HRCT characteristics of severe emphysema patients: Interobserver variability among expert readers and comparison with quantitative software

Quantitative analysis of interstitial lung abnormalities on computed tomography to predict symptomatic radiation pneumonitis after lung stereotactic body radiotherapy

BACKGROUND AND PURPOSE

Symptomatic radiation pneumonitis (SRP) is a complication of thoracic stereotactic body radiotherapy (SBRT). As visual assessments pose limitations, artificial intelligence-based quantitative computed tomography image analysis software (AIQCT) may help predict SRP risk. We aimed to evaluate high-resolution computed tomography (HRCT) images with AIQCT to develop a predictive model for SRP.

MATERIALS AND METHODS

AIQCT automatically labelled HRCT images of patients treated with SBRT for stage I lung cancer according to lung parenchymal pattern. Quantitative data including the volume and mean dose (Dmean) were obtained for reticulation + honeycombing (Ret + HC), consolidation + ground-glass opacities, bronchi (Br), and normal lungs (NL). After associations between AIQCT's quantified metrics and SRP were investigated, we developed a predictive model using recursive partitioning analysis (RPA) for the training cohort and assessed its reproducibility with the testing cohort.

RESULTS

Overall, 26 of 207 patients developed SRP. There were significant between-group differences in the Ret + HC, Br-volume, and NL-Dmean in patients with and without SRP. RPA identified the following risk groups: NL-Dmean ≥ 6.6 Gy (high-risk, n = 8), NL-Dmean < 6.6 Gy and Br-volume ≥ 2.5 % (intermediate-risk, n = 13), and NL-Dmean < 6.6 Gy and Br-volume < 2.5 % (low-risk, n = 133). The incidences of SRP in these groups within the training cohort were 62.5, 38.4, and 7.5 %; and in the testing cohort 50.0, 27.3, and 5.0 %, respectively.

CONCLUSION

AIQCT identified CT features associated with SRP. A predictive model for SRP was proposed based on AI-detected Br-volume and the NL-Dmean.

Radiomics-Assisted Computed Tomography-Based Analysis to Evaluate Lung Morphology Characteristics after Congenital Diaphragmatic Hernia

Purpose: Children with congenital diaphragmatic hernia suffer from long-term morbidity, including lung function impairment. Our study aims to analyze lung morphology characteristics via radiomic-assisted extraction of lung features in patients after congenital diaphragmatic hernia repair. Materials and Methods: 72 patients were retrospectively analyzed after approval by the local research ethics committee. All the image data were acquired using a third-generation dual-source CT (SOMATOM Force, Siemens Healthineers, Erlangen, Germany). Dedicated software was used for image analysis, segmentation, and processing. Results: Radiomics analysis of pediatric chest CTs of patients with status after CDH was possible. Between the ipsilateral (side of the defect) and contralateral lung, three shape features and two higher-order texture features were considered statistically significant. Contralateral lungs in patients with and without ECMO treatment showed significant differences in two shape features. Between the ipsilateral lungs in patients with and without the need for ECMO 1, a higher-order texture feature was depicted as statistically significant. Conclusions: By adding quantitative information to the visual assessment of the radiologist, radiomics-assisted feature analysis could become an additional tool in the future to assess the degree of lung hypoplasia in order to further improve the therapy and outcome of CDH patients.

Response to endobronchial valve treatment: it's all about the target lobe

Background

Bronchoscopic lung volume reduction using endobronchial valves (EBV) has been shown to be beneficial for severe emphysema patients. The most important predictor of treatment response is absence of collateral ventilation between the treatment target and ipsilateral lobe. However, there are still a substantial number of nonresponders and it would be useful to improve the pre-treatment identification of responders. Presumably, predictors of response will be multifactorial, and therefore our aim was to explore whether we can identify response groups using a cluster analysis.

Methods

At baseline and 1 year follow-up, pulmonary function, exercise capacity and quality of life were measured. A quantitative chest computed tomography scan analysis was performed at baseline and 2-6 months follow-up. The cluster analysis was performed using a hierarchical agglomerative method.

Results

In total, 428 patients (69% female, mean±sd age 61±8 years, forced expiratory volume in 1 s 27±8% predicted, residual volume 254±50% pred) were included in our analysis. Three clusters were generated: one nonresponder cluster and two responder clusters. Despite solid technical procedures, the nonresponder cluster had significantly less clinical response after treatment compared to the other clusters. The nonresponder cluster was characterised by significantly less emphysematous destruction, less air trapping and a higher perfusion of the target lobe, and a more homogeneous distribution of emphysema and perfusion between the target and ipsilateral lobe.

Conclusions

We found that target lobe characteristics are the discriminators between responders and nonresponders, which underlines the importance of visual and quantitative assessment of the potential treatment target lobe when selecting patients for EBV treatment.

The impact of emphysema heterogeneity on treatment response after endobronchial valve treatment

Background

Lung volume reduction with endobronchial valves can significantly improve functional outcomes in patients with advanced emphysema. The extent and spatial distribution pattern of emphysema shows considerable heterogeneity, which might affect response to endobronchial valve treatment. Our aim was to study the effect of emphysema heterogeneity on change in clinical outcomes after endobronchial valve treatment.

Methods

Data were collected from our national registry of patients who received endobronchial valve treatment between 2016 and 2020. We assessed the association between the heterogeneity index, absolute difference in destruction between the target and ipsilateral lobe, and relative change in forced expiratory volume in 1 s (FEV1), residual volume (RV), St George's Respiratory Questionnaire (SGRQ) and 6-min walk distance (6MWD) at 6-week, 6-month and 12-month follow-up.

Results

In total, 236 patients were included. Heterogeneity index was significantly associated with improvements in FEV1, RV and 6MWD at all follow-up visits, and in SGRQ at the 6- and 12-month follow-up visits. The majority of patients, independent of heterogeneity index, showed clinically meaningful improvements based on minimal important difference thresholds.

Conclusions

Heterogeneity index influences the degree of clinical improvement after endobronchial valve treatment where in general a more heterogeneous distribution translates to larger improvements. However, patients with a more homogeneous distribution also showed clinically meaningful improvements. Therefore, we believe emphysema heterogeneity alone should not be used as a decisive patient selection criterion, but should be weighed in the context of all other relevant patient and target lobe characteristics when deciding on a patient's treatment eligibility.

Evaluation of emphysema on thoracic low-dose CTs through attention-based multiple instance deep learning

In addition to lung cancer, other thoracic abnormalities, such as emphysema, can be visualized within low-dose CT scans that were initially obtained in cancer screening programs, and thus, opportunistic evaluation of these diseases may be highly valuable. However, manual assessment for each scan is tedious and often subjective, thus we have developed an automatic, rapid computer-aided diagnosis system for emphysema using attention-based multiple instance deep learning and 865 LDCTs. In the task of determining if a CT scan presented with emphysema or not, our novel Transfer AMIL approach yielded an area under the ROC curve of 0.94 ± 0.04, which was a statistically significant improvement compared to other methods evaluated in our study following the Delong Test with correction for multiple comparisons. Further, from our novel attention weight curves, we found that the upper lung demonstrated a stronger influence in all scan classes, indicating that the model prioritized upper lobe information. Overall, our novel Transfer AMIL method yielded high performance and provided interpretable information by identifying slices that were most influential to the classification decision, thus demonstrating strong potential for clinical implementation.