Probe-based confocal laser endomicroscopy imaging features of malignant lung nodules: results from a prospective in vitro study

Introduction

Extensive trials worldwide have shown that low-dose computed tomography (CT) lung cancer screening leads to a significant improvement in the detection rate of early-stage lung cancer, a significant reduction in the detection rate of advanced-stage lung cancer, and a 20% decrease in the risk of lung cancer mortality (1). The extensive use of low-dose CT screening has resulted in the clinical identification of a considerable number of patients with pulmonary nodules, a substantial proportion of which comprise indeterminate pulmonary nodules. Minimally invasive biopsy is a critical diagnostic modality for indeterminate pulmonary nodules. Common minimally invasive biopsy methods for pulmonary nodules include CT-guided percutaneous lung biopsy, bronchoscopic lung biopsy, and thoracoscopic lung biopsy. While these methods can effectively achieve pathological diagnosis of pulmonary nodules, they carry risks associated with invasive procedures. Additionally, for smaller pulmonary nodules, the positive rate of biopsy is relatively low, potentially failing to provide an accurate diagnosis (2).

Confocal laser endomicroscopy (CLE) is an endoscopic instrument based on confocal microscopy principles, enabling micrometer-level high-resolution imaging in the body, with real-time, dynamic, and relatively non-invasive characteristics (3). Currently, the two types of CLE applied in respiratory systems are needle-based confocal laser endomicroscopy (nCLE) and probe-based confocal laser endomicroscopy (pCLE) (4). The nCLE is a flexible needle-type catheter with a diameter of 0.9 mm, featuring a lateral resolution of 3.5 µm, an observation depth of 40–70 µm, and a field of view of 325 µm × 325 µm. The pCLE is a semi-flexible probe with a diameter of 1.4 mm, comprising 3,000 microfibers. It offers a lateral resolution of 0.91 µm, an observation depth of 50 µm, a field of view of 600 µm × 600 µm, and a scanning speed of up to 12 frames per second (5). Compared to nCLE, pCLE offers a broader imaging range and higher image resolution, making it suitable for in vivo diagnosis of pulmonary nodules. Substantial research has been conducted on the combination of navigation bronchoscopy and pCLE, using the intravenous injection of sodium fluorescein to facilitate the real-time imaging of lung nodules. This approach has been reported to have a diagnostic accuracy rate of 90%; however, this figure needs to be verified, as it may refer to the localization accuracy rate rather than the diagnostic accuracy rate (4-9). Furthermore, CLE has been applied to the diagnosis of interstitial lung disease, granulomatous lesions, cellular detection in pleural effusion, and pleural invasion in lung cancer (10-13). Notably, there are certain risks associated with the intravenous injection of sodium fluorescein. For example, some patients may experience hypersensitivity reactions or even anaphylactic shock, while others may develop severe adverse reactions affecting multiple systems. The use of sodium fluorescein has been shown to increase risks for both patients and operators (5-6,10), while also decreasing patient compliance.

This study sought to use CLE technology for the in vitro imaging of lung nodules without fluorescence injection. By analyzing the characteristics of CLE images of lung nodules, this study sought to establish a map of malignant lung nodules for in vivo CLE imaging without fluorescence injection. We present this article in accordance with the STARD reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1805/rc).

Methods

Study population

This prospective study was approved by the Ethics Committee of Chinese PLA General Hospital (No. S2023-590-01), and conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was registered at Chinese Clinical Trial Registry (www.chictr.org.cn, 26/12/2023, ChiCTR2300079176). Before the study, all the patients were asked to provide individual informed consent in the form of a signed document.

From December 2023 to August 2024, patients scheduled to undergo lung nodule resection surgery at the Department of Thoracic Surgery at Chinese PLA General Hospital were screened for inclusion in this study. Patients were excluded from the study if they met any of the following exclusion criteria: (I) did not undergo the surgical procedure as initially scheduled; (II) lacked CLE images; (III) lacked definitive pathological results following surgery; and/or (IV) were deemed inappropriate for participation in the study by the investigator for any reason.

Study process

The study was divided into three phases. In phase 1, the patients underwent standard surgical procedures. Following the removal of the lesion, the surgeon confirmed its location and extent. A researcher employed CLE (Cellvizio, Mauna Kea Technologies, Paris, France) to image the lesion specimen in the following sequence: target normal tissue, tissue adjacent to the lesion, and lesion tissue. Videos and images were captured during the process. To ensure the standard of visual documentation, each video imaging session had to last longer than 30 seconds, while ensuring the total imaging time remained under 10 minutes. The final diagnosis was based on the postoperative pathological diagnosis of the nodule. In phase 2, a collaborative analysis of the CLE images of the pulmonary nodules was conducted by pathologists, pulmonologists, and interventional pulmonologists, who also extracted the CLE imaging features. In phase 3, evaluators who had not participated in the CLE imaging feature analysis received training on the CLE imaging features of pulmonary nodules, and reviewed and evaluated new CLE images. After the initial training validation, a two-week washout period was observed, after which, a second training session and image assessment were conducted. The analysis of videos and images was facilitated by the use of the Cellvizio Viewer (Mauna Kea Technologies). The study workflow is shown in Figure 1.

Figure 1 Flow chart of the study. CLE, confocal laser endomicroscopy; pCLE, probe-based confocal laser endomicroscopy; nCLE, needle-based confocal laser endomicroscopy.

Acquisition of CLE images

The present study employed a pCLE (Cellvizio, Mauna Kea Technologies) platform system for imaging purposes. The patients enrolled in the study underwent routine lung nodule resection surgery performed by thoracic surgeons. The thoracic surgeons confirmed the presence of lung nodules in the resected tissue, after which the researchers used pCLE to image the tissue in the following order: normal tissue, tissue adjacent to the lesion, and lesion tissue, while recording the CLE videos. To ensure optimal imaging quality, each video was recorded for a minimum duration of 30 seconds. The specimens examined using pCLE were then sent for pathological examination, after which, postoperative follow-up of the pathological results was conducted.

Analysis of CLE imaging features

The videos were reviewed by three researchers (a pulmonologist with over 10 years of experience, a pathologist with over 15 years of experience, and a bronchoscopist with over 10 years of experience). The CLE imaging features of normal tissue, tissue adjacent to the lesion, and lesion tissue were analyzed in conjunction with the pathological diagnosis results of the lung nodules to identify the CLE imaging features of the lung nodule lesions.

Inter- and intra-observer consistency analysis

The evaluation was conducted by five evaluators, comprising two senior pulmonologists with over 10 years of experience each, one attending pulmonologist with over five years of experience, one resident pulmonologist with over 2 years of experience, and one pulmonology doctoral student. The inter- and intra-observer consistency of the evaluators was analyzed. None of the evaluators participated in the experimental design, image acquisition, nor image feature analysis. The researchers presented the imaging features of lung nodules using CLE (including textual explanations, images, and video demonstrations) to the five evaluators, distinguishing between the imaging features of normal tissue and lung nodule tissue. The evaluators underwent training in CLE imaging feature recognition, with the entire training process lasting no more than 20 minutes. After the training, the initial validation was undertaken, wherein observers reviewed CLE videos of lung nodules to determine whether the CLE-detected lesions were normal or abnormal, and whether malignant nodule imaging features were present. The objective of this study was to evaluate the inter-observer agreement (IOA) of the evaluators. A fortnight after the initial evaluation, the five evaluators undertook a second evaluation for the intra-observer reliability (IOR) analysis.

Statistical analysis

The statistical analysis of the IOR, IOA, sensitivity, specificity, positive predictive value, negative predictive value, and diagnostic rate was performed using SPSS version 26.0 (IBM Corporation, Somers, NY, USA). The five evaluators assessed 95 CLE videos to distinguish between normal/lesion-adjacent, and lesion tissues, and determine the presence of malignant features, and the diagnostic rates of the assessors were calculated. Each evaluator conducted two evaluations with a 2-week interval between evaluations. Cohen’s κ was used to compare the results of each evaluator’s first and second validation sessions to analyze the consistency of individual evaluators, and the multi-observer Fleiss κ statistic was used to calculate IOA and IOR. The IOA and IOR results were interpreted using the Landis-Koch interpretation system as follows: ≤0.20: poor; 0.21–0.40: fair; 0.41–0.60: moderate; 0.61–0.80: good; and 0.81–1.00: excellent.

Results

Patient characteristics

A total of 29 patients were included in this study, of whom four had multiple nodules. In total, 33 pulmonary nodule lesions were surgically resected. In one patient with multiple pulmonary nodules, a single nodule lesion was pathologically diagnosed postoperatively as a mucinous tumor of the lung, with the possibility of mucinous adenocarcinoma or bronchioloalveolar carcinoma being excluded. However, due to the ambiguity of the pathological results, the case was excluded from further analysis. Thus, the final study comprised a total of 29 patients with 32 pulmonary nodules. Of the 32 pulmonary nodules identified, 30 were malignant, including 29 adenocarcinomas (of which, five were minimally invasive adenocarcinomas, and 24 were invasive adenocarcinomas), and one squamous cell carcinoma. Two of the lesions were determined to be benign, with one pathological report indicating a substantial accumulation of inflammatory cells, and the other report indicating a bronchial adenoma. Each patient underwent a minimum of three CLE imaging sessions, encompassing the imaging of the lesion itself, adjacent area, and normal tissue. The clinical details of the patients are provided in Table 1.

CLE imaging features of normal lung tissue and tissue adjacent to lesions

The CLE images of the normal tissue displayed regular and intact alveolar structures, connected in a trapezoidal pattern, with fine and continuous elastic fibers, and occasional intact micro vessels. The tissue adjacent to the lesion displayed features consistent with normal tissue, including relatively intact and regular alveolar structures, with no discernible structural damage. The mean alveolar septa thickness of the normal tissue was 7.6±3.2 µm, and that of the tissue adjacent to the lesion was 9.51±11.49 µm.

CLE imaging features of malignant pulmonary nodules

The CLE images of the lesion tissue showed significant differences compared to those of the surrounding tissue and normal tissue. The mean thickness of the lesion tissue was 15.33±8.5 µm. The following eight common CLE imaging features of malignant pulmonary nodules were identified: (I) thickened alveolar septa; (II) diffuse hyperplasia of elastic fibers (cloud shadows); (III) curled alveolar septa; (IV) destruction of alveolar structures; (V) broken elastic fibers; (VI) fragmented elastin; (VII) defective alveolar structures (loss of elastin); and (VIII) microvascular disruption. Examples of the CLE images are provided in Figure 2.

Figure 2 Confocal laser endomicroscopy imaging features of normal lung tissue, tissue adjacent to pulmonary nodules, and pulmonary nodule tissue. (A) The normal alveolar structure of lung tissue. (B) Thickening of alveolar walls. (C) Cloud shadows: diffuse, homogeneous, pale grey shadows in alveolar spaces. (D) Curled alveolar septa: elastic fibers exhibiting curling changes. (E) Destruction of alveolar structures: alveolar collapse and adhesion. (F,G) Broken elastic fibers. (H) Fragmented elastin. (I) Defective alveolar structures: loss of elastin. (J) Normal vessels. (K) Microvascular disruption. (L) Benign lesions (inflammatory lesions).

The eight features were subsequently grouped into the following three overarching categories: (I) hyperplasia, which is primarily characterized by the thickening of the alveolar septa and diffuse proliferation of elastic fibers; (II) deformation, which primarily manifests as curling of the alveolar walls, alveolar collapse, and adhesion of the alveolar walls; and (III) destruction, which primarily manifests as significant destruction of the alveolar structure, including the rupture of elastic fiber, the accumulation of elastic protein in fragments, structural defects (an absence of elastic protein), and microvasculature destruction. Lung nodules exhibiting these features are often moderately differentiated adenocarcinomas rather than poorly differentiated adenocarcinomas.

Meanwhile, as the malignancy of the lesion increased, the CLE images began to show significant destruction of the alveolar structure, including the rupture of elastic fibers, the aggregation of elastic proteins, and large-scale accumulation. Microvascular structure destruction was also observed. In malignant lesions exhibiting micropapillary or solid growth, there was an evident deficiency of elastic protein, manifesting as large black areas in the CLE field of view. In the context of malignancy, there was an observable increase in the severity of alveolar structural destruction. Initially, the changes were primarily proliferative, but as the extent of the invasion increased, significant structural destruction became evident. In severe cases, there was an absence of elastin due to its extensive degradation (Figures 3-5).

Figure 3 Confocal laser endomicroscopy imaging features of lung adenocarcinoma. The features may vary due to heterogeneity. (A) Thickened alveolar septa. (B) Curled alveolar septa. (C) Fragmented elastin. (D) Loss of elastin.

Figure 4 Pathology and confocal laser endomicroscopy features of minimally invasive adenocarcinoma. (A) H&E staining, 20× magnification. (B) Elastic-Van Gieson’s staining, 20× magnification. (C) Confocal laser endomicroscopy image showing thickened alveolar septa, curled alveolar septa, and alveolar collapse. H&E, hematoxylin and eosin.

Figure 5 Pathology and confocal laser endomicroscopy features of invasive adenocarcinoma. (A) H&E staining, 20× magnification. (B) Elastic-Van Gieson’s staining, 20× magnification. (C) Confocal laser endomicroscopy image showing thickened alveolar septa, alveolar collapse and adhesion, broken elastic fibers, and fragmented elastin. H&E, hematoxylin and eosin.

CLE imaging features of benign pulmonary nodules

Two benign pulmonary nodules were included in this study, one of which was characterized by a substantial accumulation of inflammatory cells, and the other by a bronchial adenoma. The CLE image of the lesion, which exhibited a substantial accumulation of inflammatory cells, showed a mixture of bright and dark areas, with the absence of alveolar structure, and the presence of no normal or abnormal elastic fibers. A significant number of inflammatory cells were also observed. Conversely, the CLE image of bronchial adenoma exhibited certain characteristics that were analogous to those of malignant nodules, including the disruption of the alveolar structure. However, the most distinguishing feature was the presence of markedly enlarged glandular openings, a hallmark of bronchial adenoma. A summary of the CLE imaging features for all the lesions is presented in Table 2.

Consistency in determining the nature of lesions based on CLE images

Among the five evaluators, the overall sensitivity and specificity in distinguishing between lesion tissue and non-lesion tissue (normal tissue/lesion-adjacent tissue) in the CLE images were 100% and 93.7%, respectively. The positive predictive value was 88.9%, the negative predictive value was 100%, the accuracy rate was 95.8%, and the IOA was 0.811. In the non-lesion tissue, a separate assessment was conducted of both the paralyzed tissue and normal tissue. The sensitivity for identifying normal tissue, lesion tissue, and tissue adjacent to the lesion was 71.9%, 100%, and 64.5%, respectively, while the specificity was 85.7%, 93.6%, and 89.1%, respectively. The positive predictive values were 71.9%, 88.9%, and 74.1%, respectively, and the negative predictive values were 85.7%, 100%, and 83.8%, respectively. The overall accuracy of the model was 72.8%. Following a two-week washout period, the IOR obtained by comparing each evaluator’s initial and subsequent judgments was 0.789±0.035. The feature that showed the highest level of consistency was curling of the alveolar walls [0.75, 95% confidence interval (CI): 0.687–0.814], while the feature that exhibited the lowest level of consistency was elastin fragment-like accumulation (0.574, 95% CI: 0.51–0.638). Details are presented in Table 3.

Discussion

In this study, we identified comprehensive CLE imaging features of malignant pulmonary nodules by observing lung specimens from surgical procedures, and established a preliminary CLE feature map of peripheral malignant pulmonary nodules.

The present study found that the CLE imaging features of malignant pulmonary nodules could be grouped into three overarching categories: proliferation, deformation, and destruction. These features have been found to be significantly associated with malignancy (10,14). A total of eight imaging features were identified, all of which were associated with changes in elastic fibers and alterations in alveolar structure. The results of pathophysiological studies on elastin in lung cancer have reached analogous conclusions (6-9). Research has shown that lung cancer tissues exhibiting adherent growth patterns may either retain normal elastin, or display elastin thickening without the obvious destruction of elastic fibers (10-11). The destruction and degradation of elastic fibers primarily occur in the central regions of invasive-growing lung adenocarcinomas, indicating that in the central regions of invasive tumors, the structure of elastin is disrupted or replaced by tumor cells. Consequently, this disruption of the elastic fiber framework is associated with invasion and a poor prognosis (14). Considering the observed heterogeneity among tumors, a solitary lesion may comprise elements exhibiting diverse growth patterns, including adherent, micro-papillary, and solid components. Consequently, CLE images may show a variety of features, including mild alveolar thickening with intact structure, extensive alveolar structural destruction, and elastic fiber rupture, all of which may occur simultaneously.

In the absence of contrast agents, CLE imaging largely relies on the spontaneous fluorescence of elastin components in alveolar structures. Two distinct types of elastic fibers have been identified in alveolar structures (i.e., thin and thick elastic fibers). The primary framework of alveolar structures is formed by thick elastic fibers, which create alveolar pores and the sides of polygonal alveoli. Ultrafine elastic fibers may branch out from thick elastic fibers that cross the alveolar walls, thus supporting the alveolar walls (15,16). The normal alveolar structures in CLE images have been identified (1). Normal alveoli exhibit regular alveolar structures with fine, continuous, and intact elastic fibers, and the presence of inflammatory cells such as macrophages (17). Lung cancer lesions may show alveolar wall thickening, alveolar oedema, and macrophage aggregation (18).

Tissues with a high concentration of elastin components have been observed to emit spontaneous fluorescence (520 nm) when stimulated by laser light, enabling the direct imaging of alveoli and tracheal walls using CLE (19). However, tumor tissues lacking elastin components do not emit spontaneous fluorescence. Consequently, the local administration of acridine yellow or the intravenous administration of sodium fluorescein is necessary to enhance contrast and facilitate CLE imaging. However, the use of acridine yellow in human subjects raises ethical issues, primarily due to its propensity to inflict damage to DNA. Consequently, contemporary studies predominantly use sodium fluorescein (12,20-22). Following the intravenous injection of sodium fluorescein, needle-based CLE imaging can be used to identify malignant tumors and metastatic lymph nodes, provide real-time feedback on needle placement before biopsy procedures, thus improving diagnostic accuracy, and assess lesion characteristics during surgery (23).

However, the use of sodium fluorescein has been associated with adverse reactions. Notably, 1.24–17.65% of patients experience mild adverse reactions, including urticaria, nausea, vomiting, abnormal sensations in the tongue and lips, and abdominal pain, while 0.2–6% experience moderate adverse reactions, including rash, syncope, local tissue necrosis, and hemolytic anemia (24,25). To mitigate the risks associated with this process, this study used CLE imaging without fluorescent contrast agents. This approach avoids potential adverse reactions and eliminates the influence of contrast agents on the intraoperative imaging.

This study used surgical resection specimens for in vitro research. Surgical resection specimens invariably exhibit mild iatrogenic alveolar collapse in the affected area, due, at least in part, to the loss of lung tissue aeration during the resection process. This loss leads to the contraction of the elastic fibers, resulting in the apposition of the alveolar walls, which in turn leads to an artificial collapse. Furthermore, tactile manipulations by surgeons during lesion exploration can cause localized collapse. In malignant lesions, alveolar collapse is primarily the result of a concentration of elastic fibers in the affected area, resulting in a reduced alveolar cavity space.

Two benign lesions were found in the present study. In the inflammatory lesion, CLE imaging revealed the presence of inflammatory cells, devoid of alveolar structures. A thorough pathological examination under the microscope revealed extensive necrosis, the presence of histiocytes, chronic inflammatory cell infiltration, focal calcification, and the absence of malignant epithelial components. This finding is consistent with the CLE imaging results. In the other case of bronchial adenoma, CLE imaging revealed features such as hazy shadows and alveolar collapse, which are characteristics of malignant lesions. It is hypothesized that the hazy shadows and alveolar collapse are primarily due to the pathological characteristics of bronchial adenoma (26-29).

The main limitation of this study is the small number of benign lesions. Most nodular lesions removed via surgical resection are malignant, and in this study, only two benign lesions were removed. To avoid bias caused by differences in data volume, no diagnostic analysis was conducted between the benign and malignant lesions. Based on the current evidence, this method remains in the experimental research phase and has not yet been widely adopted in routine clinical practice. We intend to incorporate data on more benign lesions, small cell tumors, squamous cell carcinomas, and other lesions in the future to further analyze CLE images.

Conclusions

In conclusion, this study showed that malignant pulmonary nodules may exhibit a variety of CLE imaging features. Typical CLE imaging features without fluorescent injection may aid in the identification of malignant lung nodules.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1805/rc

Data Sharing Statement: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1805/dss

Peer Review File: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1805/prf

Funding: This work was supported by the Noncommunicable Chronic Diseases-National Science and Technology Major Project (Nos. 2024ZD0529503 and 2024ZD052950) and the Military Scientific Research Project (No. 24BJZ08).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1805/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was approved by the Ethics Committee of Chinese PLA General Hospital (No. S2023-590-01) and was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Prior to the study, all patients were asked to provide individual informed consent in the form of a signed document.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Adams SJ, Stone E, Baldwin DR, et al. Lung cancer screening. Lancet 2023;401:390-408. [Crossref] [PubMed]
2. Folch EE, Pritchett MA, Nead MA, et al. Electromagnetic Navigation Bronchoscopy for Peripheral Pulmonary Lesions: One-Year Results of the Prospective, Multicenter NAVIGATE Study. J Thorac Oncol 2019;14:445-58. [Crossref] [PubMed]
3. Kramer T, Wijsman PC, Kalverda KA, et al. Advances in bronchoscopic optical coherence tomography and confocal laser endomicroscopy in pulmonary diseases. Curr Opin Pulm Med 2023;29:11-20. [Crossref] [PubMed]
4. Manley CJ, Kramer T, Kumar R, et al. Robotic bronchoscopic needle-based confocal laser endomicroscopy to diagnose peripheral lung nodules. Respirology 2023;28:475-83. [Crossref] [PubMed]
5. Wijmans L, Yared J, de Bruin DM, et al. Needle-based confocal laser endomicroscopy for real-time diagnosing and staging of lung cancer. Eur Respir J 2019;53:1801520. [Crossref] [PubMed]
6. Kramer T, Wijmans L, de Bruin M, et al. Bronchoscopic needle-based confocal laser endomicroscopy (nCLE) as a real-time detection tool for peripheral lung cancer. Thorax 2022;77:370-7. [Crossref] [PubMed]
7. Diaz-Churion F, Yu Lee-Mateus A, Abia-Trujillo D, et al. Real-time visualization of lung malignancy with needle-based confocal laser endomicroscopy during shape-sensing robotic-assisted bronchoscopy. Respirol Case Rep 2023;11:e01092. [Crossref] [PubMed]
8. Zuo CY, Xue KY, Wu XM, et al. Value of needle confocal laser microendoscopy combined with endobronchial ultrasound bronchoscopy in the diagnosis of hilar and mediastinal lymph node lesions. Kaohsiung J Med Sci 2023;39:936-42. [Crossref] [PubMed]
9. Seitlinger J, Stasiak F, Streit A, et al. Assessment of needle-based confocal laser endomicroscopy (nCLE) as a tool for real-time diagnosis of non-small cell lung cancer. J Thorac Dis 2024;16:4986-93. [Crossref] [PubMed]
10. Kramer T, Wijmans L, van Heumen S, et al. Needle-based confocal laser endomicroscopy for real-time granuloma detection. Respirology 2023;28:934-41. [Crossref] [PubMed]
11. Salaün M, Guisier F, Dominique S, et al. In vivo probe-based confocal laser endomicroscopy in chronic interstitial lung diseases: Specific descriptors and correlation with chest CT. Respirology 2019;24:783-91. [Crossref] [PubMed]
12. Zirlik S, Hildner K, Rieker RJ, et al. Confocal Laser Endomicroscopy for Diagnosing Malignant Pleural Effusions. Med Sci Monit 2018;24:5437-47. [Crossref] [PubMed]
13. Sawada T, Takizawa H, Aoyama M, et al. Diagnosis of visceral pleural invasion using confocal laser endomicroscopy during lung cancer surgery. J Thorac Dis 2021;13:4742-52. [Crossref] [PubMed]
14. Thunnissen E, Motoi N, Minami Y, et al. Elastin in pulmonary pathology: relevance in tumours with a lepidic or papillary appearance. A comprehensive understanding from a morphological viewpoint. Histopathology 2022;80:457-67. [Crossref] [PubMed]
15. Uehara T, Honda T, Sano K, et al. A three-dimensional analysis of blood vessels in bronchioloalveolar carcinoma. Lung 2004;182:343-53. [Crossref] [PubMed]
16. Honda T, Ishida K, Hayama M, et al. Type II pneumocytes are preferentially located along thick elastic fibers forming the framework of human alveoli. Anat Rec 2000;258:34-8. [Crossref] [PubMed]
17. Thiberville L, Salaün M, Lachkar S, et al. Human in vivo fluorescence microimaging of the alveolar ducts and sacs during bronchoscopy. Eur Respir J 2009;33:974-85. [Crossref] [PubMed]
18. Wellikoff AS, Holladay RC, Downie GH, et al. Comparison of in vivo probe-based confocal laser endomicroscopy with histopathology in lung cancer: A move toward optical biopsy. Respirology 2015;20:967-74. [Crossref] [PubMed]
19. Pauly JL, Allison EM, Hurley EL, et al. Fluorescent human lung macrophages analyzed by spectral confocal laser scanning microscopy and multispectral cytometry. Microsc Res Tech 2005;67:79-89. [Crossref] [PubMed]
20. Fuchs FS, Zirlik S, Hildner K, et al. Confocal laser endomicroscopy for diagnosing lung cancer in vivo. Eur Respir J 2013;41:1401-8. [Crossref] [PubMed]
21. Takemura M, Kurimoto N, Hoshikawa M, et al. Probe-based confocal laser endomicroscopy for rapid on-site evaluation of transbronchial biopsy specimens. Thorac Cancer 2019;10:1441-7. [Crossref] [PubMed]
22. Fuchs FS, Zirlik S, Hildner K, et al. Fluorescein-aided confocal laser endomicroscopy of the lung. Respiration 2011;81:32-8. [Crossref] [PubMed]
23. van Heumen S, Kramer T, Korevaar DA, et al. Bronchoscopy with and without needle-based confocal laser endomicroscopy for peripheral lung nodule diagnosis: protocol for a multicentre randomised controlled trial (CLEVER trial). BMJ Open 2024;14:e081148. [Crossref] [PubMed]
24. Kornblau IS, El-Annan JF. Adverse reactions to fluorescein angiography: A comprehensive review of the literature. Surv Ophthalmol 2019;64:679-93. [Crossref] [PubMed]
25. Yun D, Kim DK, Lee JP, et al. Can sodium fluorescein cause contrast-induced nephropathy? Nephrol Dial Transplant 2021;36:819-25. [Crossref] [PubMed]
26. Bo J, Chen X, Zhang T, et al. Clinicopathological features and genomic analysis of bronchiolar adenoma. Histol Histopathol 2023;38:1465-74. [Crossref] [PubMed]
27. Ruini C, Schlingmann S, Jonke Ž, et al. Machine Learning Based Prediction of Squamous Cell Carcinoma in Ex Vivo Confocal Laser Scanning Microscopy. Cancers (Basel) 2021;13:5522. [Crossref] [PubMed]
28. Șerbănescu MS, Streba L, Demetrian AD, et al. Transfer Learning-Based Integration of Dual Imaging Modalities for Enhanced Classification Accuracy in Confocal Laser Endomicroscopy of Lung Cancer. Cancers (Basel) 2025;17:611. [Crossref] [PubMed]
29. Gheorghe AG, Streba L, Demetrian AD, et al. Ex vivo use of probe-based confocal laser endomicroscopy in lung cancer visualization: an ex vivo pilot study. Rom J Morphol Embryol 2025;66:163-71. [Crossref] [PubMed]

(English Language Editor: L. Huleatt)