Dataset Information

Diagnosis of Idiopathic Pulmonary Fibrosis in High-Resolution Computed Tomography Scans Using a Combination of Handcrafted Radiomics and Deep Learning.

ABSTRACT:

Purpose

To develop handcrafted radiomics (HCR) and deep learning (DL) based automated diagnostic tools that can differentiate between idiopathic pulmonary fibrosis (IPF) and non-IPF interstitial lung diseases (ILDs) in patients using high-resolution computed tomography (HRCT) scans.

Material and methods

In this retrospective study, 474 HRCT scans were included (mean age, 64.10 years ± 9.57 [SD]). Five-fold cross-validation was performed on 365 HRCT scans. Furthermore, an external dataset comprising 109 patients was used as a test set. An HCR model, a DL model, and an ensemble of HCR and DL model were developed. A virtual in-silico trial was conducted with two radiologists and one pulmonologist on the same external test set for performance comparison. The performance was compared using DeLong method and McNemar test. Shapley Additive exPlanations (SHAP) plots and Grad-CAM heatmaps were used for the post-hoc interpretability of HCR and DL models, respectively.

Results

In five-fold cross-validation, the HCR model, DL model, and the ensemble of HCR and DL models achieved accuracies of 76.2 ± 6.8, 77.9 ± 4.6, and 85.2 ± 2.7%, respectively. For the diagnosis of IPF and non-IPF ILDs on the external test set, the HCR, DL, and the ensemble of HCR and DL models achieved accuracies of 76.1, 77.9, and 85.3%, respectively. The ensemble model outperformed the diagnostic performance of clinicians who achieved a mean accuracy of 66.3 ± 6.7% (p < 0.05) during the in-silico trial. The area under the receiver operating characteristic curve (AUC) for the ensemble model on the test set was 0.917 which was significantly higher than the HCR model (0.817, p = 0.02) and the DL model (0.823, p = 0.005). The agreement between HCR and DL models was 61.4%, and the accuracy and specificity for the predictions when both the models agree were 93 and 97%, respectively. SHAP analysis showed the texture features as the most important features for IPF diagnosis and Grad-CAM showed that the model focused on the clinically relevant part of the image.

Conclusion

Deep learning and HCR models can complement each other and serve as useful clinical aids for the diagnosis of IPF and non-IPF ILDs.

SUBMITTER: Refaee T

PROVIDER: S-EPMC9259876 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BackgroundIdiopathic pulmonary fibrosis (IPF) is a fatal lung disease involving progressive degeneration of lung capacity. Current diagnosis of IPF heavily relies on visual evaluation of fibrotic features in high-resolution computed tomography (HRCT) images of the lungs. Although the characteristics of this disease have been studied at the molecular and cellular levels, little is known about the mechanical characteristics of IPF lungs inferred from HRCT images. To this end, we performed a pilot study to investigate the radiographic and volumetric characteristics of lungs in subjects with IPF.MethodsWe collected HRCT images of healthy (N = 13) and IPF (N = 9) lungs acquired at breath-holds after full inspiration (expanded state) and full expiration (contracted state). We performed statistical analyses on Hounsfield unit (HU) histograms, lobar volumes (V: lobe volume normalized by the lung volume), and lobar flows (Q: the difference in lobe volume divided by the difference in lung volume between the expanded and contracted states).ResultsParameters characterizing the HU histograms (i.e., mean, median, skewness, and kurtosis) significantly differed between healthy and IPF subjects, for all lobes in both expanded and contracted states. The distribution of V across lobes differed significantly between the groups in both states. The distribution of Q also differed significantly between the groups: Q values of the lower lobes for the IPF group were 33% (right) and 22% (left) smaller than those for the healthy group, consistent with the observation that radiographic scores were highest in the lower lung section in IPF. Notably, the root-mean-squared difference (RMSD) of Q, a measure of distance from the mean value of the healthy group, clearly distinguished the IPF subjects (RMSD of Q > 1.59) from the healthy group (RMSD of Q < 0.67).ConclusionThis study shows that lung volume and flow distribution change heterogeneously across the lung lobes of IPF subjects, with reduced capacity in the lower lobes. These volumetric changes may improve our understanding of the pathophysiology in IPF lungs.