. Recent advances in pathology: the 2022 Annual Review Issue of The Journal of Pathology. J Pathol. 2022;257(4):379382.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Collapse
  • Expand
  • 1

    Freudiger CW, Min W, Saar BG, et al. Label-free biomedical imaging with high sensitivity by stimulated Raman scattering microscopy. Science. 2008;322(5909):18571861.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 2

    Ji M, Orringer DA, Freudiger CW, et al. Rapid, label-free detection of brain tumors with stimulated Raman scattering microscopy. Sci Transl Med. 2013;5(201):201ra119.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 3

    Hollon T, Lewis S, Freudiger CW, et al. Improving the accuracy of brain tumor surgery via Raman-based technology. Neurosurg Focus. 2016;40(3):E9.

  • 4

    Reinecke D, von Spreckelsen N, Mawrin C, et al. Novel rapid intraoperative qualitative tumor detection by a residual convolutional neural network using label-free stimulated Raman scattering microscopy. Acta Neuropathol Commun. 2022;10(1):109.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 5

    Hollon TC, Pandian B, Adapa AR, et al. Near real-time intraoperative brain tumor diagnosis using stimulated Raman histology and deep neural networks. Nat Med. 2022;26(1):5258.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 6

    Wadiura LI, Kiesel B, Roetzer-Pejrimovsky T, et al. Toward digital histopathological assessment in surgery for central nervous system tumors using stimulated Raman histology. Neurosurg Focus. 2022;53(6):E12.

    • Search Google Scholar
    • Export Citation
  • 7

    Shmatko A, Laleh NG, Gerstung M, et al. Artificial intelligence in histopathology: enhancing cancer research and clinical oncology. Nat Cancer. 2022;3(9):10261038.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8

    Herrington CS, Poulsom R, Pillay N, et al. Recent advances in pathology: the 2022 Annual Review Issue of The Journal of Pathology. J Pathol. 2022;257(4):379382.

    • Crossref
    • Search Google Scholar
    • Export Citation

Metrics

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 264 264 32
PDF Downloads 246 246 20
EPUB Downloads 0 0 0