Lung ultrasound in UIP and NSIP: clinical and functional associations Authors: Ge Du, Min Li, Dongxia Hao, Min Cao, Jiali Wang, Huiying Feng, Dongchang Wang, Huan Ye Published: 2026 ABSTRACT BackgroundThis study evaluated lung ultrasound (LUS) for distinguishing nonspecific interstitial pneumonia (NSIP) from usual interstitial pneumonia (UIP) and its association with clinical and functional parameters.MethodsSeventy-six patients with HRCT-diagnosed UIP (n = 42) or NSIP (n = 34) underwent LUS. We analyzed clinical characteristic and LUS parameters, including coalescent (c-LUSS), quantitative (q-LUSS), and pleural scores. Logistic and linear regressions identified LUS parameters associated with UIP/NSIP and pulmonary function tests. ROC analysis evaluated the diagnostic performance of sex combined with pleural line total score.ResultsPatients with UIP had higher pleural line total, morphology, disruption, and fragmentation scores (all P < 0.05). Adjusted models identified several independent predictors distinguishing NSIP from UIP, notably the c-LUSS plus pleural line total score (OR: 1.06; 95% CI: 1.01–1.12), q-LUSS plus pleural line total score (OR: 1.07; 95% CI: 1.02–1.13), and pleural line total score (OR: 1.12; 95% CI: 1.03–1.22), alongside specific morphological subscores. The c-LUSS plus pleural line total score (OR: 1.12, 95% CI: 1.02–1.23), q-LUSS plus pleural line total score (OR: 1.1, 95% CI: 1.01–1.2), pleural line total score (OR: 1.14, 95% CI: 1–1.29), pleural thickness score (OR: 1.37, 95% CI: 1.02–1.85) remained independently associated with the severe FEV1. Meanwhile, the q-LUSS (β: −5.88; 95% CI: −11.33 to − 0.44) and B-line score of 3 or more (β: −7.01; 95% CI: −13.48 to − 0.54) were independently associated with 6MWD in UIP/NSIP patients. Whereas, these LUS indicators cannot reliably predict the severity of diffusion dysfunction indexed by DLCO % predicted (all P > 0.05). Combining sex and pleural line total score yielded an AUC of 0.782 (95% CI: 0.677–0.887, bootstrap C-index: 0.772).ConclusionsLUS, especially pleural assessment, may support differentiation between UIP and NSIP, but it cannot replace HRCT or multidisciplinary assessment. It exhibits relatively low performance in evaluating the severity of pulmonary diffusion dysfunction. Further external validation is still needed.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12890-026-04543-1. Introduction Interstitial lung diseases (ILDs) are a group of disorders involving the lung parenchyma, characterized by varying degrees of pulmonary inflammation and fibrosis [1]. These disorders often share similar clinical manifestations, including progressive dyspnea, dry cough, impaired lung function, and poor clinical outcomes, which pose a substantial threat to patient survival and quality of life. However, regarding the underlying pathology, ILDs exhibit substantial heterogeneity in treatment response and clinical outcomes [2]. Usual interstitial pneumonia (UIP) and nonspecific interstitial pneumonia (NSIP) are two common subtypes of ILDs, each with distinct imaging, clinical, and pathological features [3]. UIP typically involves subpleural and basal regions, presenting with heterogeneous fibrosis, fibroblastic foci, and honeycombing [1, 4]. Fibrosis is typically spatially and temporally heterogeneous. Clinically, patients with UIP generally have a poor prognosis, with limited treatment response, and irreversible pulmonary function decline [5]. Conversely, NSIP often presents with more uniform ground-glass and irregular reticular opacities, usually lacking destructive honeycombing or traction bronchiectasis [6]. Patients with NSIP usually have a younger age of onset, with a higher prevalence among females and nonsmokers [7]. These patients respond well to corticosteroid therapy and have a better long-term prognosis. Owing to their different imaging features, high-resolution computed tomography (HRCT) is generally used in clinical practice to distinguish UIP and NSIP [8, 9]. However, its application is often limited in vulnerable populations, particularly critically ill patients who are unfit for transportation and pregnant women who require strict avoidance of ionizing radiation. In contrast, lung ultrasound (LUS), a nonradiative, portable, and noninvasive imaging method, may serve as a valuable bedside tool for assessing patients with ILDs [10]. By visualizing pleural abnormalities and detecting B-lines (vertical artifacts caused by the thickening of the interlobular septa), LUS is increasingly used to assess pulmonary involvement in ILDs [11]. Evidence has shown that B-lines correlate with the extent of fibrotic lesions on HRCT [12]. Furthermore, pleural line abnormalities such as irregularities, thickening, or fragmentation, may reflect fibrosis beneath the pleura, offering insights specific to certain ILD subtypes [13]. Multiple scoring systems have been developed to quantify these findings, including the coalescent LUS score (c-LUSS), quantitative LUS score (q-LUSS), and specialized pleural line scoring systems [14]. In this study, we enrolled 76 patients with HRCT-confirmed UIP (n = 42) or NSIP (n = 34), compared the differences in LUS parameters between the two groups, and further evaluated the associations between LUS scores and functional impairment indicators, including hemoglobin-adjusted diffusing capacity of the lungs for carbon monoxide (DLCO) % predicted, 6-min walk distance (6MWD) and forced expiratory volume in 1 s (FEV1). This exploratory study sought to preliminarily explore whether LUS could help differentiate NSIP from UIP and reflect functional impairment in affected patients. Materials and methods Study design and population This cross-sectional study was conducted at Beijing Chest Hospital in 2025. This is a single-center exploratory study with a small sample size, which was constrained by the number of eligible cases recruited during the study period. A total of 307 adult patients with suspected ILD who presented with dry cough and dyspnea were screened. All patients underwent HRCT for the initial assessment. The confirmation of UIP and NSIP diagnoses in this study was achieved via a multidisciplinary discussion among pulmonology and radiology specialists, utilizing HRCT imaging features. Patients with imaging features inconsistent with those of UIP or NSIP were excluded (n = 219). Among the remaining 88 patients with HRCT-diagnosed UIP or NSIP, 12 were excluded because of missing LUS data. Seventy-six patients were included in the final analysis (Fig. 1). According to the 2018 ATS/ERS/JRS/ALAT diagnostic criteria for UIP pattern on HRCT [15], we classified all patients with UIP manifestations into definite UIP pattern (n = 30), probable UIP pattern (n = 8), and indeterminate UIP pattern (n = 4) based on radiographic features. Written informed consent was obtained from all participants. The exclusion criteria were pulmonary edema (Clinical manifestations: orthopnea, paroxysmal nocturnal dyspnea, frothy sputum or other typical manifestations of acute/chronic heart failure; Chest HRCT: bilateral symmetric ground-glass opacities, interlobular septal thickening or other signs of cardiogenic pulmonary edema; LUS: diffuse B-lines symmetrically distributed in bilateral lung fields combined with obvious pleural effusion) and obstructive lung disease. The study protocol was approved by the Institutional Review Board of Beijing Chest Hospital (Grant No. YJS-2025-02). Fig. 1Flowchart of patient selection. ILD, Interstitial lung disease; HRCT, high-resolution computed tomography; UIP, usual interstitial pneumonia; NSIP, nonspecific interstitial pneumonia Flowchart of patient selection. ILD, Interstitial lung disease; HRCT, high-resolution computed tomography; UIP, usual interstitial pneumonia; NSIP, nonspecific interstitial pneumonia Data collection Demographic and clinical information was collected, including age, sex, body mass index (BMI; ≤25 and > 25 kg/m2), smoking status (never, former, or current), alcohol consumption (never, former, or current), education level (< 9, 9–12, > 12 years), occupation (retired, household unemployment, or employment), marital status (living with a partner or living alone), and physical activity level (sedentary, moderate, or vigorous). Physical activity was categorized as sedentary, moderate, or vigorous using the World Health Organization metabolic equivalent (MET) classifications [16]. Comorbidities such as connective tissue disease (CTD), diabetes mellitus, hypertension, coronary heart disease, and reflux were documented. Detailed exposure histories, including occupational, environmental, and medication-related exposures were recorded. Pulmonary function tests performed concurrently with LUS were documented, primarily including hemoglobin-adjusted DLCO % predicted, 6MWD and FEV1. Ventilatory impairment severity was classified according to the American Thoracic Society/European Respiratory Society (ATS/ERS) 2005 criteria [17] based on the postbronchodilator FEV1 as a percentage of the predicted value: mild (≥ 70% predicted), moderate (60%-69% predicted), or severe (< 60% predicted, combining the moderately severe to very severe categories). LUS examination and scoring The LUS examinations were performed according to standardized protocols. All patients were scanned in the supine or semi-recumbent position. Chests were divided into 12 scanning zones across the anterior, lateral, and posterior walls. All LUS assessments were conducted using the LOGIQ E9 system (GE Healthcare). A curved-array transducer (2–5 MHz) was used for general scanning, and a linear-array transducer (5–12 MHz) was used to evaluate the pleural line and diaphragm in detail. A range of LUS parameters were derived for analysis, including the c-LUSS, q-LUSS, pleural line total score, and several pleural subscores (morphology, thickness, disruption, and fragmentation). Additional parameters included B-line counts (score of 3 or more), B-line confluence score, and diaphragm excursion. Supplementary Table S1 presents the scoring criteria for different lung ultrasound scoring systems. The total score was the sum of the scores of the 12 zones. All LUS images were acquired by an experienced sonographer, and the results were interpreted and confirmed by two sonographers who had undergone standardized training, both of whom were blinded to the clinical and HRCT findings. Statistical analysis Continuous variables were expressed as mean ± standard deviation (SD) or median with interquartile range (IQR) based on data distribution, whereas categorical variables were presented as numbers and frequencies. Comparisons between the UIP and NSIP groups were made using independent t-tests, Mann–Whitney U tests, or chi-square tests, as appropriate. We conducted univariate and multivariate logistic regression analyses to identify LUS parameters independently associated with the differentiation between UIP and NSIP, coding NSIP as 0 and UIP as 1. The multivariate analysis was adjusted for age, sex, smoking status, and CTD. Subgroup analysis was performed based on age, sex, smoking status, BMI, and CTD to assess whether the associations remained consistent across subgroups, and interaction terms were introduced to test for effect modifications. We performed a sensitivity analysis comparing the NSIP group with the definite UIP subgroup to verify whether the direction and statistical significance of differences in LUS pleural-related scores remained consistent with the overall findings. For outcome variables with missing values, including hemoglobin-adjusted DLCO % predicted (n = 67), 6MWD (n = 60), and FEV1 (n = 70), a complete case analysis was performed. Only patients with complete data for the respective outcomes were included in the analysis. No imputation of missing data was conducted, given the low proportion of missing values (11.8% and 21.1%, respectively) and the confirmation of missing data completely at random using Little’s MCAR test (P > 0.05). Multivariate linear regression was used to assess the associations between the LUS scores and continuous functional outcomes (hemoglobin-adjusted DLCO % predicted, and 6MWD) adjusted for the same set of covariates. Additionally, multivariate logistic regression was applied to evaluate the LUS parameters in relation to FEV1 severity (mild, moderate and severe). ROC curve analyses were performed to evaluate the discriminatory ability of the sex, the pleural line total score and combined predictive model adjusted for sex and pleural line total score. The area under the ROC characteristic curve (AUC) was calculated for each model. All statistical tests were two-tailed, and P < 0.05 was considered statistically significant. Given the relatively small sample size in the present study (n = 76) and the number of covariates included in the multivariable regression model, there was a risk of overfitting. To evaluate the stability of the model, we performed internal validation using the bootstrap method with 1000 resamples, and the corrected C-statistic was calculated. All the analyses were performed with R statistics software (version 2.4.3; http://www.R-project.org, The R Foundation) and Free Statistics software (version 2.4). Results Baseline characteristics The baseline characteristics of the study participants are summarized in Table 1. Significant differences were observed in sex distribution and smoking status between the two groups. The UIP group had a higher proportion of males (76.2% vs. 38.2%, P < 0.001) and a higher proportion of smokers (61.9% vs. 23.5%, P = 0.003). Regarding age, BMI, education level, occupation, marital status, comorbidities (CTD, diabetes, hypertension, coronary heart disease, reflux), medication use, contact history, and physical activity level, no significant differences were found between the two groups (all P > 0.05). No patients met the diagnostic criteria for chronic heart failure (including chronic diastolic heart failure). Table 1Baseline characteristics of study participantsVariablesNSIP (n = 34)UIP (n = 42)P valueSex< 0.001* Male13 (38.2%)32 (76.2%) Female21 (61.8%)10 (23.8%)Age (year), Mean ± SD65.4 ± 9.066.5 ± 9.60.608Smoke Status0.003* Never26 (76.5%)16 (38.1%) Former4 (11.8%)17 (40.5%) Current4 (11.8%)9 (21.4%)Drink0.21 Never26 (76.5%)24 (57.1%) Former4 (11.8%)9 (21.4%) Current4 (11.8%)9 (21.4%)BMI, Mean ± SD25.5 ± 3.324.7 ± 3.90.364Education level0.192 < 919 (55.9%)15 (35.7%) 9–125 (14.7%)11 (26.2%) > 1210 (29.4%)16 (38.1%)Occupation0.334 Retire17 (50%)28 (66.7%) Household unemployment11 (32.4%)10 (23.8%) Employment6 (17.6%)4 (9.5%)Marital status0.373 Living with partner33 (97.1%)38 (90.5%) Living alone1 (2.9%)4 (9.5%)CTD0.097 No18 (52.9%)30 (71.4%) Yes16 (47.1%)12 (28.6%)Diabetes0.738 No27 (79.4%)32 (76.2%) Yes7 (20.6%)10 (23.8%)Hypertension0.801 No22 (64.7%)26 (61.9%) Yes12 (35.3%)16 (38.1%)Coronary heart disease0.978 No25 (73.5%)31 (73.8%) Yes9 (26.5%)11 (26.2%)Reflux0.408 No23 (67.6%)32 (76.2%) Yes11 (32.4%)10 (23.8%)Medication0.45 No32 (94.1%)37 (88.1%) Yes2 (5.9%)5 (11.9%)Contact history0.624 No33 (97.1%)39 (92.9%) Yes1 (2.9%)3 (7.1%)Activity0.627 Sedentary6 (17.6%)5 (11.9%) Moderate19 (55.9%)28 (66.7%) Vigorous9 (26.5%)9 (21.4%)* Statistically significant difference. UIP Usual interstitial pneumonia, NSIP Nonspecific interstitial pneumonia, SD Standard deviation, BMI Body mass index, CTD Connective tissue disease Table 1Baseline characteristics of study participantsVariablesNSIP (n = 34)UIP (n = 42)P valueSex< 0.001* Male13 (38.2%)32 (76.2%) Female21 (61.8%)10 (23.8%)Age (year), Mean ± SD65.4 ± 9.066.5 ± 9.60.608Smoke Status0.003* Never26 (76.5%)16 (38.1%) Former4 (11.8%)17 (40.5%) Current4 (11.8%)9 (21.4%)Drink0.21 Never26 (76.5%)24 (57.1%) Former4 (11.8%)9 (21.4%) Current4 (11.8%)9 (21.4%)BMI, Mean ± SD25.5 ± 3.324.7 ± 3.90.364Education level0.192 < 919 (55.9%)15 (35.7%) 9–125 (14.7%)11 (26.2%) > 1210 (29.4%)16 (38.1%)Occupation0.334 Retire17 (50%)28 (66.7%) Household unemployment11 (32.4%)10 (23.8%) Employment6 (17.6%)4 (9.5%)Marital status0.373 Living with partner33 (97.1%)38 (90.5%) Living alone1 (2.9%)4 (9.5%)CTD0.097 No18 (52.9%)30 (71.4%) Yes16 (47.1%)12 (28.6%)Diabetes0.738 No27 (79.4%)32 (76.2%) Yes7 (20.6%)10 (23.8%)Hypertension0.801 No22 (64.7%)26 (61.9%) Yes12 (35.3%)16 (38.1%)Coronary heart disease0.978 No25 (73.5%)31 (73.8%) Yes9 (26.5%)11 (26.2%)Reflux0.408 No23 (67.6%)32 (76.2%) Yes11 (32.4%)10 (23.8%)Medication0.45 No32 (94.1%)37 (88.1%) Yes2 (5.9%)5 (11.9%)Contact history0.624 No33 (97.1%)39 (92.9%) Yes1 (2.9%)3 (7.1%)Activity0.627 Sedentary6 (17.6%)5 (11.9%) Moderate19 (55.9%)28 (66.7%) Vigorous9 (26.5%)9 (21.4%)* Statistically significant difference. UIP Usual interstitial pneumonia, NSIP Nonspecific interstitial pneumonia, SD Standard deviation, BMI Body mass index, CTD Connective tissue disease Baseline characteristics of study participants * Statistically significant difference. UIP Usual interstitial pneumonia, NSIP Nonspecific interstitial pneumonia, SD Standard deviation, BMI Body mass index, CTD Connective tissue disease Comparison of LUS scores between UIP and NSIP A comparison of the LUS parameters between the two groups is shown in Table 2. Patients with UIP had significantly higher values for several LUS scores than those with NSIP, including the c-LUSS plus pleural line total score (28.8 ± 9.9 vs. 22.9 ± 11.3, P = 0.019), q-LUSS plus pleural line total score (27.0 ± 9.1 vs. 20.6 ± 10.2, P = 0.006), pleural line total score (16.0 ± 6.5 vs. 11.5 ± 6.4, P = 0.004), pleural morphology score (7.0 [4.2–10.8] vs. 5.0 [2.0–6.0], P = 0.006), pleural disruption score (3.0 [1.2-4.0] vs. 1.0 [0.0-2.8], P = 0.005), and pleural fragmentation score (1.3 ± 1.2 vs. 0.6 ± 0.9, P = 0.013). No significant differences were observed in the c-LUSS, q-LUSS, pleural thickness score, B-line score of 3 or more, B-line confluence score, or diaphragm excursion between the two groups. Table 2Comparison of lung ultrasound scores between UIP and NSIPVariablesNSIP (n = 34)UIP (n = 42)P valuec-LUSS plus pleural line total score, Mean ± SD22.9 ± 11.328.8 ± 9.90.019*q-LUSS plus pleural line total score, Mean ± SD20.6 ± 10.227.0 ± 9.10.006*c-LUSS, Mean ± SD11.6 ± 6.513.0 ± 5.30.3q-LUSS, Mean ± SD9.3 ± 4.911.0 ± 3.50.091Pleural line total score, Mean ± SD11.5 ± 6.416.0 ± 6.50.004*Pleural morphology score, Median (IQR)5.0 (2.0, 6.0)7.0 (4.2, 10.8)0.006*Pleural thickness score, Mean ± SD6.8 ± 3.58.1 ± 2.80.059Pleural disruption score, Mean ± SD1.0 (0.0, 2.8)3.0 (1.2, 4.0)0.005*Pleural fragmentation score, Mean ± SD0.6 ± 0.91.3 ± 1.20.013*B-line score of 3 or more, Mean ± SD6.5 ± 3.57.7 ± 3.10.106B-line confluence score, Median (IQR)1.0 (0.0, 2.0)1.0 (0.0, 2.0)0.464Diaphragm excursion (cm), Mean ± SD1.9 ± 0.8 (n = 28)1.9 ± 0.8 (n = 36)0.875* Statistically significant difference. UIP Usual interstitial pneumonia, NSIP Nonspecific interstitial pneumonia, c-LUSS Coalescent lung ultrasound score, q-LUSS Quantitative lung ultrasound score, SD Standard deviation, IQR Interquartile range Table 2Comparison of lung ultrasound scores between UIP and NSIPVariablesNSIP (n = 34)UIP (n = 42)P valuec-LUSS plus pleural line total score, Mean ± SD22.9 ± 11.328.8 ± 9.90.019*q-LUSS plus pleural line total score, Mean ± SD20.6 ± 10.227.0 ± 9.10.006*c-LUSS, Mean ± SD11.6 ± 6.513.0 ± 5.30.3q-LUSS, Mean ± SD9.3 ± 4.911.0 ± 3.50.091Pleural line total score, Mean ± SD11.5 ± 6.416.0 ± 6.50.004*Pleural morphology score, Median (IQR)5.0 (2.0, 6.0)7.0 (4.2, 10.8)0.006*Pleural thickness score, Mean ± SD6.8 ± 3.58.1 ± 2.80.059Pleural disruption score, Mean ± SD1.0 (0.0, 2.8)3.0 (1.2, 4.0)0.005*Pleural fragmentation score, Mean ± SD0.6 ± 0.91.3 ± 1.20.013*B-line score of 3 or more, Mean ± SD6.5 ± 3.57.7 ± 3.10.106B-line confluence score, Median (IQR)1.0 (0.0, 2.0)1.0 (0.0, 2.0)0.464Diaphragm excursion (cm), Mean ± SD1.9 ± 0.8 (n = 28)1.9 ± 0.8 (n = 36)0.875* Statistically significant difference. UIP Usual interstitial pneumonia, NSIP Nonspecific interstitial pneumonia, c-LUSS Coalescent lung ultrasound score, q-LUSS Quantitative lung ultrasound score, SD Standard deviation, IQR Interquartile range Comparison of lung ultrasound scores between UIP and NSIP * Statistically significant difference. UIP Usual interstitial pneumonia, NSIP Nonspecific interstitial pneumonia, c-LUSS Coalescent lung ultrasound score, q-LUSS Quantitative lung ultrasound score, SD Standard deviation, IQR Interquartile range Logistic regression analysis for differentiating NSIP from UIP Univariate logistic regression analysis of demographic and clinical variables (Table 3) revealed that female sex (OR: 0.19, 95% CI: 0.07–0.52, P = 0.001) were significantly associated with NSIP (coded as 0) and former smoking status (OR: 6.91, 95% CI: 1.97–24.22, P = 0.003) were significantly associated with UIP (coded as 1). Univariate and multivariate logistic regression analyses of LUS variables (Table 4) showed that after adjusting for age, sex, smoking status, and CTD, the c-LUSS plus pleural line total score (adjusted OR: 1.06, 95% CI: 1.01–1.12, P = 0.019), q-LUSS plus pleural line total score (adjusted OR: 1.07, 95% CI: 1.02–1.13, P = 0.012), pleural line total score (adjusted OR: 1.12, 95% CI: 1.03–1.22, P = 0.009), pleural morphology score (adjusted OR: 1.15, 95% CI: 1.02–1.3, P = 0.022), pleural thickness score (adjusted OR: 1.22, 95% CI: 1.02–1.45, P = 0.029), pleural disruption score (adjusted OR: 1.47, 95% CI: 1.06–2.05, P = 0.022), and pleural fragmentation score (adjusted OR: 2.03, 95% CI: 1.15–3.6, P = 0.015) remained independently associated with the differentiation between NSIP and UIP. Higher scores were associated with higher odds of having a UIP pattern, indicating that patients with UIP had more severe pleural abnormalities. Table 3Univariate logistic regression analysis of demographic and clinical variables for differentiating NSIP from UIPVariableOR (95% CI)P valueSex (female vs. male)0.19 (0.07–0.52)0.001*Age1.01 (0.96–1.06)0.603Smoking status Former vs. never6.91 (1.97–24.22)0.003* Current vs. never3.66 (0.96–13.86)0.056Alcohol consumption Former vs. never2.44 (0.66–8.96)0.18 Current vs. never2.44 (0.66–8.96)0.18BMI0.94 (0.83–1.07)0.36Education level 9–12 years vs. <9 years2.79 (0.79–9.78)0.11 > 12 years vs. <9 years2.03 (0.72–5.74)0.183Occupation Unemployed/Household vs. retired0.55 (0.19–1.57)0.266 Employed vs. retired0.4 (0.1–1.64)0.206Marital status (living alone vs. with partner)3.47 (0.37–32.64)0.276Physical activity Moderate vs. sedentary1.77 (0.47–6.63)0.398 Vigorous vs. sedentary1.2 (0.27–5.4)0.812CTD (yes vs. no)0.45 (0.17–1.16)0.099Diabetes (yes vs. no)1.21 (0.4–3.6)0.738Hypertension (yes vs. no)1.13 (0.44–2.89)0.801CHD (yes vs. no)0.99 (0.35–2.75)0.978Reflux (yes vs. no)0.65 (0.24–1.79)0.409Medication (yes vs. no)2.16 (0.39–11.92)0.376Contact history (yes vs. no)2.54 (0.25–25.58)0.429* Statistically significant difference. UIP Usual interstitial pneumonia, NSIP Nonspecific interstitial pneumonia, OR Odds ratio, CI Confidence interval, BMI Body mass index, CTD Connective tissue disease, CHD Coronary heart disease. NSIP was coded as 0 and UIP as 1 in the logistic regression model Table 3Univariate logistic regression analysis of demographic and clinical variables for differentiating NSIP from UIPVariableOR (95% CI)P valueSex (female vs. male)0.19 (0.07–0.52)0.001*Age1.01 (0.96–1.06)0.603Smoking status Former vs. never6.91 (1.97–24.22)0.003* Current vs. never3.66 (0.96–13.86)0.056Alcohol consumption Former vs. never2.44 (0.66–8.96)0.18 Current vs. never2.44 (0.66–8.96)0.18BMI0.94 (0.83–1.07)0.36Education level 9–12 years vs. <9 years2.79 (0.79–9.78)0.11 > 12 years vs. <9 years2.03 (0.72–5.74)0.183Occupation Unemployed/Household vs. retired0.55 (0.19–1.57)0.266 Employed vs. retired0.4 (0.1–1.64)0.206Marital status (living alone vs. with partner)3.47 (0.37–32.64)0.276Physical activity Moderate vs. sedentary1.77 (0.47–6.63)0.398 Vigorous vs. sedentary1.2 (0.27–5.4)0.812CTD (yes vs. no)0.45 (0.17–1.16)0.099Diabetes (yes vs. no)1.21 (0.4–3.6)0.738Hypertension (yes vs. no)1.13 (0.44–2.89)0.801CHD (yes vs. no)0.99 (0.35–2.75)0.978Reflux (yes vs. no)0.65 (0.24–1.79)0.409Medication (yes vs. no)2.16 (0.39–11.92)0.376Contact history (yes vs. no)2.54 (0.25–25.58)0.429* Statistically significant difference. UIP Usual interstitial pneumonia, NSIP Nonspecific interstitial pneumonia, OR Odds ratio, CI Confidence interval, BMI Body mass index, CTD Connective tissue disease, CHD Coronary heart disease. NSIP was coded as 0 and UIP as 1 in the logistic regression model Univariate logistic regression analysis of demographic and clinical variables for differentiating NSIP from UIP * Statistically significant difference. UIP Usual interstitial pneumonia, NSIP Nonspecific interstitial pneumonia, OR Odds ratio, CI Confidence interval, BMI Body mass index, CTD Connective tissue disease, CHD Coronary heart disease. NSIP was coded as 0 and UIP as 1 in the logistic regression model Table 4Univariate and multivariate logistic regression analysis of lung ultrasound variables for differentiating NSIP from UIPVariableCrude OR (95% CI)Crude P valueAdjusted OR (95% CI)Adjusted P valuec-LUSS plus pleural line total score1.06 (1.01–1.11)0.023*1.06 (1.01–1.12)0.019*q-LUSS plus pleural line total score1.07 (1.02–1.13)0.009*1.07 (1.02–1.13)0.012*c-LUSS1.04 (0.96–1.13)0.2971.06 (0.97–1.16)0.22q-LUSS1.1 (0.98–1.23)0.0951.1 (0.97–1.25)0.128Pleural line total score1.12 (1.03–1.21)0.006*1.12 (1.03–1.22)0.009*Pleural morphology score1.16 (1.04–1.3)0.009*1.15 (1.02–1.3)0.022*Pleural thickness score1.15 (0.99–1.34)0.0621.22 (1.02–1.45)0.029*Pleural disruption score1.53 (1.13–2.08)0.007*1.47 (1.06–2.05)0.022*Pleural fragmentation score1.8 (1.11–2.94)0.018*2.03 (1.15–3.6)0.015*B-line score of 3 or more1.12 (0.98–1.29)0.1071.15 (0.97–1.36)0.097B-line confluence score1.1 (0.83–1.45)0.5031.2 (0.87–1.67)0.263* Statistically significant difference. UIP Usual interstitial pneumonia, NSIP Nonspecific interstitial pneumonia, c-LUSS Coalescent lung ultrasound score, q-LUSS Quantitative lung ultrasound score, OR Odds ratio, CI Confidence interval. NSIP was coded as 0 and UIP as 1 in the logistic regression model. Adjusted age, sex, smoking status, and CTD Table 4Univariate and multivariate logistic regression analysis of lung ultrasound variables for differentiating NSIP from UIPVariableCrude OR (95% CI)Crude P valueAdjusted OR (95% CI)Adjusted P valuec-LUSS plus pleural line total score1.06 (1.01–1.11)0.023*1.06 (1.01–1.12)0.019*q-LUSS plus pleural line total score1.07 (1.02–1.13)0.009*1.07 (1.02–1.13)0.012*c-LUSS1.04 (0.96–1.13)0.2971.06 (0.97–1.16)0.22q-LUSS1.1 (0.98–1.23)0.0951.1 (0.97–1.25)0.128Pleural line total score1.12 (1.03–1.21)0.006*1.12 (1.03–1.22)0.009*Pleural morphology score1.16 (1.04–1.3)0.009*1.15 (1.02–1.3)0.022*Pleural thickness score1.15 (0.99–1.34)0.0621.22 (1.02–1.45)0.029*Pleural disruption score1.53 (1.13–2.08)0.007*1.47 (1.06–2.05)0.022*Pleural fragmentation score1.8 (1.11–2.94)0.018*2.03 (1.15–3.6)0.015*B-line score of 3 or more1.12 (0.98–1.29)0.1071.15 (0.97–1.36)0.097B-line confluence score1.1 (0.83–1.45)0.5031.2 (0.87–1.67)0.263* Statistically significant difference. UIP Usual interstitial pneumonia, NSIP Nonspecific interstitial pneumonia, c-LUSS Coalescent lung ultrasound score, q-LUSS Quantitative lung ultrasound score, OR Odds ratio, CI Confidence interval. NSIP was coded as 0 and UIP as 1 in the logistic regression model. Adjusted age, sex, smoking status, and CTD Univariate and multivariate logistic regression analysis of lung ultrasound variables for differentiating NSIP from UIP * Statistically significant difference. UIP Usual interstitial pneumonia, NSIP Nonspecific interstitial pneumonia, c-LUSS Coalescent lung ultrasound score, q-LUSS Quantitative lung ultrasound score, OR Odds ratio, CI Confidence interval. NSIP was coded as 0 and UIP as 1 in the logistic regression model. Adjusted age, sex, smoking status, and CTD Subgroup analysis Subgroup analysis (Supplementary Figure S1) was performed to assess potential effect modifications on the relationship between pleura line total score and UIP. No significant interactions were found for age, sex, smoking status, BMI, or CTD status. Sensitivity analysis We conducted a sensitivity analysis by comparing NSIP group with definite UIP group (Supplementary Table S2), The result showed that the direction and significance of LUS pleural-related scores differences remained consistent with the overall results. Associations between lung ultrasound scores and functional impairment After adjusting for age, sex, smoking status and CTD, multivariate logistic regression of LUS parameters with FEV1 (Mild-Moderate vs. Severe) showed that the c-LUSS plus pleural line total score (adjusted OR: 1.12, 95% CI: 1.02–1.23, P = 0.022), q-LUSS plus pleural line total score (adjusted OR: 1.1, 95% CI: 1.01–1.2, P = 0.034), pleural line total score (adjusted OR: 1.14, 95% CI: 1-1.29, P = 0.043), pleural thickness score (adjusted OR: 1.37, 95% CI: 1.02–1.85, P = 0.037) remained independently associated with the severe FEV1 (Table 5). For 6MWD, the q-LUSS (adj. β: −5.88, 95% CI: −11.33 to − 0.44, P = 0.039) and B-line score of 3 or more (adj. β: −7.01, 95% CI: −13.48 to − 0.54, P = 0.038) were significantly and negatively associated with walking distance. No significant associations were observed between LUS parameters and hemoglobin-corrected DLCO % predicted (all P > 0.05) (Supplementary Table S3). Table 5Univariate and multivariate logistic regression of LUS scores with FEV1 (Mild-Moderate(n = 60) vs. Severe(n = 10))VariableCrude OR (95% CI)Crude PAdjusted OR (95% CI)Adjusted Pc-LUSS plus pleural line total score1.05(0.99–1.12)0.1111.12(1.02–1.23)0.022*q-LUSS plus pleural line total score1.05(0.98–1.12)0.1531.1(1.01–1.2)0.034*c-LUSS1.10(0.96–1.25)0.1591.19(1-1.43)0.054q-LUSS1.12(0.95–1.32)0.1881.21(0.98–1.49)0.072Pleural line total score1.07(0.97–1.18)0.1991.14(1-1.29)0.043*Pleural morphology score1.05(0.91–1.20)0.5221.11(0.95–1.3)0.197Pleural thickness score1.23(0.97–1.57)0.0911.37(1.02–1.85)0.037*Pleural disruption score1.10(0.76–1.60)0.6151.27(0.83–3.11)0.269Pleural fragmentation score1.29(0.73–2.26)0.3771.58(0.81–3.11)0.181B-line score of 3 or more1.08(0.87–1.34)0.4891.21(0.92–1.58)0.17B line confluence score1.11(0.77–1.60)0.5901.16(0.74–1.82)0.514The outcome variable was FEV1 severity: severe FEV1 impairment was coded as 1, mild-moderate impairment as 0.The sample sizes for FEV1 were 70, which was attributable to the fact that some patients were unable to complete the corresponding functional tests. Complete case analysis was applied. Adjusted age, sex, smoking status, and CTDFEV1 Forced expiratory volume in 1 second, c-LUSS Coalescent lung ultrasound score, q-LUSS Quantitative lung ultrasound score, OR Odds ratio, CI Confidence interval Table 5Univariate and multivariate logistic regression of LUS scores with FEV1 (Mild-Moderate(n = 60) vs. Severe(n = 10))VariableCrude OR (95% CI)Crude PAdjusted OR (95% CI)Adjusted Pc-LUSS plus pleural line total score1.05(0.99–1.12)0.1111.12(1.02–1.23)0.022*q-LUSS plus pleural line total score1.05(0.98–1.12)0.1531.1(1.01–1.2)0.034*c-LUSS1.10(0.96–1.25)0.1591.19(1-1.43)0.054q-LUSS1.12(0.95–1.32)0.1881.21(0.98–1.49)0.072Pleural line total score1.07(0.97–1.18)0.1991.14(1-1.29)0.043*Pleural morphology score1.05(0.91–1.20)0.5221.11(0.95–1.3)0.197Pleural thickness score1.23(0.97–1.57)0.0911.37(1.02–1.85)0.037*Pleural disruption score1.10(0.76–1.60)0.6151.27(0.83–3.11)0.269Pleural fragmentation score1.29(0.73–2.26)0.3771.58(0.81–3.11)0.181B-line score of 3 or more1.08(0.87–1.34)0.4891.21(0.92–1.58)0.17B line confluence score1.11(0.77–1.60)0.5901.16(0.74–1.82)0.514The outcome variable was FEV1 severity: severe FEV1 impairment was coded as 1, mild-moderate impairment as 0.The sample sizes for FEV1 were 70, which was attributable to the fact that some patients were unable to complete the corresponding functional tests. Complete case analysis was applied. Adjusted age, sex, smoking status, and CTDFEV1 Forced expiratory volume in 1 second, c-LUSS Coalescent lung ultrasound score, q-LUSS Quantitative lung ultrasound score, OR Odds ratio, CI Confidence interval Univariate and multivariate logistic regression of LUS scores with FEV1 (Mild-Moderate(n = 60) vs. Severe(n = 10)) The outcome variable was FEV1 severity: severe FEV1 impairment was coded as 1, mild-moderate impairment as 0.The sample sizes for FEV1 were 70, which was attributable to the fact that some patients were unable to complete the corresponding functional tests. Complete case analysis was applied. Adjusted age, sex, smoking status, and CTD FEV1 Forced expiratory volume in 1 second, c-LUSS Coalescent lung ultrasound score, q-LUSS Quantitative lung ultrasound score, OR Odds ratio, CI Confidence interval ROC curve analyses ROC curve analyses were performed to evaluate the discriminatory ability of sex alone, pleural line total score alone and the combined model, which included sex and pleural line total score. The AUCs of sex and pleural line total score were 0.69 and 0.686, respectively. Their combined model yielded an AUC of 0.782 and the corrected C-index for bootstrap sampling validation was 0.772, indicating better discriminatory performance for distinguishing NSIP from UIP (Fig. 2). The sensitivity, specificity, accuracy, and negative and positive predictive values are presented in Table 6. Fig. 2ROC analyses to evaluate the discrimination for distinguishing NSIP from UIP ROC analyses to evaluate the discrimination for distinguishing NSIP from UIP Table 6Diagnostic performance of three prediction modelssexpleural line total scoreModel 1*AUC, % (95% CI)69.0(58.4–79.5)68.6(56.5–80.7)78.2(67.7–88.7)Sensitivity (%)76.247.683.8Specificity (%)61.882.467.7Accuracy (%)68.763.276.3Positive predictive value (%)71.176.976.1Negative predictive value (%)67.756.076.7*The model 1 adjusted sex and pleural line total score Table 6Diagnostic performance of three prediction modelssexpleural line total scoreModel 1*AUC, % (95% CI)69.0(58.4–79.5)68.6(56.5–80.7)78.2(67.7–88.7)Sensitivity (%)76.247.683.8Specificity (%)61.882.467.7Accuracy (%)68.763.276.3Positive predictive value (%)71.176.976.1Negative predictive value (%)67.756.076.7*The model 1 adjusted sex and pleural line total score Diagnostic performance of three prediction models *The model 1 adjusted sex and pleural line total score The model 1 adjusted sex and pleural line total score. Discussion This single-center, cross-sectional study is the first to explore the utility of LUS scores in differentiating NSIP from UIP. In this study, we demonstrated the potential role of LUS in distinguishing between NSIP and UIP, while their value for evaluating pulmonary diffusion impairment remains limited. First, patients with UIP exhibited significantly higher pleural line-related scores, including the pleural line total score, pleural morphology score, pleural disruption score, and pleural fragmentation score, than those with NSIP (all P < 0.05). These findings highlight the distinct pleural abnormalities that may help differentiate between these two entities. Second, multivariate analysis identified multiple independent predictors for distinguishing UIP from NSIP: the c-LUSS plus pleural line total score, q-LUSS plus pleural line total score, pleural line total score, pleural morphology score, pleural thickness score, pleural disruption score, and pleural fragmentation score. All indices showed statistically significant adjusted odds ratios, supporting their potential as diagnostic markers. Third, for FEV1, the c-LUSS plus pleural line total score, q-LUSS plus pleural line total score, pleural line total score and pleural thickness score were independently associated with the severe FEV1. However, no significant correlation was observed between LUS parameters and hemoglobin-corrected DLCO % predicted in either the overall study population or the UIP subgroup. In terms of 6MWD, elevated q-LUSS and B-line score of 3 or more were independently associated with shorter walking distance. These results indicate that lung ultrasound performs better in evaluating ventilatory dysfunction than diffusion dysfunction. Finally, the model incorporating sex and pleural line total score yielded an AUC of 0.782 (95%CI: 0.677–0.887), demonstrating good discriminative ability for distinguishing NSIP from UIP. After 1000 bootstrap resampling internal validations, the bias-corrected C-index was 0.772. Pathologically, UIP exhibits heterogeneous fibrosis, fibroblastic foci, and eventual honeycombing, leading to pleural irregularities and architectural distortion [18]. The inevitable pleural involvement in UIP results in pleural irregularities, thickening, and architectural distortion [19, 20]. On LUS, these features manifest as pleural thickening, irregularity, disruption, and fragmentation, alongside abundant B-lines corresponding to interlobular septal remodeling. Conversely, NSIP features more uniform inflammation and fibrosis without fibroblastic foci or honeycombing, resulting in a relatively preserved pleura and correspondingly fewer LUS abnormalities [21]. Our findings are consistent with those of previous studies that have explored the utility of LUS in ILDs. Sperandeo et al. reported that patients with pulmonary fibrosis commonly exhibit pleural line thickening and irregularity, and these abnormalities correlate with the extent of fibrotic involvement [13]. Tardella et al. demonstrated a significant correlation between B-lines on LUS and HRCT fibrosis scores in patients with CTD-associated ILD [12]. This study is the first attempt to extend these observations by specifically comparing UIP and NSIP using a panel of LUS scores, highlighting the importance of pleural line abnormalities as key discriminators. Notably, after multivariate adjustment, pleural morphology, thickness, disruption, and fragmentation scores remained independently associated with ILD subtype, underscoring the value of a detailed pleural assessment beyond simple B-line quantification. Moreover, in our study, lung ultrasound was superior for evaluating ventilatory rather than diffusion dysfunction. According to previous published literature [22], we hypothesize two potential explanations as follow. First, LUS mainly reflects subpleural aeration status and peripheral airway obstruction, which closely matches ventilation indicators including FEV1. Nevertheless, DLCO evaluates the gas exchange capacity of the entire alveolar-capillary membrane, most of which locates in deep lung tissue beyond the detection range of LUS. Second, DLCO is modulated by extrapulmonary factors including hemoglobin level and cardiac output. LUS has inherent limitations in characterizing diffusion dysfunction. Notably, the number of cases with pulmonary function data was relatively limited, further external validation is still needed. In our study, the q-LUSS, B-line score of 3 or more were negatively associated with 6MWD, suggesting that greater ultrasound-detected pulmonary involvement correlates with reduced exercise capacity. This may occur because extensive fibrosis leads to restrictive physiology, increased work of breathing, and impaired oxygen delivery during exercise [23]. Combining sex with pleural line total score yielded favorable discriminative performance for differentiating NSIP from UIP, offering a potential noninvasive quantitative indicator for the differential diagnosis of ILD. A meta-analysis by Ebner et al. [24] demonstrated that male predominance and pleural abnormalities are key distinguishing features between UIP and NSIP, supporting the inclusion of sex and pleura-related markers in a diagnostic model. In agreement with these findings, Milad et al. [25] reported that pleural line alterations on LUS independently predicted UIP, reinforcing the clinical utility of sonographic pleural assessment as a noninvasive tool for ILD subtyping. This study has several limitations. First, it was a single-center cross-sectional study with a relatively small sample size, which may limit the generalizability of the results. Second, all patients were diagnosed with UIP and NSIP according to HRCT and multidisciplinary discussion, without histopathological confirmation. This approach may lead to misclassification bias, especially in patients with early disease or atypical imaging manifestations. Therefore, our findings should be interpreted with caution given the absence of a pathological gold standard. Future multicenter studies with larger sample sizes and pathologically confirmed cases are warranted to externally validate our findings. Third, the limited sample size may have introduced instability and overfitting risks into the multivariate regression model. Although bootstrap internal validation supported the robustness of our model, further validation in an independent external cohort is warranted. Fourth, the diagnostic model established in the present study did not undergo external validation. The AUC value obtained from the internal validation (0.782) may have been somewhat overestimated. Multicenter studies with independent external validation cohorts are required to assess the generalizability of this model. Furthermore, although all LUS scans were interpreted and cross- verified by two uniformly trained sonographers, we did not conduct formal quantitative assessments of intra- and inter-observer scoring agreement, which may introduce subtle subjective scoring bias. In conclusion, LUS may support differentiation between UIP and NSIP, but it cannot replace HRCT or multidisciplinary assessment. It exhibits relatively low performance in evaluating the severity of pulmonary diffusion dysfunction. The current findings are hypothesis-generating results, further external validation is still needed before LUS can be clinically promoted as a differential diagnostic tool for UIP and NSIP. Supplementary Information Supplementary Material 1: Supplementary Table S1. Criteria for different lung ultrasound scoring systems. Supplementary Table S2. Univariate and Multivariate Logistic Regression Analysis of LUS Variables for Differentiating NSIP from definite UIP. Supplementary Table S3. Multivariate Linear Regression of Lung Ultrasound Scores with DLCO % predicted and 6MWD. Supplementary Figure S1. Subgroup Analysis of the Association Between pleura line total score and UIP/NSIP. UIP, usual interstitial pneumonia; NSIP, nonspecific interstitial pneumonia; OR, odds ratio; CI, confidence interval; BMI, body mass index; CTD, connective tissue disease. Notes: NSIP was coded as 0 and UIP as 1 in the logistic regression model. Age and BMI cutoffs were based on median values (age: 65 years; BMI: 25 kg/m²). Adjusted age, sex, smoking status, and CTD Supplementary Material 1: Supplementary Table S1. Criteria for different lung ultrasound scoring systems. Supplementary Table S2. Univariate and Multivariate Logistic Regression Analysis of LUS Variables for Differentiating NSIP from definite UIP. Supplementary Table S3. Multivariate Linear Regression of Lung Ultrasound Scores with DLCO % predicted and 6MWD. Supplementary Figure S1. Subgroup Analysis of the Association Between pleura line total score and UIP/NSIP. UIP, usual interstitial pneumonia; NSIP, nonspecific interstitial pneumonia; OR, odds ratio; CI, confidence interval; BMI, body mass index; CTD, connective tissue disease. Notes: NSIP was coded as 0 and UIP as 1 in the logistic regression model. Age and BMI cutoffs were based on median values (age: 65 years; BMI: 25 kg/m²). Adjusted age, sex, smoking status, and CTD