Translation of Quantitative Imaging Biomarkers into Clinical Chest CT

External validation, radiological evaluation, and development of deep learning automatic lung segmentation in contrast-enhanced chest CT

OBJECTIVES

There is a need for CT pulmonary angiography (CTPA) lung segmentation models. Clinical translation requires radiological evaluation of model outputs, understanding of limitations, and identification of failure points. This multicentre study aims to develop an accurate CTPA lung segmentation model, with evaluation of outputs in two diverse patient cohorts with pulmonary hypertension (PH) and interstitial lung disease (ILD).

METHODS

This retrospective study develops an nnU-Net-based segmentation model using data from two specialist centres (UK and USA). Model was trained (n = 37), tested (n = 12), and clinically evaluated (n = 176) on a diverse 'real-world' cohort of 225 PH patients with volumetric CTPAs. Dice score coefficient (DSC) and normalised surface distance (NSD) were used for testing. Clinical evaluation of outputs was performed by two radiologists who assessed clinical significance of errors. External validation was performed on heterogenous contrast and non-contrast scans from 28 ILD patients.

RESULTS

A total of 225 PH and 28 ILD patients with diverse demographic and clinical characteristics were evaluated. Mean accuracy, DSC, and NSD scores were 0.998 (95% CI 0.9976, 0.9989), 0.990 (0.9840, 0.9962), and 0.983 (0.9686, 0.9972) respectively. There were no segmentation failures. On radiological review, 82% and 71% of internal and external cases respectively had no errors. Eighteen percent and 25% respectively had clinically insignificant errors. Peripheral atelectasis and consolidation were common causes for suboptimal segmentation. One external case (0.5%) with patulous oesophagus had a clinically significant error.

CONCLUSION

State-of-the-art CTPA lung segmentation model provides accurate outputs with minimal clinical errors on evaluation across two diverse cohorts with PH and ILD.

CLINICAL RELEVANCE

Clinical translation of artificial intelligence models requires radiological review and understanding of model limitations. This study develops an externally validated state-of-the-art model with robust radiological review. Intended clinical use is in techniques such as lung volume or parenchymal disease quantification.

KEY POINTS

• Accurate, externally validated CT pulmonary angiography (CTPA) lung segmentation model tested in two large heterogeneous clinical cohorts (pulmonary hypertension and interstitial lung disease). • No segmentation failures and robust review of model outputs by radiologists found 1 (0.5%) clinically significant segmentation error. • Intended clinical use of this model is a necessary step in techniques such as lung volume, parenchymal disease quantification, or pulmonary vessel analysis.

Radiopaque Recreations of Lung Pathologies From Clinical Computed Tomography Images Using Potassium Iodide Inkjet 3-dimensional Printing: Proof of Concept

PURPOSE

The purpose of this study was to develop a 3-dimensional (3D) printing method to create computed tomography (CT) realistic phantoms of lung cancer nodules and lung parenchymal disease from clinical CT images.

MATERIALS AND METHODS

Low-density paper was used as substrate material for inkjet printing with potassium iodide solution to reproduce phantoms that mimic the CT attenuation of lung parenchyma. The relationship between grayscale values and the corresponding CT numbers of prints was first established through the derivation of exponential fitted equation from scanning data. Next, chest CTs from patients with early-stage lung cancer and coronavirus disease 2019 (COVID-19) pneumonia were chosen for 3D printing. CT images of original lung nodule and the 3D-printed nodule phantom were compared based on pixel-to-pixel correlation and radiomic features.

RESULTS

CT images of part-solid lung cancer and 3D-printed nodule phantom showed both high visual similarity and quantitative correlation. R2 values from linear regressions of pixel-to-pixel correlations between 5 sets of patient and 3D-printed image pairs were 0.92, 0.94, 0.86, 0.85, and 0.83, respectively. Comparison of radiomic measures between clinical CT and printed models demonstrated 6.1% median difference, with 25th and 75th percentile range at 2.4% and 15.2% absolute difference, respectively. The densities and parenchymal morphologies from COVID-19 pneumonia CT images were well reproduced in the 3D-printed phantom scans.

CONCLUSION

The 3D printing method presented in this work facilitates creation of CT-realistic reproductions of lung cancer and parenchymal disease from individual patient scans with microbiological and pathology confirmation.

Head-to-head Comparison of Qualitative Radiologist Assessment With Automated Quantitative Computed Tomography Analysis for Bronchiolitis Obliterans Syndrome After Hematopoietic Cell Transplantation

PURPOSE

Computed tomography (CT) findings of bronchiolitis obliterans syndrome (BOS) can be nonspecific and variable. This study aims to measure the incremental value of automated quantitative lung CT analysis to clinical CT interpretation. A head-to-head comparison of quantitative CT lung density analysis by parametric response mapping (PRM) with qualitative radiologist performance in BOS diagnosis was performed.

MATERIALS AND METHODS

Inspiratory and end-expiratory CTs of 65 patients referred to a post-bone marrow transplant lung graft-versus-host-disease clinic were reviewed by 3 thoracic radiologists for the presence of mosaic attenuation, centrilobular opacities, airways dilation, and bronchial wall thickening. Radiologists' majority consensus diagnosis of BOS was compared with automated PRM air trapping quantification and to the gold-standard diagnosis of BOS as per National Institutes of Health (NIH) consensus criteria.

RESULTS

Using a previously established threshold of 28% air trapping on PRM, the diagnostic performance for BOS was as follows: sensitivity 56% and specificity 94% (area under the receiver operator curve [AUC]=0.75). Radiologist review of inspiratory CT images alone resulted in a sensitivity of 80% and a specificity of 69% (AUC=0.74). When radiologists assessed both inspiratory and end-expiratory CT images in combination, the sensitivity was 92% and the specificity was 59% (AUC=0.75). The highest performance was observed when the quantitative PRM report was reviewed alongside inspiratory and end-expiratory CT images, with a sensitivity of 92% and a specificity of 73% (AUC=0.83).

CONCLUSIONS

In the CT diagnosis of BOS, qualitative expert radiologist interpretation was noninferior to quantitative PRM. The highest level of diagnostic performance was achieved by the combination of quantitative PRM measurements with qualitative image feature assessments.

Cardiac computed tomography radiomics: an emerging tool for the non-invasive assessment of coronary atherosclerosis

In the last decades, significant advances have been made in the preventive approaches to cardiovascular disease. Even so, coronary artery disease remains one of the main causes of morbidity and mortality worldwide. Invasive imaging modalities, such as intravascular ultrasound or optical coherence tomography, have played a key role in the comprehension of the pathological processes underlying myocardial infarction and cerebrovascular disease. These imaging techniques have contributed greatly to the identification and phenotyping of the culprit lesion, the so-called vulnerable plaque. Coronary computed tomographic angiography (CCTA) has emerged in more recent years as the non-invasive modality of choice in the study of coronary atherosclerosis, showing in many studies a diagnostic yield comparable to invasive approaches. Moreover, being able to describe extra-luminal characteristics of the affected vessel, CCTA has greatly contributed towards shifting the attention of researchers from the mere quantification of luminal stenosis to the identification of adverse plaque features, which appear to have a stronger prognostic value. However, the identification of some of the hallmarks of vulnerable plaques is qualitative in nature and, therefore, subject to some degree of inter-reader variability. Moreover, CCTA is still unable to identify some fine markers of plaque vulnerability which can be detected by invasive techniques, such as neovascularization and plaque erosion, among others. Nonetheless, radiological images can be viewed as vast 3-D datasets which, via the use of recent technology, allow for the extraction of numerous quantitative features that may be used to accurately phenotype a given lesion. Radiomics is the process of extrapolating innumerable parameters from a given region of interest, with the goal of establishing correlations between quantitative variables and clinical data. These datasets can then be manipulated to create predictive models via the use of automated algorithms in a process called machine learning. As a result of these approaches, radiological images may offer information regarding the characterization of a plaque which can go much beyond the boundaries of what can be qualitatively asserted by the human eye, contributing to expanding the knowledge of the disease and ultimately assist clinical decisions. Thus far, radiomics has found its more consistent area of application in the field of oncology; to present date, the amount of clinical data regarding coronary artery disease is still relatively small, partly due to the technical difficulties associated with the implementation of such techniques to the study of a small and geometrically complex lesion such as the coronary plaque. The present review, after a summary of the imaging modalities most commonly used nowadays in the study of coronary plaques, will provide a perspective on the application of radiomic analysis to coronary artery disease.

CT Phantom Evaluation of 67,392 American College of Radiology Accreditation Examinations: Implications for Opportunistic Screening of Osteoporosis Using CT

OBJECTIVE. The purpose of this study was to investigate whether systematic bias in attenuation measurements occurs among CT scanners made by four major manufacturers and the relevance of this bias regarding opportunistic screening for osteoporosis. MATERIALS AND METHODS. Data on attenuation measurement accuracy were acquired using the American College of Radiology (ACR) accreditation phantom and were evaluated in a blinded fashion for four CT manufacturers (8500 accreditation submissions for manufacturer A; 18,575 for manufacturer B; 8278 for manufacturer C; and 32,039 for manufacturer D). The attenuation value for water, acrylic (surrogate for trabecular bone), and Teflon (surrogate for cortical bone; Chemours) materials for an adult abdominal CT technique (120 kV, 240 mA, standard reconstruction algorithm) was used in the analysis. Differences in attenuation value across all manufacturers were assessed using the Kruskal-Wallis test followed by a post hoc test for pairwise comparisons. RESULTS. The mean attenuation value for water ranged from -0.3 to 2.7 HU, with highly significant differences among all manufacturers (p < 0.001). For the trabecular bone surrogate, differences in attenuation values across all manufacturers were also highly significant (p < 0.001), with mean values of 120.9 (SD, 3.5), 124.6 (3.3), 126.9 (4.4), and 123.9 (3.4) HU for manufacturers A, B, C, and D, respectively. For the cortical bone surrogate, differences in attenuation values across all manufacturers were also highly significant (p < 0.001), with mean values of 939.0 (14.2), 874.3 (13.3), 897.6 (11.3), and 912.7 (13.4) HU for manufacturers A, B, C, and D, respectively. CONCLUSION. CT scanners made by different manufacturers show systematic offsets in attenuation measurement when compared with each other. Knowledge of these off-sets is useful for optimizing the accuracy of opportunistic diagnosis of osteoporosis.