Arterial versus Venous Watershed Localization for Nonpalpable Pulmonary Nodules: A Retrospective Study

Summary

This study evaluates the watershed localization method (WALM) for intraoperative localization of nonpalpable pulmonary nodules. Both arterial and venous WALM using 3D reconstruction and indocyanine green (ICG) fluorescence are compared against CT-guided localization (CT-GL), showing shorter procedure time, fewer complications, and lower costs, but increased postoperative drainage and chest tube duration.

Abstract

Accurate localization of nonpalpable pulmonary nodules is critical for successful surgical resection. This study aimed to evaluate the feasibility and safety of the watershed localization method (WALM), a technique that uses three-dimensional reconstruction and temporary occlusion of watershed arteries or veins combined with indocyanine green (ICG) fluorescence staining for real-time intraoperative navigation. Clinical data were collected from patients who underwent video-assisted thoracoscopic pulmonary wedge resection with arterial-WALM (A-WALM), venous-WALM (V-WALM), or CT-guided localization (CT-GL) from October 2023 to October 2025. Patients were divided into three groups according to the localization method. Perioperative data were compared to evaluate localization efficacy. Both A-WALM and V-WALM were associated with shorter localization procedure time, lower incidence of localization-related complications, and reduced total hospitalization costs as compared with CT-GL. However, they were associated with greater postoperative drainage volume and longer chest tube indwelling time, with no significant differences in other indicators. When the pulmonary fissure is completely developed, the A-WALM approach may be better suited for nodules located in the posterior segment of the right upper lobe, right middle lobe, apicoposterior segment of the left upper lobe, or dorsal and anteromedial basal segments of either lower lobe; V-WALM may be preferable for nodules located in the anterior and lingular segments of the left lung and the posterolateral basal segment of both lungs, while either A-WALM or V-WALM is suitable for nodules in the apical and anterior segments of the right upper lobe. When the pulmonary fissure is incompletely developed, V-WALM may be preferred for localizing nodules in the right middle lobe and apicoposterior segment of the left upper lobe. In conclusion, both A-WALM and V-WALM appear to be feasible intraoperative localization alternatives to CT-GL in selected patients with nonpalpable pulmonary nodules, although they are associated with increased postoperative drainage and longer chest tube duration.

Introduction

With the increasingly widespread application of low-dose computed tomography (LDCT) in the early screening and diagnosis of pulmonary nodules, the detection rate of early-stage lung cancer has been significantly improved¹. Pulmonary wedge resection constitutes one of the most commonly employed surgical procedures for early-stage lung cancer, and the accurate localization of pulmonary nodules directly determines the success of the operation². Currently, CT-guided localization (CT-GL) is the commonly used clinical localization method. However, it is associated with various complications, including dislodgement, pneumothorax, hemothorax, pleural reactions, and even rare but life-threatening events such as air embolism. Furthermore, the success rate of CT-GL is strongly influenced by the location of pulmonary nodules; for example, the puncture path is often obstructed by structures such as the great vessels of the heart, scapulae, and ribs. Other methods, such as electromagnetic bronchoscopic navigation localization, are difficult to popularize due to drawbacks such as high cost and complex operation^3,4. Thus, exploring a simple, efficient localization method with minimal complications remains one of the urgent issues to be addressed in the management of early-stage lung cancer.

The watershed localization method (WALM) is a fluorescence-based technique that uses the concept of pulmonary vascular territories. The boundary between two adjacent vascular territories is termed the "watershed." By temporarily occluding the segmental artery or vein supplying the nodule-bearing area and injecting indocyanine green (ICG) intravenously, the occluded territory remains unstained while the surrounding lung fluoresces, creating a visible boundary on the lung surface. This allows precise wedge resection without preoperative puncture, avoiding radiation exposure and puncture-related complications. Against this backdrop, Chu et al. proposed the WALM, which uses three-dimensional (3D) reconstruction technology for pulmonary nodule localization⁵. This method enables clear visualization of key structures, including pulmonary segments, pulmonary arteries, pulmonary veins, and bronchi; facilitates real-time intraoperative localization; preserves more healthy lung tissues; and reduces the physical, mental, and economic burdens on patients⁵. Nevertheless, there is currently no consensus on whether WALM should be implemented as the arterial-WALM (A-WALM) or venous-WALM (V-WALM)^6,7.

Therefore, this study aimed to further elucidate the advantages of WALM over CT-GL and to systematically analyze the strengths, limitations, and optimal selection criteria for A-WALM and V-WALM in the intraoperative localization of pulmonary nodules. The completion of this study will be of significant value in formulating personalized localization and surgical plans for pulmonary nodules and will provide a further theoretical basis for the application of WALM in intraoperative localization of pulmonary nodules.

Protocol

This retrospective study was conducted in accordance with the Declaration of Helsinki (revised in 2013) and was approved by the Medical Ethics Committee of The Affiliated People’s Hospital of Ningbo University (Approval Number: 2025-135). All participants provided written informed consent prior to enrollment. The study was registered with the Chinese Clinical Trial Registry (Registration Number: ChiCTR2500115862). All procedures involving human participants were performed in compliance with institutional guidelines.

Study design and patient selection
Clinical data were collected from patients who underwent video-assisted thoracoscopic pulmonary wedge resection with A-WALM, V-WALM, or CT-GL localization at the Department of Cardiothoracic Surgery, The Affiliated People's Hospital of Ningbo University, between October 1, 2023, and October 1, 2025. The inclusion criteria were as follows: (1) a pulmonary nodule measuring 8–20 mm in diameter located within the peripheral one-third of the lung field; (2) preoperative biopsy demonstrating preinvasive lesions or a ground-glass opacity (GGO) component ≥ 50%; (3) availability of high-quality chest computed tomography (CT) images suitable for accurate three-dimensional (3D) reconstruction; and (4) a clear indication for wedge resection without surgical contraindications. The exclusion criteria were: (1) a history of pneumoconiosis, tuberculosis, chronic obstructive pulmonary disease (COPD), or other significant pulmonary comorbidities; (2) allergy to iodine contrast agents; (3) a requirement for segmentectomy or lobectomy; (4) previous ipsilateral lung surgery; or (5) inability to tolerate surgery.

The patients were assigned to the A-WALM, V-WALM, or CT-GL group according to the localization method used. Propensity score matching (PSM) was performed to minimize potential confounding bias. A total of 148 patients were included in the study. After PSM, the A-WALM, V-WALM, and CT-GL groups comprised 48, 53, and 47 patients, respectively (Table 1). All covariates achieved adequate balance after matching, with standardized mean differences (SMDs) below 0.1 for all variables (P > 0.05; Table 1).

The choice of localization method (A-WALM, V-WALM, or CT-GL) was determined jointly by the surgical team based on preoperative 3D reconstruction findings, nodule location, pulmonary fissure completeness, vascular anatomy, and the feasibility of CT-guided puncture. The final decision was reached through multidisciplinary discussion and documented in the medical records. Surgeon preference and patient willingness were also considered, and all decisions were made before surgery.

Preoperative 3D reconstruction
Thin-section chest CT images (slice thickness, 1 mm) were acquired during end-inspiratory breath-hold. The Digital Imaging and Communications in Medicine (DICOM) data were imported into the Imaging Analysis System. Three-dimensional (3D) models were generated using automatic segmentation followed by manual correction to reconstruct the pulmonary arteries, veins, and bronchial tree. In cases with anatomical variations, the preoperative 3D model was used to guide surgical dissection planning. The segmental pulmonary artery within the nodule's watershed region was designated as the target vessel for A-WALM, whereas the corresponding segmental pulmonary vein was selected as the target vessel for V-WALM. Watershed regions were reconstructed to simulate reverse staining on the lung surface following peripheral indocyanine green (ICG) injection after temporary occlusion of the target vessel (Figure 1A1,A2,B1,B2).

CT-Guided Localization (CT-GL)
Pulmonary nodules in the CT-GL group were localized preoperatively using repeated CT-guided fine-needle puncture (Figure 1C1–C4).

Surgical procedure
Video-assisted thoracoscopic surgery was performed using the 3D fluorescence thoracoscopy system. The pre-identified watershed vessels were dissected and isolated, after which the target vessel was temporarily occluded using either a bulldog clamp or a 7-0 silk suture. Within 10 s of vessel occlusion, 3 mL of ICG solution (2.5 mg/mL; total dose, 7.5 mg) was injected through a peripheral vein. The boundary between the stained and unstained lung tissue was then marked with an electrosurgical knife. The vessel occlusion was typically maintained for 1–2 min. If the fluorescence staining was inadequate, the clamp was released for 3–5 min before repeating the ICG injection, with a maximum of two repeat injections. After the boundary had been marked, the clamp or suture was removed.

Wedge resection was subsequently performed using a linear stapler while ensuring adequate surgical margins (Figure 1A3–A5,B3–B5). In the CT-GL group, wedge resection was performed according to the preoperative localization needle position (Figure 1C5).

Outcome measures
The primary localization-related outcomes included technical localization success, initial wedge resection success, and the final R0 resection rate. Technical localization success was defined as successful completion of the assigned localization procedure without immediate failure requiring abandonment of the assigned method, including successful vessel occlusion with adequate fluorescence staining for WALM or successful marker placement at the target site for CT-GL. Initial wedge resection success was defined as achieving an R0 resection (microscopically negative margins) following the first wedge resection without requiring additional resection, and corresponded to the localization success rate reported as the primary outcome of this study. The final R0 resection rate was defined as the achievement of R0 resection after all remedial procedures, including extended wedge resection, conversion to segmental localization, or repeat puncture localization, when the initial wedge resection was unsuccessful.

Secondary outcomes included localization procedure time (defined as the duration from initial vessel dissection to complete boundary visualization for WALM and from the start of CT scanning to successful marker placement for CT-GL), total postoperative drainage volume, chest tube indwelling time, surgical margin distance, hospital stay, total hospitalization costs (including localization-related expenses, surgical fees, and postoperative care), localization-associated complications (pain, pneumothorax, marker displacement, hemorrhage, and prolonged air leak lasting > 5 days), changes in localization strategy due to intraoperative difficulty, vessel identification difficulty (rated by the operating surgeon as easy, moderate, or difficult) and its association with incomplete pulmonary fissures, and pathological findings.

Statistical analysis
Continuous variables were first evaluated for normality using the Shapiro–Wilk test. Variables meeting the assumption of normality are presented as mean ± standard deviation (SD) and were analyzed by one-way analysis of variance (ANOVA), with Bonferroni-adjusted post hoc comparisons applied where appropriate. Variables that were not normally distributed were compared using the Kruskal–Wallis H test, followed by pairwise Mann–Whitney U tests with Bonferroni correction. Differences in categorical variables were assessed using the χ² test. To determine whether the primary findings remained independent of important clinical factors, additional multivariable analyses were performed using generalized linear models and logistic regression, incorporating nodule-to-pleura distance, nodule location, and operating surgeon as covariates.

Given the retrospective, non-randomized study design, propensity score matching (PSM) was used to improve the comparability of the three treatment groups. A logistic regression model was constructed to generate propensity scores, with localization technique (A-WALM, V-WALM, or CT-GL) specified as the dependent variable and age, sex, body mass index, smoking history, nodule diameter, nodule-to-pleura distance, nodule location, pulmonary fissure completeness, and Charlson Comorbidity Index entered as matching variables. Patients were matched in a 1:1 ratio using nearest-neighbor matching without replacement, with a caliper width equal to 0.2 times the standard deviation of the logit-transformed propensity score. The success of the matching procedure was evaluated using standardized mean differences (SMDs), with values below 0.1 indicating satisfactory balance between groups. The number of patients included before and after matching was recorded for each study group.

Results

Demographic characteristics and nodule distribution among the groups
There were no significant differences in clinical characteristics, pulmonary nodule features, or nodule distributions among the three groups. All the enrolled patients successfully underwent wedge resection for solitary pulmonary nodules with adequate surgical margins (≥2 cm or ≥maximum tumor diameter). No ICG-related complications or perioperative mortality were observed. All 148 patients ultimately achieved final R0 resection after remedial procedures when needed, yielding a final R0 resection rate of 100%.

Patient characteristics	A-WALM (n=48)	V-WALM (n=53)	CT-GL (n=47)	χ2/F-value	P-value
Sex [n (%)]				0.824	0.662
Male	21 (43.75)	19 (35.85)	17 (36.17)
Female	27 (56.25)	34 (64.15)	30 (63.83)
Age (years)	52.74 ± 12.16	56.49 ± 9.43	54.98 ± 10.20	1.927	0.149
Nodule diameter (mm)	12.89 ± 2.67	13.42 ± 2.31	13.45 ± 2.54	0.778	0.461
Disease duration (month)	19.36 ± 6.58	17.20 ± 5.63	16.99 ± 5.42	2.986	0.054
Distance to pleura (mm)	13.29 ± 1.29	13.52 ± 1.31	13.19 ± 1.52	0.714	0.492
GGO component (%)	61.68 ± 6.23	60.99 ± 6.39	60.92 ± 6.37	0.009	0.991
Degree of pulmonary fissure development [n (%)]				0.973	0.192
Incomplete	12 (25.00)	17 (32.08)	14 (29.79)
complete	36 (75.00)	36 (67.92)	33 (70.21)
Nodule location [n (%)]				6.546	0.586
RUL	5 (10.42)	8 (15.09)	7 (14.89)
RML	8 (16.67)	11 (20.76)	6 (12.77)
RLL	13 (27.08)	12 (22.64)	11 (23.40)
LUL	9 (18.75)	16 (30.19)	14 (29.79)
LLL	13 (27.08)	6 (11.32)	9 (19.15)

Table 1: Baseline demographic and clinical characteristics of the propensity score-matched study population. Data are expressed as mean ± SD or n (%), as appropriate. Comparisons of baseline variables showed no statistically significant differences among the three groups (all P > 0.05). Covariate balance following propensity score matching was evaluated using standardized mean differences (SMDs), with values <0.1 indicating satisfactory matching. Abbreviations: GGO = ground-glass opacity; RUL = right upper lobe; RML = right middle lobe; RLL = right lower lobe; LUL = left upper lobe; LLL = left lower lobe; SD = standard deviation; SMD = standardized mean difference.

Comparison of perioperative quantitative parameters among the groups
Comparative analysis revealed that both the A-WALM and V-WALM groups exhibited significantly shorter localization times, lower overall complication rates, and reduced total hospitalization costs compared with the CT-GL group (P < 0.05, Table 2). However, the A-WALM and V-WALM groups showed greater postoperative drainage volumes and longer chest tube indwelling times (P < 0.05). When directly comparing the A-WALM and V-WALM groups, the A-WALM group demonstrated shorter localization times but higher drainage volumes, longer chest tube indwelling times, and increased complication rates (P < 0.05), whereas no significant difference was observed in total hospitalization costs (P > 0.05). No significant differences were detected among the three groups in terms of the rate of successful R0 resection with the initial wedge resection (localization success rate), total hospital stay duration, margin distance of tumor resection, or pathological outcomes (P > 0.05). When complications were categorized by timing, all CT-GL-related complications were preoperative puncture-related events, whereas no such preoperative events occurred in the WALM groups. Intraoperative complications occurred exclusively in the WALM groups. Postoperative complications, including prolonged air leak and hemorrhage, were observed across all three groups, with higher rates in the WALM groups (see Table 3 for details). The results of the multivariate analysis, adjusted for nodule-to-pleura distance, lobar location, and operating surgeon, confirmed the findings from the univariate comparisons. The localization method remained a significant independent factor associated with both localization procedure time and total postoperative drainage volume (P < 0.001 for both). The significant inter-group differences (A-WALM vs. V-WALM vs. CT-GL) identified in the univariate analysis persisted following adjustment for these potential confounders. Notably, the key findings remained consistent after accounting for potential confounders in the multivariate logistic regression model (see Supplemental Table S1).

Parameter	A-WALM ( (n=48)	V-WALM (n=53)	CT-GL (n=47)	χ2/F-value/H-value	P-value
Successful localization [n (%)]				1.177	0.555
Yes	45 (93.75)	48 (90.57)	41 (87.23)
No	3 (6.25)	5 (9.43)	6 (12.77)
Localization procedure time (min)	17.47 ± 1.69*&	20.45 ± 1.21#	27.89 ± 5.07	6.253	<0.001
Postoperative drainage volume (ml)	137.5 (8)*&	118 (8.5)#	103 ( 14)	121.782	<0.001
Chest tube indwelling times (days)	2.13 ± 0.42*&	1.89 ± 0.19#	1.73 ± 0.15	28.810	<0.001
Length of hospital stay (days)	3.86 ± 0.28	3.77 ± 0.27	3.56 ± 0.28	0.007	0.993
Localization-associated complications [n (%)]				6.182	0.033
Present	8 (16.66)*&	5 (9.43)#	13 (27.66)
Absent	42 (83.34)	48 (90.57)	34 (72.34)
Total hospitalization costs (CNY)	20085.04± 1243.37&	20126.44± 1182.14#	22562.98± 1396.95	59.177	<0.001
Margin distance of tumor resection (cm)	2.13 ± 0.76	2.09 ± 0.87	2.16 ± 0.52	0.010	0.529
Final pathological diagnosis [n (%)]				6.274	0.617
Benign lesions	2 (4.17)	4 (7.54)	1 (2.13)
Atypical adenomatous hyperplasia	1 (2.08)	2 (3.77)	1 (2.13)
Adenocarcinoma in situ	18 (37.50)	10 (18.87)	12 (25.53)
Minimally invasive adenocarcinoma	15 (31.25)	20 (37.74)	17 (36.17)
Invasive adenocarcinoma	12 (25.00)	17 (32.08)	16 (34.04)

Table 2: Perioperative outcomes of wedge resections among the groups. Successful localization was uniformly defined for all groups as achieving an R0 resection (microscopically negative margins) with the initial wedge resection, without requiring any additional resection (i.e., initial wedge resection success). Cases requiring remedial procedures were counted as failures for this primary outcome but were included in the intention-to-treat analysis for all other perioperative outcomes. The localization procedure time was measured as the duration from the initial dissection of the watershed vessel to the complete visualization of target tissue boundaries for A-WALM/V-WALM and from the commencement of CT-guided scanning to the successful placement of the marker for CT-GL. It is important to note that these measurements represent different phases of the clinical workflow. The time for WALM reflects a purely intraoperative step, whereas the time for CT-GL encompasses the entire preoperative localization procedure, including patient positioning and scanning. Therefore, a direct comparison of these numerical values should be interpreted with caution. The observed shorter recorded times for WALM reflect a workflow advantage of intraoperative localization rather than superior technical speed, as the CT-GL time includes additional preoperative steps. Statistical significance is indicated as follows: "*" denotes significant differences between A-WALM and V-WALM (P < 0.05, Bonferroni-adjusted); "#" indicates significant differences between A-WALM and CT-GL (P < 0.05, Bonferroni-adjusted); "&" marks significant differences between V-WALM and CT-GL (P < 0.05, Bonferroni-adjusted). Abbreviations: A-WALM = Arterial watershed localization method; CT = computed tomography; ICG = indocyanine green; V-WALM = Venous watershed localization method; CT-GL = CT-guided localization.

Group	Cases (n)	Preoperative complications			Intra-/Postoperative complications		Overall complication
		Pain	Pneumothorax	Marker displacement	Hemorrhage	Prolonged air leak
A-WALM	48	0	0	0	4 (8.33)	4 (8.33)	8 (16.66)
V-WALM	53	0	0	0	2 (3.76)	3 (5.67)	5 (9.43)
CT-GL	47	6 (12.77)	2 (4.26)	2 (4.26)	1 (2.12)	2 (4.26)	13 (27.67)

Table 3: Incidence of localization-related complications in thoracoscopic wedge resections among the groups [n (%)]. Complications are categorized by timing: (1) Preoperative complications (exclusive to CT-GL): pain, pneumothorax, and marker displacement; (2) Intraoperative complications: bleeding and marker dislodgement; (3) Postoperative complications (occurring in any group): prolonged air leak (＞5 days) and hemorrhage (drainage volume ≥200 mL on postoperative day 1). Prolonged air leak: prolonged air leakage ＞ 5 days; Postoperative hemorrhage: drainage volume ≥ 200 mL on the first postoperative day. Preoperative complications are specific to the CT-GL procedure. Intra-/postoperative complications (Postoperative hemorrhage: drainage volume ≥ 200 mL on the first postoperative day; Prolonged air leak ＞ 5 days) could occur in any of the groups. The "overall complication rate" is provided as a summary reference for the total burden of adverse events associated with each localization strategy. However, because the timing and nature of complications differ fundamentally between CT-GL and WALM, direct comparison of overall rates is not appropriate for safety assessment. Readers are advised to refer to the specific complication profiles to meaningfully compare the distinct risk patterns between the techniques. Abbreviations: A-WALM = Arterial watershed localization method; CT = computed tomography; ICG = indocyanine green; V-WALM = Venous watershed localization method; CT-GL = CT-guided localization.

Supplemental Table S1. Multivariable generalized linear models and logistic regression analyses for key perioperative endpoints in A-WALM, V-WALM, and CT-GL groups. Note: (1) Model type & sample size alignment: continuous outcomes (localization procedure time, postoperative drainage volume, total hospitalization costs, chest tube indwelling time ) use generalized linear models; binary outcomes (localization success, localization-related complications) use logistic regression. (2) Reference groups: "Group" uses CT-GL as the reference; "Location of the nodule" uses the right upper lobe as the reference; "Surgeon" uses surgeon 1 as the reference. For binary outcomes, the reference category represents the "baseline state" (e.g., "Failure" for localization success). (3) Parameter interpretation: for continuous outcomes, "Coefficient" indicates the change in the endpoint per 1-unit increase in the independent variable (or vs. reference group for categorical variables). For binary outcomes, an OR > 1 indicates increased odds of the endpoint, while an OR < 1 indicates decreased odds. (4) Model validity: generalized linear models meet assumptions (linearity, homoscedasticity) via residual analysis; logistic regression shows good fit via Hosmer-Lemeshow test (success rate: χ² = 5.28, P = 0.721; complications: χ² = 4.85, P = 0.774). VIF < 1.5 for all variables, confirming no severe multicollinearity. Abbreviations: A-WALM = Arterial watershed localization method; CT = computed tomography; V-WALM = Venous watershed localization method; CT-GL = CT-guided localization; OR = odds ratio; CI = confidence interval; VIF = variance inflation factor. Please click here to download this file.

Guidance for selecting A-WALM or V-WALM in pulmonary nodule wedge resection (Table 4)
The preferential principles for selecting between A-WALM and V-WALM were formulated based on a comprehensive evaluation of several perioperative metrics, including the feasibility of vessel identification on preoperative 3D reconstruction, intraoperative difficulty of vascular exposure, duration of fluorescent staining, and postoperative outcomes such as drainage volume. These principles also account for the inherent anatomical characteristics of the segmental vasculature and the degree of pulmonary fissure completeness, which critically influence the technical feasibility and risk profile of each method.

In cases with complete pulmonary fissures, A-WALM may be considered for the nodules located in (1) the posterior segment of the right upper lobe; (2) the right middle lobe; (3) the apicoposterior segment of the left upper lobe; (4) the dorsal segment of both lower lobes; (5) the anteromedial basal segments of left lower lobes; and (6) the medial and anterior basal segments of the right lower lobe. Conversely, V-WALM is recommended for nodules located in (1) the anterior segment and lingular segment of the left upper lobe, and (2) the posterior and lateral basal segments of both lungs. For nodules situated in the apical and anterior segments of the right upper lobe, either A-WALM or V-WALM may be employed with comparable efficacy. In cases of incomplete pulmonary fissures, V-WALM may be the preferred approach for nodules in the right middle lobe and the apicoposterior segment of the left upper lobe, while the selection criteria for other segments remain consistent with those for complete fissures. This stratified approach accounts for both vascular accessibility and technical feasibility under different anatomical conditions.

Nodule distribution in pulmonary segments			A-WALM	V-WALM		A-WALM	V-WALM
			Complete pulmonary fissure			Incomplete pulmonary fissure
RUL	Apical segment			△			△
	Posterior segment		√			√
	Anterior segment			△			△
RML	Lateral segment		√				√
	Medial segment		√				√
RLL	Dorsal segment		√			√
	Medial basal segment		√			√
	Anterior basal segment		√			√
	Lateral basal segment			√			√
	Posterior basal segment			√			√
LUL	Apicoposterior segment		√				√
	Anterior segment			√			√
	Superior lingular segment			√			√
	Inferior lingular segment			√			√
LLL	Dorsal segment		√			√
	Anteromedial basal segment		√			√
	Lateral basal segment			√			√
	Posterior basal segment			√			√

Table 4: Exploratory guidance on the selection of A-WALM and V-WALM in thoracoscopic lung wedge resection (based on anatomical reasoning and institutional experience). "√" indicates the preferred localization method; "△" indicates no significant difference between the two localization methods. These recommendations are exploratory and derived from anatomical reasoning and institutional experience rather than from subgroup analysis of segment-specific outcome data. They are intended to provide practical guidance for clinical decision-making and require prospective validation in future studies. For incomplete fissure cases, V-WALM is recommended only when the relevant segmental vein can be safely identified and occluded; if venous dissection is also challenging, alternative strategies should be considered. Abbreviations: A-WALM = Arterial watershed localization method; V-WALM = Venous watershed localization method.

Data Availability
The datasets supporting the conclusions of this article are included within the article and in Supplemental Table S1 and Supplemental Table S2.

Figure 1. Localization diagrams of A-WALM, V-WALM, and CT-GL. (A1–A5) A-WALM: (A1) Preoperative chest CT showing nodule location and planned resection margin. (A2) 3D reconstruction identifying the target watershed artery and the lung segment to be resected. (A3) Intraoperative dissection and temporary occlusion of the target artery. (A4) Fluorescence visualization and marking of the tissue boundary after ICG injection. (A5) Wedge resection completed and the artery released. (B1–B5) V-WALM: (B1) Preoperative chest CT showing nodule location and planned resection margin. (B2) 3D reconstruction identifying the target watershed vein and the lung segment to be resected. (B3) Intraoperative dissection and temporary occlusion of the target vein. (B4) Fluorescence visualization and marking of the tissue boundary after ICG injection. (B5) Wedge resection completed and the vein released. (C1–C5) CT-GL: (C1) Preoperative chest CT showing nodule location. (C2) Pre-procedural planning of puncture needle depth and angle. (C3) Advancement of the guide needle to the intercostal space adjacent to the nodule. (C4) Deployment of the puncture needle at the target site. (C5) Intraoperative view of the puncture needle within the lung parenchyma. Abbreviations: A-WALM = Arterial watershed localization method; CT = computed tomography; ICG = indocyanine green; V-WALM = Venous watershed localization method; CT-GL = CT-guided localization. Please click here to view a larger version of this figure.

Supplemental Table S2. Raw data supporting conclusions of this study. Please click here to download this file.

Discussion

The widespread use of low-dose CT has increased early-stage lung cancer detection, with GGO as a predominant imaging finding^8,9. The 3D reconstruction technology has now been widely applied in preoperative planning for lung cancer treatment, providing high-quality digital models as references for pulmonary wedge resection^10,11,12. In addition, the method of visualizing the target pulmonary segment plane by blocking pulmonary vascular blood flow has been widely proven to be safe and reliable^13,14. WALM integrates 3D reconstruction technology with the watershed vessel occlusion principle for intersegmental plane identification. By preoperatively reconstructing the relationship between the watershed vessel staining area and nodule location, this technique temporarily occludes the watershed vessel during surgery to achieve real-time intraoperative localization, visualize the lung segment containing the nodule, and complete wedge resection of the target lung tissue. A-WALM functions on the principle of occluding arterial blood flow, which induces a loss of blood perfusion in the watershed lung tissue. Since ICG distributes in a blood flow–dependent manner, its ability to reach the distal vascular bed is impaired under such conditions. This results in a local "fluorescence defect" that enables retrograde staining of the watershed lung tissue¹⁵.

In contrast, V-WALM modulates blood flow pressure to impede venous return within the occluded region, leading to capillary pressure exceeding segmental arterial pressure. This physiological change prevents ICG from entering the local pleural surface vessels, thereby also achieving retrograde staining of the watershed lung tissue¹⁶. Thus, occlusion of either the target artery or vein can clearly visualize the plane of the target lung tissue. For readers without a clinical background, the surgical workflow is as follows: preoperative 3D reconstruction identifies the target vessel; intraoperatively, the vessel is temporarily occluded; ICG is injected intravenously; the boundary between fluorescent and non-fluorescent lung is marked; the vessel is released; and wedge resection is performed along the marked line. The entire localization step takes approximately 1–2 min. However, there is currently a lack of systematic evaluation or analysis of the selection between A-WALM and V-WALM in clinical practice.

This study explored the advantages and disadvantages of different localization methods in 148 patients who underwent pulmonary wedge resection. The results showed that, compared with CT-GL, both A-WALM and V-WALM were associated with shorter localization times, a lower incidence of localization-related complications, and reduced total hospitalization costs. However, WALM was associated with increased postoperative drainage volumes and longer chest tube indwelling time. This may be attributed to the additional vascular dissection required for WALM during localization, which increases the risk of hemorrhage and air leakage, thereby delaying tube removal. No significant differences were observed between WALM and CT-GL in localization success rate, tumor resection margin distance, hospitalization duration, or the distribution of postoperative pathological results. This indicates that, similar to traditional CT-GL, WALM can achieve precise resection of pulmonary nodules without compromising postoperative pathological evaluation, highlighting its potential for nodule localization.

In addition, three patients in the CT-GL group did not undergo CT-GL because their nodules were obscured by the great cardiac vessels, ribs, and scapulae. Instead, WALM localization was adopted, further supporting the applicability of the watershed analysis localization method to the localization of pulmonary nodules in special positions. This study also revealed that A-WALM required a shorter localization time than V-WALM, which may be attributed to the complex and variable branching patterns of veins, making them more difficult to identify. However, compared with V-WALM, A-WALM was associated with a greater postoperative drainage volume, a longer chest tube indwelling time, and a higher incidence of localization-related complications. These differences likely reflect the increased risks of bleeding and air leakage during A-WALM procedures. There were no significant differences in hospital stay duration or total hospitalization costs between the two approaches. Notably, while A-WALM consistently produced a dye-stained area that persisted for approximately 3–4 min, V-WALM produced a stable stain for only 1–2 min. Nevertheless, this 1–2-min window was sufficient for complete margin demarcation, a finding consistent with previous reports¹⁷.

A total of 14 cases of failed initial wedge resection were observed across the three groups in this study, including 3 cases in the A-WALM group, 5 cases in the V-WALM group, and 6 cases in the CT-GL group. The main causes of technical localization failure were as follows: anatomical variations of segmental pulmonary vessels (1 case in the A-WALM group and 2 cases in the V-WALM group), poor ICG fluorescence staining (2 cases in the A-WALM group and 3 cases in the V-WALM group), special nodule location obscured by cardiac great vessels/bony structures (3 cases in the CT-GL group), and localization needle displacement/dislodgement (3 cases in the CT-GL group). All failed cases received timely intraoperative remedial measures, including extending the wedge resection scope, converting to segmental localization, or performing repeated puncture localization. Final R0 resection was successfully achieved in all 148 patients. For the calculation of the primary outcome (initial wedge resection success rate), the 14 failed cases were included in the denominator as failures.

For secondary outcomes (e.g., localization procedure time, postoperative drainage volume, hospital stay), all 148 patients were included in the analysis according to their assigned group (intention-to-treat basis), regardless of whether remedial measures were required. No severe postoperative complications occurred, and the prognosis did not differ significantly from that of the successful localization group. Complication analysis revealed distinct safety profiles between the techniques. Because the timing and nature of complications differ fundamentally between CT-GL and WALM, the overall complication rate should not be used as a direct measure for safety comparison. The higher overall complication rate with CT-GL stemmed primarily from preoperative risks (e.g., puncture-related pain, pneumothorax), inherent to percutaneous needle placement. In contrast, complications of A-WALM and V-WALM were exclusively intra- or post-operative, chiefly hemorrhage and prolonged air leak, resulting from hilar vessel dissection and occlusion. In addition, although both WALM techniques showed significantly shorter recorded times than CT-GL, this comparison requires careful interpretation due to fundamental differences in what was measured. The WALM time reflects a discrete intraoperative step (vessel dissection to boundary marking), while the CT-GL time encompasses the entire preoperative localization process (from CT scanning to marker deployment). Thus, the observed time advantage for WALM primarily reflects its streamlined, operating room–integrated workflow rather than solely faster technical execution. This integration eliminates the need for a separate preoperative procedure, potentially offering greater overall efficiency.

When selecting an appropriate pulmonary nodule localization method, the localization success rate is the primary consideration, as it directly determines surgical success and significantly affects patient recovery and medical experience. This study found no significant differences in localization success rates among A-WALM, V-WALM, and CT-GL, which may be attributed to the high-quality CT data, advanced 3D reconstruction technology, and proficient clinical expertise utilized in this study. For clinicians and medical institutions with varying practice levels, CT-GL may be a more appropriate choice if any of the following barriers exist: (1) failure to obtain high-resolution CT data; (2) inability to perform precise 3D reconstruction of pulmonary vessels; (3) difficulty in accurately localizing watershed vessels due to extensive vascular variations or limited technical proficiency. In addition, the pulmonary veins exhibit complex and variable branching patterns, whereas the pulmonary arteries often run parallel to the bronchi with fewer variations and more regular trajectories. Thus, A-WALM may be the preferred option for WALM localization when intraoperative identification of watershed pulmonary veins is challenging.

For complete pulmonary fissures, A-WALM is preferred for the right upper lobe posterior segment, right middle lobe, left upper lobe apicoposterior segment, and bilateral lower lobe dorsal and anteromedial basal segments. This preference is based on the clearer anatomical structure of arteries in these regions, their ease of exposure, longer visualization time after occlusion, and fewer branches or variations. In contrast, V-WALM may offer advantages for the left upper lobe anterior segment, left upper lobe lingular segment, and bilateral lateral-posterior basal segments, owing to the superficial distribution of the pulmonary veins in these areas. In the right upper lobe apical and anterior segments, both arteries and veins are superficial and easily exposed, with comparable surgical approaches and anatomical complexity, enabling the use of either A-WALM or V-WALM. In cases involving incomplete pulmonary fissures that require division, A-WALM may induce additional trauma, thereby rendering V-WALM—where superficial veins serve as watershed vessels—more suitable. For example, in the right middle lobe or left upper lobe apicoposterior segment with incomplete fissures, anatomical dissection and temporary occlusion of watershed veins are recommended. However, this advantage is not universal and depends on the specific nodule location and individual vascular anatomy. Therefore, our recommendation of V-WALM for incomplete fissure cases should be applied with the caveat that V-WALM is preferred only when the relevant segmental vein can be safely identified and occluded. If venous anatomy is also unfavorable, alternative strategies such as CT-GL should be considered. When nodules are located near the hilum, the deep anatomical location of the pulmonary arteries in this region complicates arterial dissection, elevating the risk of tissue injury and potential bleeding. In such cases, V-WALM may offer a lower-risk alternative by avoiding deep arterial dissection.

Although WALM demonstrated a high overall success rate, instances of localization failure in our cohort warrant a detailed analysis to inform future practice. The primary contributing factors can be categorized into the following three areas: complex venous drainage patterns, a mismatch between the watershed boundary and the nodule location, and the inherent technical learning curve.

This study has several limitations: (1) Long-term complications associated with WALM, such as venous thrombosis secondary to V-WALM or hilar anatomical disturbances resulting from A-WALM, as well as the long-term impacts of both methods on lung function, require further monitoring and follow-up. (2) Special anatomical variations or the presence of emphysema may interfere with the retrograde staining area of the target lung tissue; however, this was not evaluated in this study. (3) The proposed preferential principles for localization methods are exploratory and were formulated under the premise that all three localization methods are free of restrictive factors, without considering multiple objective variables. Thus, further research involving clinicians and medical institutions with varying levels of practice is warranted. (4) Given the procedural differences among the interventions, it was not possible to mask the physicians or patients, which may have introduced performance bias. Nevertheless, measurement bias was rigorously controlled by implementing masking of the outcome assessors. (5) This was a retrospective study without randomization, and the choice of localization method was influenced by nodule location, fissure completeness, vascular anatomy, and surgeon preference, which may have introduced selection bias. Although propensity score matching and multivariate analyses were performed to adjust for confounders, residual or unmeasured confounding cannot be entirely excluded. Future prospective randomized studies are needed to validate the comparative efficacy of the three localization methods. The development of novel methodologies in future research will be crucial for overcoming these methodological constraints.

WALM functions as an intuitive intraoperative guidance tool, complementing rather than replacing meticulous preoperative assessment. It is optimally indicated for wedge resection in carefully selected patients (e.g., those with small, predominantly ground-glass, or preinvasive nodules), providing real-time visualization to establish precise margins. However, given the retrospective nature of this study and the potential for selection bias, our findings should be interpreted with caution. The increased postoperative drainage and longer chest tube duration associated with WALM should also be considered when selecting the localization method. Therefore, WALM may be a feasible alternative to CT-GL in selected patients, but broader claims of superiority are not supported by the current data. Its potential role as an adjunct in anatomical segmentectomy warrants further investigation.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
7?0 silk suture	Ethicon (or local supplier)	Not available	Not available
Bulldog clamp	Medline (or local supplier)	Not available	Not available
Indocyanine green (ICG)	Generic (e.g., Yichuang Pharmaceutical)	Not available	Not available
Linear stapler	Ethicon or Medtronic	Not available	Not available
SPSS software	IBM Corp., Armonk, NY, USA	Version 26.0	RRID: SCR_002865
Storz IMAGE1 S 3D fluorescence system	Karl Storz SE&Co.KG, Tuttlingen, Germany	Not available	Not available
United Imaging Analysis System (3D reconstruction software)	United Imaging Healthcare, Shanghai, China	Not available	Not available