Radiomic Measures from Chest HRCT Associated with Lung Function in Sarcoidosis

Abstract

Introduction.

Pulmonary sarcoidosis is a rare heterogenous lung disease of unknown etiology, with limited treatment options. Phenotyping relies on clinical testing including visual scoring of chest radiographs. Objective radiomic measures from high resolution computed tomography (HRCT) may provide additional information to assess disease status. As the first radiomics analysis in sarcoidosis, we investigate the potential of radiomics as biomarkers for sarcoidosis, by assessing (1) differences in HRCT between sarcoidosis and controls, (2) associations between radiomic measures and spirometry, and (3) trends between Scadding stages.

Methods.

Radiomic features were computed on HRCT in three anatomical planes. Linear regression compared global radiomic features between sarcoidosis (N=73) and controls (N=78) and identified associations with spirometry. Spatial differences in associations across the lung were investigated using functional data analysis. A sub-analysis compared radiomic features between Scadding stages.

Results.

Global radiomic measures differed significantly between sarcoidosis and control (p<0.001 for skewness, kurtosis, fractal dimension and Geary’s C), with differences in spatial radiomics most apparent in superior and lateral regions. In sarcoidosis subjects, there were significant associations between radiomic measures and spirometry, with a large association found between Geary’s C and forced vital capacity (FVC) (p=0.008). Global radiomic measures differed significantly between Scadding stages (p<0.032), albeit non-linearly, with stage IV having more extreme radiomic values. Radiomics explained 71.1% of the variability in FVC compared to 51.4% with Scadding staging alone.

Conclusions.

Radiomic HRCT measures objectively differentiate disease abnormalities, associate with lung function, and identify trends in Scadding stage, showing promise as quantitative biomarkers for pulmonary sarcoidosis.

Summary “take home” message

Radiomic measures identify pulmonary parenchymal abnormalities in sarcoidosis and are highly associated with lung function, suggesting that radiomics could enhance visual reads and result in improved patient profiling, disease staging and monitoring.

INTRODUCTION

Pulmonary sarcoidosis is a rare heterogenous disease of unknown etiology characterized by the formation of granulomas in the lungs, as well as other organs. Compromised lung function is common with pulmonary involvement¹, which can limit daily activities, and frequently indicates the need for therapy to reduce or reverse abnormalities. Spontaneous remission has been quoted to occur in up to two thirds of patients in certain populations². Yet, mortality rates due to respiratory failure are rising in the US³.

Assessment of chest radiography, in addition to lung function, is used to clinically monitor sarcoidosis, with chest radiographs showing abnormalities in 90% of cases⁴. The Scadding system of staging, based on chest radiographs, is a common visual classification for sarcoidosis, dividing the radiographic manifestations into five stages: stage 0 (no radiograph abnormalities), stage I (bilateral hilar lymphadenopathy, BHL), stage II (pulmonary infiltration with BHL), stage III (pulmonary infiltration without BHL), and stage IV (pulmonary fibrosis with volume loss). Despite the numerical nomenclature, there is no sequential ordering of Scadding stages, and subjects may demonstrate abnormalities in multiple stages at various times. Additional chest radiographic patterns may be observed that are not well characterized by Scadding stage, such as small nodular opacities along the bronchovascular bundle, focal consolidation, consolidation of small nodular opacities (conglomerate masses), ground-glass, and fibrosis, among others².

Recently, high-resolution computed tomography (HRCT) scans have been used to characterize pulmonary sarcoidosis manifestations^5–7. Despite limited standard visual assessment tools for CT sarcoidosis manifestations^8–10, CT findings are still visually characterized, which is time-consuming, dependent on the expertise of the reader, and subject to poor inter-rater reliability¹¹. Quantitative assessment of HRCT in sarcoidosis may provide a more rapid, objective, and sensitive quantification of the various abnormalities that present in sarcoidosis.

Radiomics, an emerging field in which large numbers of quantitative imaging features are computed from medical images, has proved useful for developing quantitative biomarkers in emphysema^12–15interstitial lung disease^16–18and lung cancer^19–23. To our knowledge, radiomic measures have yet to be explored in sarcoidosis. Given the usefulness of radiomics in other diseases, we hypothesize that radiomic features will also prove useful as potential imaging biomarkers for sarcoidosis.

In this paper, we perform an early stage investigation of radiomics as a potential biomarker in sarcoidosis, where early stage references the biomarker study pipeline suggested by Pepe²⁴. Our goal is to investigate whether specific radiomic features on lung HRCT differ between sarcoidosis cases and healthy controls (Figure 1 and Table 1). In particular, we compute global and spatially-varying radiomic features, and assess how these measures differentiate sarcoidosis (N=73) from healthy controls (N=78), and between Scadding stages. Lastly, we relate radiomics and lung function to show the potential of radiomics as biomarkers for sarcoidosis.

Figure 1.

Representative HRCT scans from subjects representative of HRCT abnormalities, including no abnormalities (S1), mosaic attenuation/honeycombing (S2), nodules (S3), and fibrosis (S4). In general, a two-dimensional HRCT slice from a healthy subject (S1) appears to have mostly healthy lung tissue, with the occasional blood vessel or airway on the slice. Conversely, the HRCT slices from subject with sarcoidosis (S2-S4) have increased opacification and parenchymal abnormalities apparent, shown by the whiter areas on the CT scan. These parenchymal abnormalities on the CT scan for subjects S2-S4 alter the appearance of the Hounsfield unit (HU) histogram, resulting in less skewness and less kurtosis (“peaked”) densities, as compared to the healthy subject, S1. Further, the spatial radiomic measures also differ, with sarcoidosis subjects having smaller fractal dimension and Geary’s C, and larger Moran’s I, as compared to the healthy subject.

Table 1.

Description of radiomic features for two-dimensional HRCT slices

Measure and Equation*	Description	Clinical Hypothesis
Skewness $\frac{\frac{1}{n}{\sum }_{i=1}^n{\left(x_i-\overline{x}\right)}^3}{{\left[\frac{1}{n-1}{\sum }_{i=1}^n{\left(x_i-\overline{x}\right)}^2\right]}^{\raisebox{1ex}{$3$}\!\left/ \!\raisebox{-1ex}{$2$}\right.}}$	First-order histogram feature that measures the asymmetry of a sample distribution. Positive values indicate right-skew, or a long right tail. Values closer to 0 indicate little or no skew.	Sarcoidosis results in increased opacification (i.e. whiter regions) on HRCT from parenchymal abnormalities. This will alter the properties of the histogram of HU units to appear more normally distributed (i.e. less skew and less kurtosis). Thus, we hypothesize that sarcoidosis subjects will have less skewness and less kurtosis as compared to healthy controls.
Kurtosis $\frac{\frac{1}{n}{\sum }_{i=1}^n{\left(x_i-\overline{x}\right)}^4}{{\left[\frac{1}{n}{\sum }_{i=1}^n{\left(x_i-\overline{x}\right)}^2\right]}^2}-3$	First-order histogram feature that measures the tailed-ness of a sample distribution. Positive values indicate infrequent, extreme outliers. Smaller values indicate fewer, less extreme outliers.
Fractal Dimension $1+median\left[2-\frac{\log V\left(2\right)-\log V\left(1\right)}{\log 2}\right]$ $V\left(k\right)=\frac{{\sum }_{i=1}^n{\sum }_{j=1}^n{w}_{i,j}\left(k\right)\left|x_i-x_j\right|}{{\sum }_{i=1}^n{\sum }_{j=1}^n{w}_{i,j}\left(k\right)}$	Second-order texture feature that measures the self-similarity of pixels in space, or the image roughness29. On a two-dimensional plane, it ranges from 2 to 3, with lower values indicative of adjacent pixels appearing more similar (i.e. more smoothness).	Sarcoidosis involves the formation of micronodules in the lung that may conglomerate as disease worsens, and/or fibrosis may develop. Conglomeration and fibrosis are both represented by adjacent pixels of higher opacification on HRCT (i.e. more smoothness). Thus, we hypothesize that sarcoidosis subjects will have lower fractal dimension, higher Moran’s I and lower Geary’s C as compared to healthy controls.
Moran’s I $\frac{n{\sum }_{i=1}^n{\sum }_{j=1}^n{w}_{i,j}\left(x_i-\overline{x}\right)\left(x_j-\overline{x}\right)}{{\sum }_{i=1}^n{\sum }_{j=1}^n{w}_{i,j}\left(1\right){\sum }_{i=1}^n{\left(x_i-\overline{x}\right)}^2}$	Second-order spatial feature that measures global spatial autocorrelation, or similarity of pixels in space30. It ranges approximately from −1 to 1, with values around zero indicative of no spatial autocorrelation and higher values indicative of positive spatial autocorrelation, or adjacent pixels appearing more similar.
Geary’s C $\frac{n{\sum }_{i=1}^n{\sum }_{j=1}^n{w}_{i,j}{\left(x_i-x_j\right)}^2}{{\sum }_{i=1}^n{\sum }_{j=1}^n{w}_{i,j}\left(1\right){\sum }_{i=1}^n{\left(x_i-\overline{x}\right)}^2}$	Second-order spatial feature that measures local spatial autocorrelation31. Geary’s C, ranging from 0 to 2, is inversely proportional to Moran’s I, with values close to zero indicating strong positive spatial autocorrelation, or adjacent pixels appearing more similar.

^*$x_i$ is the HU intensity at a single pixel, i; n is the total number of pixels and $\overline{x}$ is the mean HU from a lung-masked HRCT slice; $w_{i,j}\left(k\right)$ is an indicator function for whether the distance between pixels i and j is k units

METHODS

Study Populations and Data Acquisition

Online Data Supplement Section E1 includes a detailed description of the study populations and imaging acquisition. In brief, the sarcoidosis population (N=79) was recruited at National Jewish Health (NJH) as part of the NHLBI funded Genomic Research in Alpha-1 Antitrypsin Deficiency and Sarcoidosis (GRADS) study²⁵. A non-smoking, healthy control population (N=108) was obtained from the COPDGene study²⁶. Subjects in GRADS were between 18 and 85 years old, had a confirmed diagnosis of sarcoidosis via biopsy or manifestations consistent with acute sarcoidosis (Lofgren’s syndrome)²⁵. Subjects in the COPDGene study were between 45 and 80 years old, with no history of lung disease and normal postbronchodilator spirometry²⁷. In both GRADS and COPDGene, pulmonary function testing was obtained (PFT, including pre-bronchodilator (BD) forced expiratory volume at one second (FEV1) and forced vital capacity (FVC)) and a chest HRCT ordered according to the same imaging protocol²⁶, except for a difference in tube current average (GRADS: 180–330 mA varied by BMI; COPDGene: 400 mA); according to Mackin²⁸, we do not believe the difference in tube current will affect our results. Seventy-three of the GRADS subjects had an HRCT. Only COPDGene subjects scanned on the same machine manufacturer as the GRADS subjects were analyzed in this study (N=78). All subjects provided signed informed consent through either the GRADS or COPDGene studies.

Radiomic Measures

Image data pre-processing and radiomics is described in Online Data Supplement Sections E2 and E3. In brief, lung segmentation and radiomic calculation were performed using the lungct R package (https://github.com/ryansar/lungct). Five radiomic measures (skewness, kurtosis, fractal dimension²⁹, Moran’s I³⁰ and Geary’s C³¹) were pre-selected for evaluation in our study based on their statistical properties and success in previous studies for idiopathic pulmonary fibrosis and lung nodules^32,20,33. The five radiomic measures were computed on every slice for every subject in each of the two lungs and three anatomical planes. To summarize the distribution of the radiomic features obtained for each subject, lung and plane, the median value was computed. Detailed descriptions of each measure along with directional clinical hypotheses for each measure can be found in Figure 1 and Table 1.

Statistical Analysis

A detailed statistical analysis can be found in Online Data Supplement Section E4. Descriptive statistics (e.g. mean and standard deviation for continuous variables; frequencies for categorical variables) were used to summarize demographic, radiomic, and spirometry data (Table 2). Linear regression models were used to test for differences in global radiomic measures between sarcoidosis and healthy controls, adjusted for age, gender, and BMI (Table 3 and Table E1, E2). To evaluate whether disease status (i.e. sarcoidosis vs. control) alters associations between lung function and global radiomic measures, linear regressions were fitted with an interaction between disease group and a global radiomic measure (Figure 2 and Tables E3, E4). In secondary analyses, functional regressions using penalized smoothing splines³⁴ were fitted to determine whether there was a spatially-varying association between subjects with and without sarcoidosis and a radiomic measure; that is, whether certain regions of the lung were more different between sarcoidosis and control than other regions (Figures 3, 4 and Figures E1, E2).

Table 2.

Differences in subject characteristics by case and control status and Scadding stage. Categorical variables are summarized by the frequency (%); continuous variables are summarized by mean (standard deviation). P-values for lung function measures (pre-BD FEV1, FVC, and the ratio) are adjusted for gender, age, race, ethnicity, and height.

	Control	Sarcoidosis	P-value	Stage 0	Stage I	Stage II	Stage III	Stage IV	P-value
	(N = 78)	(N = 73)		(N = 9)	(N = 8)	(N = 28)	(N = 11)	(N = 17)
Male (%)	28 (35.9)	36(49.3)	0.133	2 (22.2%)	3 (37.5%)	14 (50.0%)	7 (63.6%)	10 (58.8%)	0.329
White (%)	77 (98.7)	62 (84.9)	0.002	7 (77.8%)	8 (100.0%)	24 (85.7%)	11 (100.0%)	12 (70.6%)	0.166
Age (years)	64.51 (8.32)	54.14 (8.64)	<0.001	51.44 (7.19)	52.27 (13.79)	53.76 (9.31)	56.93 (6.81)	55.28 (6.24)	0.612
BMI	27.97 (4.76)	30.12 (6.95)	0.029	36.68 (7.20)	29.80 (5.99)	28.55 (5.72)	30.28 (8.74)	29.27 (6.58)	0.040
Pre-BD FEV1 (L)	2.76 (0.71)	2.81 (0.95)	<0.001	2.86 (0.85)	2.98 (1.07)	2.85 (1.05)	3.02 (0.80)	2.39 (0.82)	0.055
Pre-BD FVC (L)	3.61 (0.90)	3.72 (1.20)	<0.001	3.45 (1.03)	3.69 (1.31)	3.87 (1.40)	3.92 (1.06)	3.43 (0.95)	0.140
FEV1:FVC ratio	0.77 (0.05)	0.75 (0.09)	0.465	0.83 (0.03)	0.81 (0.08)	0.74 (0.08)	0.78 (0.06)	0.68 (0.10)	<0.001

Table 3.

Differences in global radiomic features between cases and controls and by Scadding stage in the left lung, axial plane. Values are summarized by mean (standard errors). P-values are adjusted for gender, age, and BMI. For right lung and coronal and sagittal results, see Online Data Supplement Table E1 and E2.

	Control	Sarcoidosis	P-value	Stage 0	Stage I	Stage II	Stage III	Stage IV	P-value
	(N = 78)	(N = 73)		(N = 9)	(N = 8)	(N = 28)	(N = 11)	(N = 17)
Skewness	3.642 (0.060)	3.166 (0.085)	<0.001	3.009 (0.131)	3.491 (0.198)	3.309 (0.150)	3.373 (0.187)	2.727 (0.172)	0.027
Kurtosis	16.653 (0.600)	12.732 (0.688)	<0.001	10.879 (1.102)	15.052 (2.003)	14.084 (1.259)	14.248 (1.741)	9.413 (1.066)	0.032
Fractal D	2.429 (0.003)	2.404 (0.004)	<0.001	2.434 (0.011)	2.437 (0.006)	2.397 (0.004)	2.421 (0.005)	2.374 (0.008)	<0.001
Moran’s I	0.691 (0.004)	0.702 (0.005)	0.047	0.667 (0.013)	0.668 (0.013)	0.715 (0.005)	0.680 (0.007)	0.729 (0.008)	<0.001
Geary’s C	0.226 (0.002)	0.212 (0.002)	<0.001	0.226 (0.006)	0.227 (0.005)	0.208 (0.003)	0.229 (0.005)	0.195 (0.002)	<0.001

Figure 2.

Association between lung function and global radiomic features by disease status, adjusted for gender, age, and BMI. P-values on subfigures test for significant associations between lung function and global radiomic features for both sarcoidosis (blue lines) and healthy controls (green lines). Shaded bars represent 95% confidence bands, predicted for a female subject, with a mean age of 59.5 years and mean BMI of 29.0; points represent raw data values per subject, colored by disease group. The global radiomic features are calculated on the left lung, axial orientation. For the right lung, and other orientations, see Online Data Supplement Tables E2 and E3.

Figure 3.

Mean radiomic features throughout the lung for sarcoidosis and healthy controls. Shaded bars represent 95% confidence bands; individual lines represent raw radiomic features throughout the lung per individual, colored by disease group. Results are shown for the left lung and all orientations. For right lung results, see Online Data Supplement Figure E1.

Figure 4.

Effect size of the absolute difference in radiomic features throughout the lung between sarcoidosis and healthy control subjects, adjusted for gender, age, and BMI. Assuming a normal approximation, values above 1.96 represent statistically significant differences at a significance level of 0.05. Results are shown for the left lung. For right lung results, see Online Data Supplement Figure E2.

The above analyses were repeated in sarcoidosis subjects to investigate potential trends in radiomic and lung function measures across Scadding stages (Figure 5 and Figures E3).

Results were considered significant at p < 0.05.

RESULTS

For brevity, the results presented here are for the left lung axial orientation, unless otherwise noted. In general, similar patterns were seen for the right lung, and coronal & sagittal anatomical planes (see Online Data Supplement for more information).

Differences in Non-Radiomic Characteristics between Sarcoidosis and Controls and across Scadding Stages in Sarcoidosis.

Table 2 shows the characteristics of the subject population in this study. Compared to controls (N=78), sarcoidosis subjects (N=73) were younger [54.1 (SD 8.6) vs. 64.5 (SD 8.3) years; p<0.001], with a higher BMI [30.1 (SD 7.0) vs. 28.0 (SD 4.8); p=0.029], larger population of white (84.9% vs. 94.1%, p=0.002) and a meaningfully, but not significantly, higher percentage of men (49.3% vs 35.9%; p=0.133). Adjusted for gender, race, ethnicity, age and height, sarcoidosis had significantly lower pre-bronchodilator FEV1 and FVC (p<0.001 for both) compared to controls, but a similar FEV1:FVC ratio (p=0.465).

Subject characteristics were similar across the Scadding stages (Table 2; p>0.05), with differences observed in BMI and the FEV1:FVC ratio (p=0.040 and p<0.001, respectively). Stage IV had the lowest FEV1:FVC ratio, while others had FEV1:FVC ratios within a normal range.

Differences in Global Radiomic Measures were Observed between Sarcoidosis and Controls

The average of the global skewness was positive, but lower in sarcoidosis subjects compared to controls [3.17 (SE 0.09) vs. 3.64 (SE 0.06); p<0.001], as was the global kurtosis [12.7 (SE 0.7) vs. 16.7 (SE 0.6); p<0.001], global fractal dimension [2.404 (SE 0.004) vs. 2.429 (SE 0.003); p<0.001], and global Geary’s C [0.212 (SE 0.002) vs. 0.226 (SE 0.002)]; the average global Moran’s I [0.702 (SE 0.005) vs. 0.691 (SE 0.004)] was higher in sarcoidosis subjects (p<0.001) (Table 3, and Tables E1 & E2).

Associations were Apparent between Global Radiomic Measures and Lung Function and Differed between Sarcoidosis and Controls

Kurtosis.

The relationship between kurtosis and pre-BD FEV1 (Figure 2A) differed between sarcoidosis and control subjects (p=0.032); in sarcoidosis subjects there was a positive association between kurtosis and FEV1 (β=0.053 L, SE=0.012; p<0.001); however, in controls, there was no significant association between kurtosis and FEV1 (p=0.189). The relationship between kurtosis and FVC and FEV1:FVC was positive (Figure 2B & 2C) and did not differ between sarcoidosis and control subjects (p>0.087).

Geary’s C.

The association between Geary’s C and pre-BD FEV1 (Figure 2D), pre-BD FVC (Figure 2E) and FEV1:FVC (Figure 2F) differed between sarcoidosis and controls (p<0.001, p=0.002, and p<0.001, respectively); in sarcoidosis subjects, there were significant positive associations between Geary’s C and FEV1, FVC, and FEV1:FVC (FEV1: β=15.7 L, SE=3.6, p<0.001; FVC: β=21.4 L, SE=4.6, p=0.008; FEV1:FVC: β=2.11 units, SE=0.42, p<0.001); however, in control subjects, there were significant negative associations between Geary’s C and FEV1 and FEV1:FVC (FEV1: β=−10.9 L, SE=4.5, p=0.016; FEV1:FVC: β=−1.06 units, SE=0.51, p=0.040).

Differences in Spatial Radiomic Measures were Noted Between Sarcoidosis and Controls

Radiomic measures differed between cases and controls throughout much of the lung across all three planes, as exhibited by separation of confidence bands in Figure 3, and t-statistics greater than 1.96 in Figure 4. Skewness (top panel in Figure 3, yellow in Figure 4) and kurtosis (second row in Figure 3, light green in Figure 4) were significantly lower in sarcoidosis compared to controls, with the largest significant differences observed in the superior and middle sagittal lung regions; smaller, but still significant differences were observed in the coronal plane. Fractal dimension (third row in Figure 3, teal in Figure 4) and Moran’s I (fourth row in Figure 3, dark blue in Figure 4) were significantly higher, and Geary’s C (bottom panel in Figure 3, dark purple in Figure 4) significantly lower in the sarcoidosis population, with the most significant differences in the superior and lateral lung regions; there were also large differences in fractal dimension in the coronal plane. See Online Data Supplement Figures E1 and E2 for right lung results.

Radiomic Measures of HRCT Differed According to Scadding Stage Classifications

Global radiomic measures differed when patients were classified by Scadding stage, across all anatomical orientations, with the significant differences driven by Stage IV (Table 3). Not surprisingly, radiomic measures did not follow a sequential increasing or decreasing pattern across Scadding stages. In addition, the various radiomic measures did not follow the same patterns across Scadding stages in terms of which stages have higher (or lower) mean values compared to others. The average of the global skewness, kurtosis and fractal dimension were highest in stage I subjects and lowest in stage IV (p=0.027, p=0.032, p<0.001, respectively). For the global Moran’s I, the average was lowest in stage 0 subjects [0.667 (SE 0.013)], and highest in stage IV subjects [0.729 (SE 0.008)] (p<0.001). Average Geary’s C was highest in stage III subjects [0.229 (SE 0.005)], and lowest in stage IV subjects [0.195 (0.002)] (p<0.001).

Importantly, in sarcoidosis subjects, global radiomic measures explained more variation in the lung function measures than Scadding stage measures. For pre-BD FEV1, Scadding stage along with gender, age and BMI explained 44.5% of the variability of pre-BD FEV1, compared to 67.7% with global radiomic measures and gender, age and BMI. Scadding stage along with gender, age and BMI explained 51.4% of the variability in pre-BD FVC, whereas, global radiomic measures along with gender, age and BMI explained 71.1% of the variation in pre-BD FVC. For the FEV1:FVC ratio, Scadding stage, gender, age and BMI explained 26.9%, compared to 40.2% of the variation explained with global radiomic measures, gender, age and BMI.

DISCUSSION

To our knowledge, this is the first radiomics analysis in sarcoidosis. We show the potential of radiomics as biomarkers for sarcoidosis, by (1) detecting differences in HRCT between sarcoidosis and healthy controls, (2) finding associations between radiomic measures and lung function, and (3) identifying trends between Scadding stages. This study can be classified as phase II evidence in the developmental evidence of radiomics as a biomarker for sarcoidosis treatment and disease course prediction²⁴.

In our findings, global radiomic measures differed significantly between cases and controls in both the left and right lungs in each anatomical plane. The distribution of HUs on CT from subjects with sarcoidosis are more normally distributed (i.e. less skew and kurtosis) as compared to controls. This is due to increased opacification (i.e. whiter areas) on CT scans of sarcoidosis subjects likely caused by parenchymal abnormalities; although it is not clear from our findings which abnormalities are contributing the most. These radiomic findings are consistent with findings via visual assessment that note increased opacification and parenchymal abnormalities on CTs from subjects with sarcoidosis². Moreover, the significant differences in our second-order radiomic features (fractal dimension, Moran’s I and Geary’s C), show adjacent pixels are more similar (in terms of HUs) on CT scans from subjects with sarcoidosis as compared to controls. This is likely a result of nodule conglomeration and/or fibrosis on CT scans of subjects with sarcoidosis, which is consistent with findings via visual assessment². Furthermore, the spatial radiomic measures displayed significant geographic variation across the lung between cases and controls, with differences most apparent in the superior, mid-to-outer sagittal regions. These spatial findings suggest an upper and lateral lobe predominance of radiographic abnormalities in sarcoidosis, which is consistent with findings on visual assessment³⁵. Thus, our radiomic findings are consistent with known visual abnormalities for pulmonary sarcoidosis. It is reassuring that these results support rather than conflict with visual scoring. A next step in our work will be to assess correlations between visual assessment and radiomic measures and to use the two together to provide an integrative radiologic assessment of pulmonary sarcoidosis.

Global radiomic measures were also significantly associated with spirometry. The association between radiomic measures and lung function was in general stronger among the sarcoidosis population. In sarcoidosis subjects, lower skewness and kurtosis, likely caused by more parenchymal abnormalities, were associated with lower spirometry. Moreover, lower fractal dimension, higher Moran’s I and lower Geary’s C, likely caused by nodule conglomeration and/or fibrosis, were associated with lower spirometry. It is particularly interesting that there are strong associations between radiomic measures and lung function considering that previous work measuring associations between lung function and visual scoring on CTs from subjects with sarcoidosis has not shown a consistent association². The associations between radiomic measures and spirometry provide further evidence of the potential clinical utility of radiomic analyses in assessment of pulmonary sarcoidosis.

Based on Scadding stage classifications, we found that global radiomics differed significantly in both the left and right lungs in each anatomical plane, with much of the significant differences driven by Scadding stage IV. For skewness, kurtosis, fractal dimension and Geary’s C, the most significant pairwise differences were between stage I and IV; for Moran’s I, between stage 0 and IV. Not surprisingly, radiomic measures did not follow a sequential increasing or decreasing pattern across Scadding stages. In addition, the various radiomic measures did not follow the same patterns across Scadding stages in terms of which stages have higher (or lower) mean values compared to others. These findings are consistent with our knowledge regarding the non-sequential ordering of Scadding stage² and the potential for CT to provide more sensitive and specific information in regards to parenchymal abnormalities in pulmonary sarcoidosis. Specifically, these results suggest that Scadding stage 0 exhibit some type of abnormality on HRCT. Additionally, these radiomic features appear to be sensitive to parenchymal abnormalities noted in Scadding stage IV, such as fibrosis and/or upper lobe volume loss with hilar retraction. Furthermore, we found that radiomic measures are a better predictor of spirometry than Scadding stage, which is consistent with other studies which note the poor predictive ability of Scadding stage in regards to lung function^36,37.

Further investigation of the utility of radiomic measures as a quantitative biomarker for pulmonary disease, including lung function and other clinical characteristics is warranted, especially given the ability to generate these measures in an automated fashion. A major benefit of a radiomic analysis is the automated computational efficiency and reproducibility, increasing the potential use of these methods in clinical settings. We analyzed 151 segmented scans in approximately three minutes per scan, which is arguably faster than visual assessment. Radiomic algorithms could be programmed into scanners to be available along with visual reads to further enhance a patient’s image profile for sarcoidosis staging and disease monitoring.

Our study is not without limitations. This investigation was performed on a modest sample size. Repeating this analysis in a larger sarcoidosis population will be important to verify generalizability and reproducibility. Based on the way the COPDGene cohort was constructed, our control population was only non-smokers. We do not believe that this introduced significant biases given the small proportion of smokers in the sarcoidosis population, and because a sensitivity analysis removing the sarcoidosis patients who smoked showed qualitatively similar results to those we describe above. Although the HRCT scans from both the controls and sarcoidosis were obtained under very similar protocols, we note that a weakness of this study is that the controls and sarcoidosis subjects were scanned on various different Siemens CT machines with different tube currents. This potentially introduces biases due to scanner and protocol differences.

To conclude, our work identifies the usefulness of radiomics on HRCT by efficiently and objectively identifying pulmonary parenchymal abnormalities on subjects with sarcoidosis. We also highlight the significant association between radiomics and lung function, particularly among those with sarcoidosis, suggesting that the radiomic measurements that we evaluated have functional implications. This work shows exciting promise for radiomics as biomarkers for disease in sarcoidosis, which should be further evaluated for application in the clinic and/or research setting. We are hopeful that future research with radiomics on HRCT will result in better understanding about disease progression, classification, and treatment options for subjects with sarcoidosis.

An earlier version of this manuscript was posted to ArXiv in June 2018³⁸. Compared to that version, the present manuscript has substantial improvements to its methods, results, and presentation.