Lan Song
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Lan Song has a total of 123 co-authors. Here are the top co-authors by number of shared publications:
Ge Hu
Department of Radiology, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China
10 shared publicationsZhenchen Zhu
Department of Radiology, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China
10 shared publicationsWei Han
Department of Evidence-based Medicine, Xuanwu Hospital, Capital Medical University, Beijing, China
9 shared publicationsZhoumeng Ying
Department of Radiology, State Key Laboratory of Complex Severe and Rare Diseases, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing, China
9 shared publicationsJinhua Wang
Department of Radiology, State Key Laboratory of Complex Severe and Rare Diseases, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Bejing, China
8 shared publicationsZhengyu Jin
Department of Radiology, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China
8 shared publications
Latest Publications
- Open Access
Computed tomography-derived body composition parameters as potential predictors of severity and adverse outcomes in viral pneumonia: a retrospective study
Background: Following the COVID-19 pandemic, co-infections with multiple respiratory viruses have become increasingly common, complicating the accurate prediction of disease progression and prognosis. This study assessed the use of computed tomography (CT)-derived body composition parameters combined with clinical risk factors to predict the severity and short-term adverse outcomes of viral pneumonia. Methods: A total of 140 hospitalized patients with viral pneumonia who had undergone chest CT were retrospectively included and stratified into severe and non-severe groups. Body composition, including visceral and subcutaneous adipose volumes (VAV and SAV) and erector spinae volume (ESV), was measured at T4, T8, and T12 costovertebral joint levels and the corresponding whole-vertebral level using 3D-Slicer. Serological indicators and 30-day adverse outcomes were recorded. Clinical, imaging, and integrated models were developed to distinguish patients’ severity and outcome. Results: The median age was 77 years (IQR: 68–85), with 91 (65%) patients being male. Sixty-five (46.4%) patients were severe patients and 26 (18.6%) had adverse outcomes within 30 days. Lower T12-ESV and higher T12-VAV were associated with viral pneumonia severity after adjusting for confounders ( OR = 0.823, 95% CI [0.733–0.924], p = 0.001; OR = 1.030, 95% CI [1.014–1.047], p < 0.001). Those with adverse outcomes had reduced subcutaneous fat ( p < 0.05). The area under the curve (AUC) of the integrated model using clinical and body composition parameters was the highest ( AUC = 0.843, p < 0.001). Conclusion: Decreased ESV and increased VAV at T12 level are independent predictors of severe viral pneumonia, while reduced SAV correlates well with adverse 30-day outcomes. CT-derived body composition parameters can predict the progression and short-term prognosis of viral pneumonia, thereby aiding clinical decision-making.
Open AccessComputed tomography-derived body composition parameters as potential predictors of severity and adverse outcomes in viral pneumonia: a retrospective study
Background: Following the COVID-19 pandemic, co-infections with multiple respiratory viruses have become increasingly common, complicating the accurate prediction of disease progression and prognosis. This study assessed the use of computed tomography (CT)-derived body composition parameters combined with clinical risk factors to predict the severity and short-term adverse outcomes of viral pneumonia. Methods: A total of 140 hospitalized patients with viral pneumonia who had undergone chest CT were retrospectively included and stratified into severe and non-severe groups. Body composition, including visceral and subcutaneous adipose volumes (VAV and SAV) and erector spinae volume (ESV), was measured at T4, T8, and T12 costovertebral joint levels and the corresponding whole-vertebral level using 3D-Slicer. Serological indicators and 30-day adverse outcomes were recorded. Clinical, imaging, and integrated models were developed to distinguish patients’ severity and outcome. Results: The median age was 77 years (IQR: 68–85), with 91 (65%) patients being male. Sixty-five (46.4%) patients were severe patients and 26 (18.6%) had adverse outcomes within 30 days. Lower T12-ESV and higher T12-VAV were associated with viral pneumonia severity after adjusting for confounders ( OR = 0.823, 95% CI [0.733–0.924], p = 0.001; OR = 1.030, 95% CI [1.014–1.047], p < 0.001). Those with adverse outcomes had reduced subcutaneous fat ( p < 0.05). The area under the curve (AUC) of the integrated model using clinical and body composition parameters was the highest ( AUC = 0.843, p < 0.001). Conclusion: Decreased ESV and increased VAV at T12 level are independent predictors of severe viral pneumonia, while reduced SAV correlates well with adverse 30-day outcomes. CT-derived body composition parameters can predict the progression and short-term prognosis of viral pneumonia, thereby aiding clinical decision-making.
- Open Access
- 17 July 2026
- Published by Springer Nature in BMC Medical Imaging
- Vol. 26 (1)
A CT-based deep learning model for the automated risk stratification of refractory Mycoplasma pneumoniae pneumonia in children
Background: The accurate identification of children with refractory Mycoplasma pneumoniae pneumonia (RMPP) remains challenging. This study aimed to develop a transformer-based model utilizing clinically indicated chest computed tomography (CT) to stratify pediatric RMPP risk at a critical decision point. Methods: Non-contrast chest CT data from a multicenter retrospective cohort of 1224 pediatric patients with Mycoplasma pneumoniae pneumonia who underwent clinically indicated CT were used to develop a transformer-based deep learning framework (trans-DLF). The primary cohort comprised training ( n = 506), validation ( n = 140), and internal testing ( n = 139) cohorts, with two independent external cohorts ( n = 331 and n = 108) used to evaluate generalizability. Model performance was assessed by the area under the receiver operating characteristic curve (AUC) and compared against a three-dimensional convolutional neural network (3D-CNN), a clinical model, and a multimodal nomogram. Interpretability was examined using gradient-weighted class activation mapping (Grad-CAM). Results: The median age was 6.83 years (interquartile range, 5.0–8.6 years), and 609 (49.8%) were male. The trans-DLF demonstrated strong performance across all cohorts: training (AUC 0.97; 95% confidence interval [CI], 0.96–0.98), validation (0.91; 0.86–0.96), internal testing (0.90; 0.85–0.95), and external testing (0.89; 0.84–0.94 and 0.89; 0.82–0.95). It significantly outperformed the clinical model ( p < 0.001), while its AUCs were not significantly different from those of the multimodal nomogram. The model maintained good performance in outpatient settings (AUC 0.87) with good calibration and net clinical benefit. Grad-CAM suggested that predictions were influenced by clinically meaningful features, particularly consolidations. Conclusion: The trans-DLF provides a streamlined and efficient approach to RMPP risk assessment in children who have already undergone clinically indicated chest CT and may support timely, evidence-based decision-making without additional tests.
- 17 July 2026
- Published by Springer Nature in BMC Medical Imaging
- Vol. 26 (1)
Open AccessA CT-based deep learning model for the automated risk stratification of refractory Mycoplasma pneumoniae pneumonia in children
Background: The accurate identification of children with refractory Mycoplasma pneumoniae pneumonia (RMPP) remains challenging. This study aimed to develop a transformer-based model utilizing clinically indicated chest computed tomography (CT) to stratify pediatric RMPP risk at a critical decision point. Methods: Non-contrast chest CT data from a multicenter retrospective cohort of 1224 pediatric patients with Mycoplasma pneumoniae pneumonia who underwent clinically indicated CT were used to develop a transformer-based deep learning framework (trans-DLF). The primary cohort comprised training ( n = 506), validation ( n = 140), and internal testing ( n = 139) cohorts, with two independent external cohorts ( n = 331 and n = 108) used to evaluate generalizability. Model performance was assessed by the area under the receiver operating characteristic curve (AUC) and compared against a three-dimensional convolutional neural network (3D-CNN), a clinical model, and a multimodal nomogram. Interpretability was examined using gradient-weighted class activation mapping (Grad-CAM). Results: The median age was 6.83 years (interquartile range, 5.0–8.6 years), and 609 (49.8%) were male. The trans-DLF demonstrated strong performance across all cohorts: training (AUC 0.97; 95% confidence interval [CI], 0.96–0.98), validation (0.91; 0.86–0.96), internal testing (0.90; 0.85–0.95), and external testing (0.89; 0.84–0.94 and 0.89; 0.82–0.95). It significantly outperformed the clinical model ( p < 0.001), while its AUCs were not significantly different from those of the multimodal nomogram. The model maintained good performance in outpatient settings (AUC 0.87) with good calibration and net clinical benefit. Grad-CAM suggested that predictions were influenced by clinically meaningful features, particularly consolidations. Conclusion: The trans-DLF provides a streamlined and efficient approach to RMPP risk assessment in children who have already undergone clinically indicated chest CT and may support timely, evidence-based decision-making without additional tests.
- 22 April 2026
- Published by Springer Nature in Nature Cancer
- Vol. 7 (5) , 773-789
DeepFAN, a transformer-based model for human–artificial intelligence collaborative assessment of incidental pulmonary nodules in CT scans: a multireader, multicase trial
The widespread adoption of computed tomography has increased the detection of lung nodules. However, deep learning methods for classification of benign and malignant nodules often fail to comprehensively integrate global and local features, and most of these methods have not been validated through clinical trials. Here we developed DeepFAN, a transformer-based model trained on more than 10,000 pathology-confirmed nodules, and conducted a multireader, multicase clinical trial (Chinese Clinical Trial Registry: ChiCTR2400084624) to evaluate its efficacy in assisting junior radiologists. DeepFAN achieved diagnostic area under the curve (AUC) values of 0.939 (95% CI 0.930–0.948) on an internal test set and 0.954 (95% CI 0.934–0.973) on a clinical trial dataset involving 400 cases across three independent medical institutions. Explainability analysis indicated higher contributions from global than local features. The average performance of 12 readers improved significantly: by 10.9% (95% CI 8.3–13.5%) for AUC, 10.0% (95% CI 8.9–11.1%) for accuracy, 7.6% (95% CI 6.1–9.2%) for sensitivity and 12.6% (95% CI 10.9–14.3%) for specificity (all P < 0.001). Nodule-level interreader diagnostic consistency improved from fair to moderate (overall κ: 0.313 versus 0.421; P = 0.019). These results indicate that DeepFAN can effectively assist junior radiologists and could help to homogenize diagnostic quality and reduce unnecessary follow-up of patients with indeterminate pulmonary nodules.
- 22 April 2026
- Published by Springer Nature in Nature Cancer
- Vol. 7 (5) , 773-789
DeepFAN, a transformer-based model for human–artificial intelligence collaborative assessment of incidental pulmonary nodules in CT scans: a multireader, multicase trial
The widespread adoption of computed tomography has increased the detection of lung nodules. However, deep learning methods for classification of benign and malignant nodules often fail to comprehensively integrate global and local features, and most of these methods have not been validated through clinical trials. Here we developed DeepFAN, a transformer-based model trained on more than 10,000 pathology-confirmed nodules, and conducted a multireader, multicase clinical trial (Chinese Clinical Trial Registry: ChiCTR2400084624) to evaluate its efficacy in assisting junior radiologists. DeepFAN achieved diagnostic area under the curve (AUC) values of 0.939 (95% CI 0.930–0.948) on an internal test set and 0.954 (95% CI 0.934–0.973) on a clinical trial dataset involving 400 cases across three independent medical institutions. Explainability analysis indicated higher contributions from global than local features. The average performance of 12 readers improved significantly: by 10.9% (95% CI 8.3–13.5%) for AUC, 10.0% (95% CI 8.9–11.1%) for accuracy, 7.6% (95% CI 6.1–9.2%) for sensitivity and 12.6% (95% CI 10.9–14.3%) for specificity (all P < 0.001). Nodule-level interreader diagnostic consistency improved from fair to moderate (overall κ: 0.313 versus 0.421; P = 0.019). These results indicate that DeepFAN can effectively assist junior radiologists and could help to homogenize diagnostic quality and reduce unnecessary follow-up of patients with indeterminate pulmonary nodules.
- Open Access
- 18 January 2026
- Published by Oxford University Press (OUP) in BJR|Open
- Vol. 8 (1)
Diffuse cystic lung diseases—a primer for radiologists
Diffuse cystic lung diseases (DCLDs) comprise a broad spectrum of disorders characterized by diffusely distributed pulmonary cysts, posing significant diagnostic challenges due to overlapping imaging features. High-resolution CT serves as the cornerstone for the diagnosis of DCLDs, yet its reliability is limited by technical constraints, including insufficient spatial resolution for subtle cysts and variable correlation with functional impairment. Furthermore, persistent discrepancies between imaging findings and underlying pathology highlight the inherent limitations of current modalities. This review synthesizes current diagnostic workflows, emphasizing the integration of imaging with clinical, genetic, and histopathological data within a multidisciplinary framework and appraising the impact of emerging technologies on diagnostic precision. Throughout this evolution, radiologists retain a central role in timely recognition, collaborative decision-making, and outcome optimization for patients with DCLDs.
- 18 January 2026
- Published by Oxford University Press (OUP) in BJR|Open
- Vol. 8 (1)
Open AccessDiffuse cystic lung diseases—a primer for radiologists
Diffuse cystic lung diseases (DCLDs) comprise a broad spectrum of disorders characterized by diffusely distributed pulmonary cysts, posing significant diagnostic challenges due to overlapping imaging features. High-resolution CT serves as the cornerstone for the diagnosis of DCLDs, yet its reliability is limited by technical constraints, including insufficient spatial resolution for subtle cysts and variable correlation with functional impairment. Furthermore, persistent discrepancies between imaging findings and underlying pathology highlight the inherent limitations of current modalities. This review synthesizes current diagnostic workflows, emphasizing the integration of imaging with clinical, genetic, and histopathological data within a multidisciplinary framework and appraising the impact of emerging technologies on diagnostic precision. Throughout this evolution, radiologists retain a central role in timely recognition, collaborative decision-making, and outcome optimization for patients with DCLDs.
- Open Access
- 16 December 2025
- Published by Springer Nature in Respiratory Research
- Vol. 27 (1)
A causal forest model integrating quantitative CT scores to predict benefit from flexible bronchoscopy in pediatric Mycoplasma pneumoniae pneumonia: a two-center retrospective study
Flexible bronchoscopy (FB) is recommended for pediatric Mycoplasma pneumoniae pneumonia (MPP) with persistent consolidation or atelectasis, though substant
- 16 December 2025
- Published by Springer Nature in Respiratory Research
- Vol. 27 (1)
Open AccessA causal forest model integrating quantitative CT scores to predict benefit from flexible bronchoscopy in pediatric Mycoplasma pneumoniae pneumonia: a two-center retrospective study
Flexible bronchoscopy (FB) is recommended for pediatric Mycoplasma pneumoniae pneumonia (MPP) with persistent consolidation or atelectasis, though substant
- Open Access
- 9 December 2025
- Published by Wolters Kluwer Health in International Journal of Surgery
- Vol. 112 (4) , 9213-9224
Machine learning-based quantitative prediction of spread through air spaces in primary lung adenocarcinoma using intratumoral heterogeneity scores
Background: Spread through air spaces (STAS) is a distinct, aggressive pattern of primary lung adenocarcinoma (LUAD) that affects both prognosis and treatment strategies for patients. This study aimed to quantify intratumoral heterogeneity (ITH) and integrate the quantitative metrics with intratumoral-peritumoral habitat features and clinical-radiologic characteristics to preoperatively predict the STAS status of primary LUAD and further explore the potential biological basis underlying the prediction model. Methods: Conventional radiomics features and habitat features were extracted from intratumoral and peritumoral regions on preoperative computerized tomography (CT) images. A new index, the ITH score, was developed to quantify ITH levels. Univariable and multivariable logistic regression analyses were conducted to identify clinical-radiologic characteristics associated with STAS. Various machine learning algorithms were used to build the prediction models. Additionally, intratumoral-peritumoral habitat features, ITH score, and clinical-radiologic characteristics were integrated into a combined model. Finally, 24 patients with RNA sequencing data were utilized for gene expression analysis. Results: A total of 1268 patients (median age, 60 years; IQR, 53.8–66.0 years; 850 female) were divided into the training set ( n = 943), validation set ( n = 236), and external test set ( n = 89). Using the Light Gradient Boosting Machine classifier, the combined model demonstrated the highest predictive performance for STAS, achieving an AUC value of 0.97 in the training, 0.98 in the validation, and 0.91 in the external test set. Differentially expressed genes in a high combined model probability group were associated with monocarboxylic acid transport and metabolism. Conclusion: The combined model demonstrated superior performance in predicting STAS in patients with primary LUAD.
- 9 December 2025
- Published by Wolters Kluwer Health in International Journal of Surgery
- Vol. 112 (4) , 9213-9224
Open AccessMachine learning-based quantitative prediction of spread through air spaces in primary lung adenocarcinoma using intratumoral heterogeneity scores
Background: Spread through air spaces (STAS) is a distinct, aggressive pattern of primary lung adenocarcinoma (LUAD) that affects both prognosis and treatment strategies for patients. This study aimed to quantify intratumoral heterogeneity (ITH) and integrate the quantitative metrics with intratumoral-peritumoral habitat features and clinical-radiologic characteristics to preoperatively predict the STAS status of primary LUAD and further explore the potential biological basis underlying the prediction model. Methods: Conventional radiomics features and habitat features were extracted from intratumoral and peritumoral regions on preoperative computerized tomography (CT) images. A new index, the ITH score, was developed to quantify ITH levels. Univariable and multivariable logistic regression analyses were conducted to identify clinical-radiologic characteristics associated with STAS. Various machine learning algorithms were used to build the prediction models. Additionally, intratumoral-peritumoral habitat features, ITH score, and clinical-radiologic characteristics were integrated into a combined model. Finally, 24 patients with RNA sequencing data were utilized for gene expression analysis. Results: A total of 1268 patients (median age, 60 years; IQR, 53.8–66.0 years; 850 female) were divided into the training set ( n = 943), validation set ( n = 236), and external test set ( n = 89). Using the Light Gradient Boosting Machine classifier, the combined model demonstrated the highest predictive performance for STAS, achieving an AUC value of 0.97 in the training, 0.98 in the validation, and 0.91 in the external test set. Differentially expressed genes in a high combined model probability group were associated with monocarboxylic acid transport and metabolism. Conclusion: The combined model demonstrated superior performance in predicting STAS in patients with primary LUAD.
- Open Access
- 1 November 2025
- Published by AME Publishing Company in Quantitative Imaging in Medicine and Surgery
- Vol. 15 (11) , 10409-10421
Quantitative analysis of pulmonary vascular alterations in children with refractory Mycoplasma pneumoniae pneumonia
Background: Mycoplasma pneumoniae pneumonia (MPP) is a major cause of community-acquired pneumonia (CAP) in children, with some cases progressing to refractory MPP (RMPP). RMPP is associated with a hypercoagulable state and pulmonary embolism. This study aimed to investigate pulmonary microvascular changes in RMPP and evaluate the predictive value of pulmonary blood volume (PBV) parameters. Methods: A retrospective study using UV-Net-based pulmonary vascular analysis included 512 pediatric MPP patients in a cross-validation cohort and 124 pediatric MPP patients in an external testing cohort. Pulmonary blood vessels were segmented and classified by cross-sectional area into three blood-volume fractions: BV5%, representing the percentage of PBV contained in vessels with a cross-sectional area less than 5 mm2; BV5–10%, representing the percentage of PBV contained in vessels with a cross-sectional area between 5 and 10 mm2; and BV10%, representing the percentage of PBV contained in vessels with a cross-sectional area greater than 10 mm2. Logistic regression and extreme gradient boosting were used to analyze associations and predict RMPP. Model performance was assessed via receiver operating characteristic (ROC) curve analysis. Results: Patients with RMPP, compared to patients with non-RMPP, had a significantly lower BV5% (median: 58.50% vs. 60.63%, P=0.007) and higher BV10% (median: 20.90% vs. 19.39%, P=0.004). Multivariate analysis revealed BV5% as a protective predictor [odds ratio (OR) =0.70, P=0.005] and BV10% as a risk factor for RMPP (OR =1.49, P=0.002). Compared with the clinical-only model, the model incorporating these computed tomography (CT)-derived parameters significantly improved performance in the cross-validation cohort, demonstrating superiority in terms of area under the ROC curve (AUC) and other metrics (combined model: AUC =0.91, 95% CI: 0.89–0.94; clinical-only model: AUC =0.88, 95% CI: 0.86–0.91; PConclusions: Quantitative analysis revealed microvascular alterations in patients with RMPP. Integrating CT-derived biomarkers can enhance RMPP prediction and facilitate early intervention.
- 1 November 2025
- Published by AME Publishing Company in Quantitative Imaging in Medicine and Surgery
- Vol. 15 (11) , 10409-10421
Open AccessQuantitative analysis of pulmonary vascular alterations in children with refractory Mycoplasma pneumoniae pneumonia
Background: Mycoplasma pneumoniae pneumonia (MPP) is a major cause of community-acquired pneumonia (CAP) in children, with some cases progressing to refractory MPP (RMPP). RMPP is associated with a hypercoagulable state and pulmonary embolism. This study aimed to investigate pulmonary microvascular changes in RMPP and evaluate the predictive value of pulmonary blood volume (PBV) parameters. Methods: A retrospective study using UV-Net-based pulmonary vascular analysis included 512 pediatric MPP patients in a cross-validation cohort and 124 pediatric MPP patients in an external testing cohort. Pulmonary blood vessels were segmented and classified by cross-sectional area into three blood-volume fractions: BV5%, representing the percentage of PBV contained in vessels with a cross-sectional area less than 5 mm2; BV5–10%, representing the percentage of PBV contained in vessels with a cross-sectional area between 5 and 10 mm2; and BV10%, representing the percentage of PBV contained in vessels with a cross-sectional area greater than 10 mm2. Logistic regression and extreme gradient boosting were used to analyze associations and predict RMPP. Model performance was assessed via receiver operating characteristic (ROC) curve analysis. Results: Patients with RMPP, compared to patients with non-RMPP, had a significantly lower BV5% (median: 58.50% vs. 60.63%, P=0.007) and higher BV10% (median: 20.90% vs. 19.39%, P=0.004). Multivariate analysis revealed BV5% as a protective predictor [odds ratio (OR) =0.70, P=0.005] and BV10% as a risk factor for RMPP (OR =1.49, P=0.002). Compared with the clinical-only model, the model incorporating these computed tomography (CT)-derived parameters significantly improved performance in the cross-validation cohort, demonstrating superiority in terms of area under the ROC curve (AUC) and other metrics (combined model: AUC =0.91, 95% CI: 0.89–0.94; clinical-only model: AUC =0.88, 95% CI: 0.86–0.91; PConclusions: Quantitative analysis revealed microvascular alterations in patients with RMPP. Integrating CT-derived biomarkers can enhance RMPP prediction and facilitate early intervention.
- Open Access
- 24 September 2025
- Published by MDPI AG in Biomedicines
- Vol. 13 (10) , 2333
Assessing the Heterogeneous Treatment Effects of Glucocorticoids in Infants and Toddlers with Severe Pneumonia
Background: Community-acquired pneumonia (CAP) is a leading cause of pediatric hospitalizations and a risk factor for chronic respiratory conditions. Glucocorticoids (GCs) are used as adjunctive therapy to reduce inflammation, but their efficacy in infants and toddlers remains unclear. Method: A retrospective study of 1116 infants and toddlers with severe CAP was conducted, using causal forest to estimate individual treatment effects (ITEs), with the duration of intensive care unit (ICU) stay as the outcome. Patients were stratified based on ITEs to investigate the heterogeneous treatment effect (HTE) and identify responders. Generalized linear models validated the HTE across subclasses, followed by comparative analyses to characterize responders. Variable importance was assessed using the causal model, and Shapley additive explanations (SHAP) quantified each variable’s contribution to the ITE. Analysis was also performed in mechanically ventilated patients (MV group). Results: GCs demonstrated significant HTE. Older patients and those with elevated inflammation markers showed better responses, whereas no such benefit was observed in respiratory syncytial virus-infected patients. These subgroups experienced shorter ICU stays both in the whole cohort (β = −0.16, p < 0.001) and MV group (β = −0.34, p < 0.001), and shorter ventilation duration was observed in the MV group (β = −0.35, p < 0.001). Age and the anion gap were key predictors of ITEs. SHAP analysis revealed a positive correlation between age and GC effectiveness. Conclusions: Significant heterogeneity in GC treatment effects exists among infants and toddlers with severe CAP, highlighting the need for optimization of GC use in this population.
- 24 September 2025
- Published by MDPI AG in Biomedicines
- Vol. 13 (10) , 2333
Open AccessAssessing the Heterogeneous Treatment Effects of Glucocorticoids in Infants and Toddlers with Severe Pneumonia
Background: Community-acquired pneumonia (CAP) is a leading cause of pediatric hospitalizations and a risk factor for chronic respiratory conditions. Glucocorticoids (GCs) are used as adjunctive therapy to reduce inflammation, but their efficacy in infants and toddlers remains unclear. Method: A retrospective study of 1116 infants and toddlers with severe CAP was conducted, using causal forest to estimate individual treatment effects (ITEs), with the duration of intensive care unit (ICU) stay as the outcome. Patients were stratified based on ITEs to investigate the heterogeneous treatment effect (HTE) and identify responders. Generalized linear models validated the HTE across subclasses, followed by comparative analyses to characterize responders. Variable importance was assessed using the causal model, and Shapley additive explanations (SHAP) quantified each variable’s contribution to the ITE. Analysis was also performed in mechanically ventilated patients (MV group). Results: GCs demonstrated significant HTE. Older patients and those with elevated inflammation markers showed better responses, whereas no such benefit was observed in respiratory syncytial virus-infected patients. These subgroups experienced shorter ICU stays both in the whole cohort (β = −0.16, p < 0.001) and MV group (β = −0.34, p < 0.001), and shorter ventilation duration was observed in the MV group (β = −0.35, p < 0.001). Age and the anion gap were key predictors of ITEs. SHAP analysis revealed a positive correlation between age and GC effectiveness. Conclusions: Significant heterogeneity in GC treatment effects exists among infants and toddlers with severe CAP, highlighting the need for optimization of GC use in this population.
- Open Access
- 30 May 2025
- Published by Springer Nature in BMC Medical Imaging
- Vol. 25 (1) , 1-12
Deep learning reconstruction improves computer-aided pulmonary nodule detection and measurement accuracy for ultra-low-dose chest CT
To compare the image quality and pulmonary nodule detectability and measurement accuracy between deep learning reconstruction (DLR) and hybrid iterative reconstruction (HIR) of chest ultra-low-dose CT (ULDCT). Participants who underwent chest standard-dose CT (SDCT) followed by ULDCT from October 2020 to January 2022 were prospectively included. ULDCT images reconstructed with HIR and DLR were compared with SDCT images to evaluate image quality, nodule detection rate, and measurement accuracy using a commercially available deep learning–based nodule evaluation system. Wilcoxon signed-rank test was used to evaluate the percentage errors of nodule size and nodule volume between HIR and DLR images. Eighty-four participants (54 ± 13 years; 26 men) were finally enrolled. The effective radiation doses of ULDCT and SDCT were 0.16 ± 0.02 mSv and 1.77 ± 0.67 mSv, respectively (P < 0.001). The mean ± standard deviation of the lung tissue noises was 61.4 ± 3.0 HU for SDCT, 61.5 ± 2.8 HU and 55.1 ± 3.4 HU for ULDCT reconstructed with HIR-Strong setting (HIR-Str) and DLR-Strong setting (DLR-Str), respectively (P < 0.001). A total of 535 nodules were detected. The nodule detection rates of ULDCT HIR-Str and ULDCT DLR-Str were 74.0% and 83.4%, respectively (P < 0.001). The absolute percentage error in nodule volume from that of SDCT was 19.5% in ULDCT HIR-Str versus 17.9% in ULDCT DLR-Str (P < 0.001). Compared with HIR, DLR reduced image noise, increased nodule detection rate, and improved measurement accuracy of nodule volume at chest ULDCT. Not applicable.
- 30 May 2025
- Published by Springer Nature in BMC Medical Imaging
- Vol. 25 (1) , 1-12
Open AccessDeep learning reconstruction improves computer-aided pulmonary nodule detection and measurement accuracy for ultra-low-dose chest CT
To compare the image quality and pulmonary nodule detectability and measurement accuracy between deep learning reconstruction (DLR) and hybrid iterative reconstruction (HIR) of chest ultra-low-dose CT (ULDCT). Participants who underwent chest standard-dose CT (SDCT) followed by ULDCT from October 2020 to January 2022 were prospectively included. ULDCT images reconstructed with HIR and DLR were compared with SDCT images to evaluate image quality, nodule detection rate, and measurement accuracy using a commercially available deep learning–based nodule evaluation system. Wilcoxon signed-rank test was used to evaluate the percentage errors of nodule size and nodule volume between HIR and DLR images. Eighty-four participants (54 ± 13 years; 26 men) were finally enrolled. The effective radiation doses of ULDCT and SDCT were 0.16 ± 0.02 mSv and 1.77 ± 0.67 mSv, respectively (P < 0.001). The mean ± standard deviation of the lung tissue noises was 61.4 ± 3.0 HU for SDCT, 61.5 ± 2.8 HU and 55.1 ± 3.4 HU for ULDCT reconstructed with HIR-Strong setting (HIR-Str) and DLR-Strong setting (DLR-Str), respectively (P < 0.001). A total of 535 nodules were detected. The nodule detection rates of ULDCT HIR-Str and ULDCT DLR-Str were 74.0% and 83.4%, respectively (P < 0.001). The absolute percentage error in nodule volume from that of SDCT was 19.5% in ULDCT HIR-Str versus 17.9% in ULDCT DLR-Str (P < 0.001). Compared with HIR, DLR reduced image noise, increased nodule detection rate, and improved measurement accuracy of nodule volume at chest ULDCT. Not applicable.