ICG-augmented hyperspectral imaging for visualization of intestinal perfusion compared to conventional ICG fluorescence imaging: an experimental study

Alexander Studier‐Fischer, Florian Marc Schwab, Maike Rees, Silvia Seidlitz, Jan Sellner, Berkin Özdemir, Leonardo Ayala, Jan Odenthal, Samuel Knoedler, Karl‐Friedrich Kowalewski, Caelán Max Haney, Maximilian Dietrich, Gabriel Alexander Salg, Hannes Götz Kenngott, Beat P. Müller‐Stich, Lena Maier‐Hein, Felix NickelInternational Journal of Surgery 2023 · DOI

TL;DR

This study introduces ICG-augmented hyperspectral imaging (IA-HSI) as a superior method for visualizing intestinal perfusion compared to conventional ICG fluorescence imaging. IA-HSI demonstrated greater accuracy in identifying the extent of malperfusion, potentially improving surgical safety and reducing the risk of unnecessary bowel resection.

What Problem Does This Paper Address?

Small bowel malperfusion (SBM) is a serious condition that can lead to significant morbidity and mortality. Accurately determining the extent of malperfused tissue is crucial for surgical intervention, particularly for deciding on resection margins. Current methods, including surgeon's visual inspection and conventional Indocyanine Green (ICG) fluorescence imaging, lack standardization and objectivity, leading to potential over- or under-resection. This can result in either sacrificing healthy bowel, impacting nutrient absorption, or leaving behind compromised tissue, increasing the risk of complications and reoperation. The study aims to address this diagnostic gap by evaluating a novel approach, IA-HSI, which combines hyperspectral imaging (HSI) with ICG.

Methods

The core of this research lies in the development and evaluation of ICG-augmented hyperspectral imaging (IA-HSI). Hyperspectral imaging (HSI) captures spectral reflectance data across a broad range of wavelengths, allowing for the assessment of tissue properties like oxygenation (StO2) without external contrast agents. In this study, HSI was augmented by the intravenous administration of Indocyanine Green (ICG), a fluorescent dye commonly used in medical imaging.

The experimental setup involved several key steps:

  • Baseline HSI: Physiological small bowel, avascular small bowel (created via mesotomy), and small bowel after ICG administration were imaged using HSI to establish baseline spectral signatures.
  • Malperfusion Models: Two models of malperfusion were created in pigs: irreversible mesotomy-induced and reversible clamp-induced. These were monitored over time to observe changes in perfusion.
  • IA-HSI Evaluation: In the final experimental phase, mesotomy-induced malperfusion was induced, and the affected bowel segments were evaluated using four methods:
  • HSI only: Standard hyperspectral imaging.
  • IA-HSI: Hyperspectral imaging with ICG.
  • Clinical evaluation: Surgeon's visual assessment.
  • Conventional ICG imaging: Standard ICG fluorescence imaging.
  • Histopathological Correlation: Tissue samples were collected from different regions identified by these modalities and analyzed histopathologically to assess pre-necrotic changes, providing a ground truth for perfusion status.
  • Vascular Corrosion Casting: This technique was used to visualize the microvasculature of the small bowel, providing a structural explanation for the observed perfusion patterns.

The TIVITA Tissue system was used for HSI acquisition, providing spectral resolution and spatial data. Data analysis involved quantitative measurements of resection border distances and statistical comparisons between the imaging modalities.

Datasets

This study utilized an experimental animal model (pigs) for data acquisition. A total of 54 pigs were used for in-vivo experiments. No pre-existing public datasets were used; instead, the researchers generated their own data through controlled experimental procedures. The study did not introduce a new benchmark dataset in the traditional machine learning sense, but rather generated experimental data to compare imaging modalities.

Key Insights and Ideas

  • ICG as a "Smart" Contrast Agent for HSI: The study demonstrates that ICG, when combined with HSI, can act as a powerful contrast agent. While ICG alone has limitations in objectivity, its fluorescence properties, when analyzed by HSI, provide enhanced information about tissue perfusion.
  • IA-HSI Reveals Sharper Resection Margins: IA-HSI was able to suggest a more defined resection line where StO2 values began to decrease, unlike conventional HSI which showed a more gradual decline. This sharper delineation is crucial for precise surgical decision-making.
  • Histopathology Validates IA-HSI Superiority: The gold standard of histopathological analysis confirmed that regions deemed sufficiently perfused by conventional ICG imaging still showed significant pre-necrotic changes, indicating that IA-HSI is more sensitive in detecting compromised tissue.
  • ICG Creates Artifacts in StO2, But These are Useful: The study acknowledges that ICG administration causes visual artifacts in StO2 values (an apparent increase), which are not true oxygenation changes. However, they ingeniously propose that this artifact can be empirically leveraged to visualize segmental SBM.
  • Vascular Architecture Explains Perfusion Patterns: Vascular corrosion casting provided a clear anatomical basis for the observed perfusion, showing how perfusomes relate to the identified resection borders.

Key Results

  • Resection Border Differences: Conventional ICG imaging suggested a resection line significantly further towards the malperfused region (53 mm ± 13 mm) compared to IA-HSI (which suggested a sharp line where StO2 decreased). Clinical evaluation suggested a line 23 mm (±7 mm) from the healthy tissue.
  • Histopathological Findings: Regions considered well-perfused by conventional ICG imaging showed a 27% (±9%) increase in pre-necrotic changes, highlighting the insensitivity of this method. IA-HSI-identified well-perfused regions showed comparable pre-necrotic changes to physiological controls.
  • IA-HSI Sensitivity: IA-HSI was found to be superior to conventional HSI, clinical evaluation, and conventional ICG imaging in terms of histopathological safety for visualizing segmental SBM.
  • StO2 Cutoff: A retrospectively fitted StO2 cutoff value of 51% after ICG application was suggested for diagnosing seriously malperfused small bowel.

Limitations and Future Work

  • "Off-Label" Use of ICG: The authors acknowledge that using ICG as a contrast agent with HSI is an "off-label" application, as the computation of physiological parameters like StO2 is based on assumptions about chromophores that are altered by ICG. This means the calculated index values may not be physiologically accurate but are empirically useful.
  • Small Sample Size: The main experimental group consisted of 10 pigs. While considered sufficient for a large-animal model, the authors suggest that consistency of results needs to be evaluated in larger, consecutive studies.
  • Animal Model to Human Translation: The study was conducted in pigs, and while small bowel physiology is highly comparable to humans, direct translation of results requires further exploration.
  • Temporal Fluctuations: StO2 values and underlying tissue spectra change over time after ICG administration, meaning the timing of imaging is critical and can lead to temporal fluctuations in assessment.
  • Subjectivity in Border Placement: The initial placement of resection borders was subjective, and while inter-rater variability was low, the comparison of distances is susceptible to this. The key takeaway is the general tendency, not the precise millimeter measurements.
  • Future Work: The authors emphasize the need for clinical trials to explore the clinical applicability and relevance of IA-HSI. They also suggest prospective testing of resection margins evaluated by different modalities and the development of more globally applicable StO2 cutoff values through in-depth prospective analysis.

Summary updated

Abstract

Not applicable. Translational animal science. Original article.
BibTeX
@article{studierfischer2023icg,
  author = {Alexander Studier‐Fischer and Florian Marc Schwab and Maike Rees and Silvia Seidlitz and Jan Sellner and Berkin Özdemir and Leonardo Ayala and Jan Odenthal and Samuel Knoedler and Karl‐Friedrich Kowalewski and Caelán Max Haney and Maximilian Dietrich and Gabriel Alexander Salg and Hannes Götz Kenngott and Beat P. Müller‐Stich and Lena Maier‐Hein and Felix Nickel},
  title = {{ICG-augmented hyperspectral imaging for visualization of intestinal perfusion compared to conventional ICG fluorescence imaging: an experimental study}},
  journal = {International Journal of Surgery},
  year = {2023},
  doi = {10.1097/js9.0000000000000706},
}
Figure