A Cunliffe

Lung Texture in Serial Thoracic Computed Tomography Scans: Correlation of Radiomics-based Features With Radiation Therapy Dose and Radiation Pneumonitis Development

PURPOSE To assess the relationship between radiation dose and change in a set of mathematical intensity- and texture-based features and to determine the ability of texture analysis to identify patients who develop radiation pneumonitis (RP). METHODS AND MATERIALS A total of 106 patients who received radiation therapy (RT) for esophageal cancer were retrospectively identified under institutional review board approval. For each patient, diagnostic computed tomography (CT) scans were acquired before (0-168 days) and after (5-120 days) RT, and a treatment planning CT scan with an associated dose map was obtained. 32- × 32-pixel regions of interest (ROIs) were randomly identified in the lungs of each pre-RT scan. ROIs were subsequently mapped to the post-RT scan and the planning scan dose map by using deformable image registration. The changes in 20 feature values (ΔFV) between pre- and post-RT scan ROIs were calculated. Regression modeling and analysis of variance were used to test the relationships between ΔFV, mean ROI dose, and development of grade ≥2 RP. Area under the receiver operating characteristic curve (AUC) was calculated to determine each features ability to distinguish between patients with and those without RP. A classifier was constructed to determine whether 2- or 3-feature combinations could improve RP distinction. RESULTS For all 20 features, a significant ΔFV was observed with increasing radiation dose. Twelve features changed significantly for patients with RP. Individual texture features could discriminate between patients with and those without RP with moderate performance (AUCs from 0.49 to 0.78). Using multiple features in a classifier, AUC increased significantly (0.59-0.84). CONCLUSIONS A relationship between dose and change in a set of image-based features was observed. For 12 features, ΔFV was significantly related to RP development. This study demonstrated the ability of radiomics to provide a quantitative, individualized measurement of patient lung tissue reaction to RT and assess RP development.

Lung texture in serial thoracic CT scans: registration-based methods to compare anatomically matched regions.

PURPOSE The aim of this study was to compare three demons registration-based methods to identify spatially matched regions in serial computed tomography (CT) scans for use in texture analysis. METHODS Two thoracic CT scans containing no lung abnormalities and acquired during serial examinations separated by at least one week were retrospectively collected from 27 patients. Over 1000 regions of interest (ROIs) were randomly placed in the lungs of each baseline scan. Anatomically matched ROIs in the corresponding follow-up scan were placed by mapping the baseline scan ROI center pixel to (1) the original follow-up scan, (2) the follow-up scan resampled to match the baseline scan voxel size, and (3) the follow-up scan aligned to the baseline scan through affine registration. Mappings used the vector field obtained through demons deformable registration of each follow-up scan variant to the baseline scan. 140 texture features distributed among five feature classes were calculated in all ROIs. Feature value differences between paired ROIs were evaluated using Bland-Altman 95% limits of agreement. For each feature, (1) the mean feature value change and (2) the difference between the upper and lower limits of agreement were normalized to the mean feature value to obtain, respectively, the normalized bias and normalized range of agreement (nRoA). Nonparametric tests were used to evaluate differences in normalized bias and nRoA across the three methods. RESULTS Because patient CT scans contained no pathology, minimal changes in feature values were expected (i.e., low nRoA and normalized bias). Seventy-five features with very large feature value variability (nRoA ≥ 100%) were excluded from further analysis. Across the remaining 65 features, significant differences in normalized bias were observed among the three methods. The lowest normalized bias (median: 0.06%) was achieved when feature values were calculated on original follow-up scans. The affine registration method achieved the lowest nRoA, though nRoA was not significantly increased using original follow-up scans. Features with low nRoA values also had low normalized bias, though the converse was not necessarily true. Using nRoA as a metric, a set of 20 features having both low nRoA and normalized bias were identified. CONCLUSIONS Three methods to facilitate texture analysis of serial CT scans using demons registration for ROI placement were evaluated. The bias in feature value change between matched ROIs was minimized when feature values were calculated on original baseline and follow-up scans. A set of features that had both low bias and variability (nRoA) in feature value change using this method were identified. This texture analysis approach could facilitate future measurement of pathologic changes between CT scans without necessitating calculation of feature values on deformed scans.PURPOSE The aim of this study was to compare three demons registration-based methods to identify spatially matched regions in serial computed tomography (CT) scans for use in texture analysis. METHODS Two thoracic CT scans containing no lung abnormalities and acquired during serial examinations separated by at least one week were retrospectively collected from 27 patients. Over 1000 regions of interest (ROIs) were randomly placed in the lungs of each baseline scan. Anatomically matched ROIs in the corresponding follow-up scan were placed by mapping the baseline scan ROI center pixel to (1) the original follow-up scan, (2) the follow-up scan resampled to match the baseline scan voxel size, and (3) the follow-up scan aligned to the baseline scan through affine registration. Mappings used the vector field obtained through demons deformable registration of each follow-up scan variant to the baseline scan. 140 texture features distributed among five feature classes were calculated in all ROIs. Feature value differences between paired ROIs were evaluated using Bland-Altman 95% limits of agreement. For each feature, (1) the mean feature value change and (2) the difference between the upper and lower limits of agreement were normalized to the mean feature value to obtain, respectively, the normalized bias and normalized range of agreement (nRoA). Nonparametric tests were used to evaluate differences in normalized bias and nRoA across the three methods. RESULTS Because patient CT scans contained no pathology, minimal changes in feature values were expected (i.e., low nRoA and normalized bias). Seventy-five features with very large feature value variability (nRoA ≥ 100%) were excluded from further analysis. Across the remaining 65 features, significant differences in normalized bias were observed among the three methods. The lowest normalized bias (median: 0.06%) was achieved when feature values were calculated on original follow-up scans. The affine registration method achieved the lowest nRoA, though nRoA was not significantly increased using original follow-up scans. Features with low nRoA values also had low normalized bias, though the converse was not necessarily true. Using nRoA as a metric, a set of 20 features having both low nRoA and normalized bias were identified. CONCLUSIONS Three methods to facilitate texture analysis of serial CT scans using demons registration for ROI placement were evaluated. The bias in feature value change between matched ROIs was minimized when feature values were calculated on original baseline and follow-up scans. A set of features that had both low bias and variability (nRoA) in feature value change using this method were identified. This texture analysis approach could facilitate future measurement of pathologic changes between CT scans without necessitating calculation of feature values on deformed scans.

Incorporation of pre‐therapy 18F‐FDG uptake data with CT texture features into a radiomics model for radiation pneumonitis diagnosis

Purpose To determine whether the addition of standardized uptake value (SUV) from PET scans to CT lung texture features could improve a radiomics‐based model of radiation pneumonitis (RP) diagnosis in patients undergoing radiotherapy. Methods and materials Anonymized data from 96 esophageal cancer patients (18 RP‐positive cases of Grade ≥ 2) were collected including pre‐therapy PET/CT scans, pre‐/post‐therapy diagnostic CT scans and RP status. Twenty texture features (first‐order, fractal, Laws’ filter and gray‐level co‐occurrence matrix) were calculated from diagnostic CT scans and compared in anatomically matched regions of the lung. Classifier performance (texture, SUV, or combination) was assessed by calculating the area under the receiver operating characteristic curve (AUC). For each texture feature, logistic regression classifiers consisting of the average change in texture feature value and the pre‐therapy SUV standard deviation (SUVSD) were created and compared with the texture feature as a lone classifier using ANOVA with correction for multiple comparisons (P < 0.0025). Results While clinical parameters (mean lung dose, smoking history, tumor location) were not significantly different among patients with and without symptomatic RP, SUV and texture parameters were significantly associated with RP status. AUC for single‐texture feature classifiers alone ranged from 0.58 to 0.81 and 0.53 to 0.71 in high‐dose (≥ 30 Gy) and low‐dose (< 10 Gy) regions of the lungs, respectively. AUC for SUVSD alone was 0.69 (95% confidence interval: 0.54–0.83). Adding SUVSD into a logistic regression model significantly improved model fit for 18, 14 and 11 texture features and increased the mean AUC across features by 0.08, 0.06, and 0.04 in the low‐, medium‐, and high‐dose regions, respectively. Conclusions Addition of SUVSD to a single‐texture feature improves classifier performance on average, but the improvement is smaller in magnitude when SUVSD is added to an already effective classifier using texture alone. These findings demonstrate the potential for more accurate assessment of RP using information from multiple imaging modalities.

TU‐A‐12A‐04: Quantitative Texture Features Calculated in Lung Tissue From CT Scans Demonstrate Consistency Between Two Databases From Different Institutions

PURPOSE To evaluate the consistency of computed tomography (CT) scan texture features, previously identified as stable in a healthy patient cohort, in esophageal cancer patient CT scans. METHODS 116 patients receiving radiation therapy (median dose: 50.4Gy) for esophageal cancer were retrospectively identified. For each patient, diagnostic-quality pre-therapy (0-183 days) and post-therapy (5-120 days) scans (mean voxel size: 0.8mm×0.8mm×2.5mm) and a treatment planning scan and associated dose map were collected. An average of 501 32×32-pixel ROIs were placed randomly in the lungs of each pre-therapy scan. ROI centers were mapped to corresponding locations in post-therapy and planning scans using the displacement vector field output by demons deformable registration. Only ROIs with mean dose <5Gy were analyzed, as these were expected to contain minimal post-treatment damage. 140 texture features were calculated in pre-therapy and post-therapy scan ROIs and compared using Bland-Altman analysis. For each feature, the mean feature value change and the distance spanned by the 95% limits of agreement were normalized to the mean feature value, yielding normalized range of agreement (nRoA) and normalized bias (nBias). Using Wilcoxon signed rank tests, nRoA and nBias were compared with values computed previously in 27 healthy patient scans (mean voxel size: 0.67mm×0.67mm×1mm) acquired at a different institution. RESULTS nRoA was significantly (p<0.001) larger in cancer patients than healthy patients. Differences in nBias were not significant (p=0.23). The 20 features identified previously as having nRoA<20% for healthy patients had the lowest nRoA values in the current database, with an average increase of 5.6%. CONCLUSION Despite differences in CT scanner type, scan resolution, and patient health status, the same 20 features remained stable (i.e., low variability and bias) in the absence of disease changes for databases from two institutions. Identification of these features is the first step towards quantifying radiation-induced changes between preand post-therapy scans. Supported, in part, by NIH Grant Nos. S10 RR021039, and P30 CA14599, the Virginia and D. K. Ludwig Fund for Cancer Research, Imaging Research Institute, Biological Sciences Division, The University of Chicago, and The Institute for Translational Medicine Pilot Award, The University of Chicago.

Comparison of demons deformable registration-based methods for texture analysis of serial thoracic CT scans

To determine how 19 image texture features may be altered by three image registration methods, “normal” baseline and follow-up computed tomography (CT) scans from 27 patients were analyzed. Nineteen texture feature values were calculated in over 1,000 32x32-pixel regions of interest (ROIs) randomly placed in each baseline scan. All three methods used demons registration to map baseline scan ROIs to anatomically matched locations in the corresponding transformed follow-up scan. For the first method, the follow-up scan transformation was subsampled to achieve a voxel size identical to that of the baseline scan. For the second method, the follow-up scan was transformed through affine registration to achieve global alignment with the baseline scan. For the third method, the follow-up scan was directly deformed to the baseline scan using demons deformable registration. Feature values in matched ROIs were compared using Bland- Altman 95% limits of agreement. For each feature, the range spanned by the 95% limits was normalized to the mean feature value to obtain the normalized range of agreement, nRoA. Wilcoxon signed-rank tests were used to compare nRoA values across features for the three methods. Significance for individual tests was adjusted using the Bonferroni method. nRoA was significantly smaller for affine-registered scans than for the resampled scans (p=0.003), indicating lower feature value variability between baseline and follow-up scan ROIs using this method. For both of these methods, however, nRoA was significantly higher than when feature values were calculated directly on demons-deformed followup scans (p<0.001). Across features and methods, nRoA values remained below 26%.

Dynamic contrast-enhanced CT for the assessment of tumour response in malignant pleural mesothelioma: a pilot study

ObjectivesThe aim of this pilot study was to investigate the utility of haemodynamic parameters derived from dynamic contrast-enhanced computed tomography (DCE-CT) scans in the assessment of tumour response to treatment in malignant pleural mesothelioma (MPM) patients.MethodsThe patient cohort included nine patients undergoing chemotherapy and five patients on observation. Each patient underwent two DCE-CT scans separated by approximately 2 months. The DCE-CT parameters of tissue blood flow (BF) and tissue blood volume (BV) were obtained within the dynamically imaged tumour. Mean relative changes in tumour DCE-CT parameters between scans were compared between the on-treatment and on-observation cohorts. DCE-CT parameter changes were correlated with relative change in tumour bulk evaluated according to the modified RECIST protocol.ResultsDiffering trends in relative change in BF and BV between scans were found between the two patient groups (p = 0.19 and p = 0.06 for BF and BV, respectively). No significant rank correlations were found when comparing relative changes in DCE-CT parameters with relative change in tumour bulk.ConclusionsDiffering trends in the relative change of BF and BV between patients on treatment and on observation indicate the potential of DCE-CT for the assessment of pharmacodynamic endpoints with respect to treatment in MPM. A future study with a larger patient cohort and unified treatment regimens should be undertaken to confirm the results of this pilot study.Key Points• CT-derived haemodynamic parameters show differing trends between malignant pleural mesothelioma patients on treatment and patients off treatment• Changes in haemodynamic parameters do not correlate with changes in tumour bulk as measured according to the modified RECIST protocol• Differing trends across the two patient groups indicate the potential sensitivity of DCE-CT to assess pharmacodynamic endpoints in the treatment of MPM

SU-E-J-251: Incorporation of Pre-Therapy 18F-FDG Uptake with CT Texture Features in a Predictive Model for Radiation Pneumonitis Development

Purpose: To determine whether the addition of standardized uptake value (SUV) statistical variables to CT lung texture features can improve a predictive model of radiation pneumonitis (RP) development in patients undergoing radiation therapy. Methods: Anonymized data from 96 esophageal cancer patients (18 RP-positive cases of Grade ≥ 2) were retrospectively collected including pre-therapy PET/CT scans, pre-/posttherapy diagnostic CT scans and RP status. Twenty texture features (firstorder, fractal, Laws’ filter and gray-level co-occurrence matrix) were calculated from diagnostic CT scans and compared in anatomically matched regions of the lung. The mean, maximum, standard deviation, and 50th–95th percentiles of the SUV values for all lung voxels in the corresponding PET scans were acquired. For each texture feature, a logistic regression-based classifier consisting of (1) the average change in that texture feature value between the pre- and post-therapy CT scans and (2) the pre-therapy SUV standard deviation (SUVSD) was created. The RP-classification performance of each logistic regression model was compared to the performance of its texture feature alone by computing areas under the receiver operating characteristic curves (AUCs). T-tests were performed to determine whether the mean AUC across texture features changed significantly when SUVSD was added to the classifier. Results: The AUC for single-texturefeature classifiers ranged from 0.58–0.81 in high-dose (≥ 30 Gy) regions of the lungs and from 0.53–0.71 in low-dose (< 10 Gy) regions. Adding SUVSD in a logistic regression model using a 50/50 data partition for training and testing significantly increased the mean AUC by 0.08, 0.06 and 0.04 in the low-, medium- and high-dose regions, respectively. Conclusion: Addition of SUVSD from a pre-therapy PET scan to a single CT-based texture feature improves RP-classification performance on average. These findings demonstrate the potential for more accurate prediction of RP using information from multiple imaging modalities. Supported, in part, by the National Institute of Biomedical Imaging and Bioengineering of the National Institutes of Health under grant number T32 EB002103; SGA receives royalties and licensing fees through the University of Chicago for computer-aided diagnosis technology. HA receives royalties through the University of Chicago for computer-aided diagnosis technology.

TU‐F‐BRF‐03: Effect of Radiation Therapy Planning Scan Registration On the Dose in Lung Cancer Patient CT Scans

PURPOSE To characterize the effect of deformable registration of serial computed tomography (CT) scans on the radiation dose calculated from a treatment planning scan. METHODS Eighteen patients who received curative doses (≥60Gy, 2Gy/fraction) of photon radiation therapy for lung cancer treatment were retrospectively identified. For each patient, a diagnostic-quality pre-therapy (4-75 days) CT scan and a treatment planning scan with an associated dose map calculated in Pinnacle were collected. To establish baseline correspondence between scan pairs, a researcher manually identified anatomically corresponding landmark point pairs between the two scans. Pre-therapy scans were co-registered with planning scans (and associated dose maps) using the Plastimatch demons and Fraunhofer MEVIS deformable registration algorithms. Landmark points in each pretherapy scan were automatically mapped to the planning scan using the displacement vector field output from both registration algorithms. The absolute difference in planned dose (|ΔD|) between manually and automatically mapped landmark points was calculated. Using regression modeling, |ΔD| was modeled as a function of the distance between manually and automatically matched points (registration error, E), the dose standard deviation (SD_dose) in the eight-pixel neighborhood, and the registration algorithm used. RESULTS 52-92 landmark point pairs (median: 82) were identified in each patients scans. Average |ΔD| across patients was 3.66Gy (range: 1.2-7.2Gy). |ΔD| was significantly reduced by 0.53Gy using Plastimatch demons compared with Fraunhofer MEVIS. |ΔD| increased significantly as a function of E (0.39Gy/mm) and SD_dose (2.23Gy/Gy). CONCLUSION An average error of <4Gy in radiation dose was introduced when points were mapped between CT scan pairs using deformable registration. Dose differences following registration were significantly increased when the Fraunhofer MEVIS registration algorithm was used, spatial registration errors were larger, and dose gradient was higher (i.e., higher SD_dose). To our knowledge, this is the first study to directly compute dose errors following deformable registration of lung CT scans.

WE‐C‐103‐09: Investigation of Demons Deformable Registration‐Based Methods to Measure Lung CT Texture Change Over Time

PURPOSE To compare three demons registration-based methods to identify spatially matched regions in serial computed tomography (CT) scans for use in texture analysis. METHODS Two normal thoracic CT scans were collected from 27 patients. Over 1,000 regions of interest (ROIs) were randomly placed in the lungs of each baseline scan. Anatomically matched ROIs in the corresponding follow-up scan were placed by mapping the baseline scan ROI center pixel to (1) the original follow-up scan, (2) the follow-up scan resampled to match the baseline scan voxel size, and (3) the follow-up scan aligned to the baseline scan through affine registration. Mappings used the vector field obtained through demons deformable registration of each follow-up scan variant to the baseline scan. 140 texture features were calculated in all ROIs. Feature value differences between paired ROIs were evaluated using Bland-Altman analysis. For each feature, (1) the mean feature value change and (2) the distance spanned by the 95% limits of agreement were normalized to the mean feature value to obtain, respectively, normalized bias (nBias) and normalized range of agreement (nRoA). Paired Students t-tests were used to compare nBias across the three methods and nRoA across the three methods. RESULTS For 20 features with low variability (nRoA<20%), significant differences in nBias existed among the three methods. For every feature, the lowest nBias was achieved when feature values were calculated on original follow-up scans. nRoA was not significantly increased, indicating low variability in feature value change. CONCLUSION Three methods to facilitate texture analysis of serial CT scans were evaluated. The bias in feature value change between matched ROIs was minimized with original follow-up scans, using demons deformation to identify corresponding ROIs between scans. This approach could facilitate future measurement of pathologic change between CT scans without necessitating calculation of feature values on deformed scans. This work was supported, in part, by The Coleman Endowment through The University of Chicago Comprehensive Cancer Center, National Science Foundation Research Experience for Undergraduates (NSF REU) Award No. 1062909, and National Institutes of Health (NIH) Grant Nos. S10 RR021039, P30 CA14599, and T32 EB002103-23.

TU‐A‐218‐05: Evaluation of Image Registration Using Landmark Matching and Texture Analysis

Purpose: To compare demons deformable registration with rigid and affineregistration methods by measuring registration accuracy and image texture changes introduced by each algorithm. Methods: Eleven temporally sequential pairs of clinical thoracic CT scans demonstrating no abnormal pathology were collected retrospectively from different patients. Rigid, affine, and demons registration methods were applied to all scan pairs using an appropriate choice of registration parameters. 150 landmarks were automatically selected in each baseline scan, and a semi‐automated approach was used to identify corresponding landmarks in the respective deformed follow‐up scans. Eleven first‐order texture features were calculated in 32x32‐pixel regions of interest (ROIs) randomly placed in the baseline scans and at anatomically matched locations in the demons‐, rigid‐, and affine‐ registered follow‐up scans. For each ROI pair, feature value differences between the three registered follow‐up scans and the baseline scan were calculated. Results: The average per‐patient Euclidean distance between baseline scan landmarks and registered follow‐up scan landmarks ranged from 0.01–0.50 mm (overall mean: 0.18 mm), 1.6–6.8 mm (overall mean: 3.95 mm), and 2.06–7.91 mm (overall mean: 4.82 mm) for demons, affine, and rigid registration methods, respectively. Across all eleven features, the variances in feature value change for rigid and affine registration methods were at least an order of magnitude larger than the corresponding feature‐ value‐change variances for demons registration. Conclusions: Demons registration achieved more accurate spatial alignment between temporally sequential CT scans than rigid or affine registration methods. Furthermore, the variance in texture feature value changes between baseline and registered follow‐up scan ROIs was smaller for demons than for rigid or affine registration methods. These advantages of demons registration observed with “normal” thoracic CT scans indicate that future studies may incorporate demons to combine serial CT scans such that automated methods may distinguish registration artifacts from actual pathologic change.