Lighting the path: a narrative review of non-molecular intraoperative lung imaging modalities

Current protocols lack validated non-destructive modalities to intraoperatively evaluate lung tissue without excision and histological processing, a critical step for confirming negative margins during cancer resections. Surgeons currently rely on preoperative radiologic imaging and intraoperative tactile feedback. Consequently, there is a persistent need to improve intraoperative identification and visualization of lung pathologies, including nodules, ground-glass opacities, and their margins relative to healthy parenchyma. Optical imaging modalities have emerged as promising tools for intraoperative lung visualization. While molecular imaging approaches have been extensively reviewed, non-molecular intraoperative optical modalities remain comparatively underrepresented. This narrative review summarizes recent clinical developments in non-molecular optical imaging of lung tissue and compares them with contrast-dependent approaches. Inclusion and exclusion criteria focused the scope on primary literature reporting intraoperative optical imaging of lung parenchyma or bronchial tissues between 2008 and 2025. Imaging modalities were evaluated based on their ability to visualize lung tissue and discriminate healthy from abnormal regions.

A PubMed search (2008-2025) identified English-language studies on intraoperative optical imaging of lung parenchyma using relevant spectroscopy-related terms. Results were screened using predefined criteria and independently reviewed for reliability. Studies not focused on lung parenchyma were excluded.

Molecular imaging techniques included indocyanine green (ICG)-based imaging, with and without targeted probes, as well as confocal laser endomicroscopy (CLE). Exogenous contrast-dependent modalities demonstrate efficacy in lung surgery, enabling identification of nodules and intersegmental planes. However, these approaches are limited by contrast spillover into the operative field, dye-associated adverse reactions, narrow imaging time windows, and reliance on surrogate biomarkers that provide limited quantitative information. In contrast, non-molecular imaging modalities operate independently of exogenous agents and directly measure intrinsic optical properties of tissue. Their limitations include shallow tissue penetrance and challenges associated with analyzing large spectral data sets. Machine learning (ML) techniques may enhance both molecular and non-molecular imaging by improving image processing, enabling data integration, and supporting complementary multimodal imaging strategies.

Non-molecular intraoperative imaging modalities are clinically feasible and safe. They enable differentiation of pathological and normal lung tissue while assessing physiologic properties without contrast agents, thereby avoiding spill-over and time constraints. These technologies offer to enhance intraoperative visualization beyond preoperative imaging and tactile assessment. With refinement and validation, non-molecular optical imaging may reduce dependence on histopathology, improve intraoperative decision-making, and enhance surgical outcomes.