Stéphane Bedwani

Assessing lung function using contrast‐enhanced dual‐energy computed tomography for potential applications in radiation therapy

Purpose: There is an increasing interest in the evaluation of lung function from physiological images in radiation therapy treatment planning to reduce the extent of postradiation toxicities. The purpose of this work was to retrieve reliable functional information from contrast‐enhanced dual‐energy computed tomography (DECT) for new applications in radiation therapy. The functional information obtained by DECT is also compared with other methods using single‐energy CT (SECT) and single‐photon emission computed tomography (SPECT) with CT. The differential function between left and right lung, as well as between lobes is computed for all methods. Methods: Five lung cancer patients were retrospectively selected for this study; each underwent a SPECT/CT scan and a contrast‐injected DECT scan, using 100 and 140 Sn kVp. The DECT images are postprocessed into iodine concentration maps, which are further used to determine the perfused blood volume. These maps are calculated in two steps: (a) a DECT stoichiometric calibration adapted to the presence of iodine and followed by (b) a two‐material decomposition technique. The functional information from SECT is assumed proportional to the HU numbers from a mixed CT image. The functional data from SPECT/CT are considered proportional to the number of counts. A radiation oncologist segmented the entire lung volume into five lobes on both mixed CT images and low‐dose CT images from SPECT/CT to allow a regional comparison. The differential function for each subvolume is computed relative to the entire lung volume. Results: The differential function per lobe derived from SPECT/CT correlates strongly with DECT (Pearsons coefficient r = 0.91) and moderately with SECT (r = 0.46). The differential function for the left lung shows a mean difference of 7% between SPECT/CT and DECT; and 17% between SPECT/CT and SECT. The presence of nonfunctional areas, such as localized emphysema or a lung tumor, is reflected by an intensity drop in the iodine concentration maps. Functional dose volume histograms (fDVH) are also generated for two patients as a proof of concept. Conclusion: The extraction of iodine concentration maps from a contrast‐enhanced DECT scan is achieved to compute the differential function for each lung subvolume and good agreement is found in respect to SPECT/CT. One promising avenue in radiation therapy is to include such functional information during treatment planning dose optimization to spare functional lung tissues.

Robust quantitative contrast‐enhanced dual‐energy CT for radiotherapy applications

Purpose The purpose of this study was to develop and validate accurate methods for determining iodine content and virtual noncontrast maps of physical parameters, such as electron density, in the context of radiotherapy. Methods A simulation environment is developed to compare three methods allowing extracting iodine content and virtual noncontrast composition: (a) two‐material decomposition, (b) three‐material decomposition with the conservation of volume constraint, and (c) eigentissue decomposition. The simulation allows comparing the performance of the methods using iodine‐based contrast agent contents in tissues from a reference dataset with variable density and elemental composition. The comparison is performed in two ways: (a) with a priori knowledge on the composition of the targeted tissue, and (b) without a priori knowledge on the base tissue. The three methods are tested with patient images scanned with dual‐energy CT and iodine‐based contrast agent. An experimental calibration adapted to the presence of iodine is performed by imaging tissue equivalent materials and diluted contrast agent solutions with known atomic composition. Results Results show that in the case of known a priori on the composition of the targeted tissue, the two‐material decomposition is robust to variable densities and atomic compositions without biasing the results. In the absence of a priori knowledge on the target tissue composition, the eigentissue decomposition method yields minimal bias and higher robustness to variations. Results from the experimental calibration and the images of two patients show that the extracted quantities are accurate and the bias is negligible for both methods with respect to clinical applications in their respective scope of use. For the patient imaged with a contrast agent, virtual noncontrast electron densities are found in good agreement with values extracted from the scan without contrast agent. Conclusion This study identifies two accurate methods to quantify iodine‐based contrast agents and virtual noncontrast composition images with dual‐energy CT. One is the two‐material decomposition with a priori knowledge of the constituent components focused on organ‐specific applications, such as kidney or lung function assessment. The other method is the eigentissue decomposition and is useful for general radiotherapy applications, such as treatment planning where accurate dose calculations are needed in the absence of contrast agent.

Improved tissue assignment using dual-energy computed tomography in low-dose rate prostate brachytherapy for Monte Carlo dose calculation

PURPOSE An improvement in tissue assignment for low-dose rate brachytherapy (LDRB) patients using more accurate Monte Carlo (MC) dose calculation was accomplished with a metallic artifact reduction (MAR) method specific to dual-energy computed tomography (DECT). METHODS The proposed MAR algorithm followed a four-step procedure. The first step involved applying a weighted blend of both DECT scans (I H/L) to generate a new image (I Mix). This action minimized Hounsfield unit (HU) variations surrounding the brachytherapy seeds. In the second step, the mean HU of the prostate in I Mix was calculated and shifted toward the mean HU of the two original DECT images (I H/L). The third step involved smoothing the newly shifted I Mix and the two original I H/L, followed by a subtraction of both, generating an image that represented the metallic artifact (I A,(H/L)) of reduced noise levels. The final step consisted of subtracting the original I H/L from the newly generated I A,(H/L) and obtaining a final image corrected for metallic artifacts. Following the completion of the algorithm, a DECT stoichiometric method was used to extract the relative electronic density (ρe) and effective atomic number (Z eff) at each voxel of the corrected scans. Tissue assignment could then be determined with these two newly acquired physical parameters. Each voxel was assigned the tissue bearing the closest resemblance in terms of ρe and Z eff, comparing with values from the ICRU 42 database. A MC study was then performed to compare the dosimetric impacts of alternative MAR algorithms. RESULTS An improvement in tissue assignment was observed with the DECT MAR algorithm, compared to the single-energy computed tomography (SECT) approach. In a phantom study, tissue misassignment was found to reach 0.05% of voxels using the DECT approach, compared with 0.40% using the SECT method. Comparison of the DECT and SECT D 90 dose parameter (volume receiving 90% of the dose) indicated that D 90 could be underestimated by up to 2.3% using the SECT method. CONCLUSIONS The DECT MAR approach is a simple alternative to reduce metallic artifacts found in LDRB patient scans. Images can be processed quickly and do not require the determination of x-ray spectra. Substantial information on density and atomic number can also be obtained. Furthermore, calcifications within the prostate are detected by the tissue assignment algorithm. This enables more accurate, patient-specific MC dose calculations.

SU‐E‐T‐501: A Sensitivity Study of Tissue Characterization for Brachytherapy Monte Carlo Dose Calculation

Purpose: To establish the reliability of electron density (ED) mapping and tissue segmentation techniques using CT images for brachytherapy, considering stochastic and systematic Hounsfield unit (HU) variations. Methods: Most common artifacts are simulated within a Monte Carlo theoretical framework. A set of CT data is first generated with the EGSnrc suite followed by their reconstruction performed with an iterative algorithm. A statistical analysis of HU values from reconstructed images is performed to determine uncertainties and systematic effects introduced by the iterative algorithm and the presence of artifacts. Maps of ED and tissue indexes are retrieved from an HU‐ED curve calibrated at 120 kVp using experimental measurements and ICRU data. The mean energy absorption coefficient for each tissue is computed from ICRU data for an Iridium‐192 source in order to evaluate the impact on dose calculations. A probabilistic approach is used to compute the uncertainty on absorbed dose generating random distributions of HU and determining the probabilistic effects on the extracted ED and absorption coefficients. Results: For an uncertainty of +/− 20 HU, absorption coefficient uncertainties raise up to 3% when HU values are near the fat‐muscle intersection and up to 9% near the muscle‐spongiosa intersection. Uncertainties on ED are found to be less than 1% for HU above 0 and up to 3% for fat. A systematic effect of 50 HU caused by typical artifacts yields absorption coefficient errors up to 30% for HU values near the muscle‐spongiosa region, and errors in ED up to 8% for fat. Conclusion: Results show that the main source of dose calculation uncertainty is caused by the sensitivity of the tissue segmentation technique. This study suggests that improvements in such techniques are yet to be achieved.

SU‐E‐J‐87: Lung Deformable Image Registration Using Surface Mesh Deformation for Dose Distribution Combination

PURPOSE To allow a reliable deformable image registration (DIR) method for dose calculation in radiation therapy. This work proposes a performance assessment of a morphological segmentation algorithm that generates a deformation field from lung surface displacements with 4DCT datasets. METHODS From the 4DCT scans of 15 selected patients, the deep exhale phase of the breathing cycle is identified as the reference scan. Varian TPS EclipseTM is used to draw lung contours, which are given as input to the morphological segmentation algorithm. Voxelized contours are smoothed by a Gaussian filter and then transformed into a surface mesh representation. Such mesh is adapted by rigid and elastic deformations to match each subsequent lung volumes. The segmentation efficiency is assessed by comparing the segmented lung contour and the TPS contour considering two volume metrics, defined as Volumetric Overlap Error (VOE) [%] and Relative Volume Difference (RVD) [%] and three surface metrics, defined as Average Symmetric Surface Distance (ASSD) [mm], Root Mean Square Symmetric Surface Distance (RMSSD) [mm] and Maximum Symmetric Surface Distance (MSSD) [mm]. Then, the surface deformation between two breathing phases is determined by the displacement of corresponding vertices in each deformed surface. The lung surface deformation is linearly propagated in the lung volume to generate 3D deformation fields for each breathing phase. RESULTS The metrics were averaged over the 15 patients and calculated with the same segmentation parameters. The volume metrics obtained are a VOE of 5.2% and a RVD of 2.6%. The surface metrics computed are an ASSD of 0.5 mm, a RMSSD of 0.8 mm and a MSSD of 6.9 mm. CONCLUSION This study shows that the morphological segmentation algorithm can provide an automatic method to capture an organ motion from 4DCT scans and translate it into a volume deformation grid needed by DIR method for dose distribution combination.

The impacts of mid‐treatment CBCT‐guided patient repositioning on target coverage during lung VMAT

The purpose of this study is quantify intrafraction motion (IFM) during lung volumetric‐modulated arc therapy (VMAT) and evaluate the impact of mid‐treatment cone beam computed tomography (CBCT)‐guided patient repositioning on target coverage.

151: Does MID-Treatment CBCT-Guided Patient Repositioning During Lung VMAT Impact Target Coverage?

Purpose: The objectives of this study are to (1) quantify intrafraction motion (IFM) during lung volumetric-modulated arc therapy (VMAT) and (2) evaluate the impact of mid-treatment patient repositioning after cone beam computed tomography (CBCT) acquisition upon target coverage. Method: This analysis included lung tumors treated with VMAT between April 2012 and June 2015 with 50-60 Gy in 3-5 fractions. Treatment planning consisted of a four-dimensional (4D) CT scan from which an internal target volume (ITV) delineation was performed. A 5 mm margin was added in all directions to obtain the final planning target volume (PTV). Treatment sessions were performed in supine position with a customized dual vacuum immobilization device (BodyFIX, Elekta, Stockholm, Sweden). All patients underwent pre and mid-treatment CBCTs to ensure proper repositioning. Following each CBCT, a two-step rigid registration was performed by an experienced radiation oncologist according to the planning CT, taking into account organs at risk (OARs). Bone shift was first assessed with a registration of the vertebrae adjacent to the lesion. Then, tumor shift was isolated with a soft tissue registration by aligning targets. IFM, combining bone and tumor shifts, was defined as the target displacement from pre to mid-treatment CBCT acquisition and was quantified in terms of anterior-posterior (AP), cranio-caudal (CC) and medio-lateral (ML) amplitudes as well as three-dimensional (3D) vector. For patients with IFM ≥ 5 mm, a post hoc dose calculation analysis was performed to assess target coverage impacts of mid-treatment CBCT-guided repositioning. Results: Ninety–seven patients, totalizing 367 fractions, were included. Mean (±SD) overall treatment time was 53:02 ± 13:08 min. Mean time from pre to midtreatment CBCT acquisition was 22:58 ± 5:33 min. Mean time to perform mid treatment CBCT scan acquisition, registrations and couch repositioning was 15:49 ± 4:14 min. Mean IFM amplitudes were 0.9 ± 1.2 mm, 0.6 ± 1.0 mm and 0.6 ± 0.8 mm in the AP, CC and ML respectively. IFM was < 3 mm and < 5 mm in all directions in respectively 315/367 (86%) and 358/367 (98%) fractions. Mean 3D IFM vector was 1.5 ± 1.4 mm (max = 8.1 mm) and was < 5 mm in 354/367 (96%). Among the 13 fractions with IFM vector ≥ 5 mm, 11/13 (85%) were dominantly induced by a tumor shift. For all these fractions, dose calculation analysis of worst-case scenario indicates that ITV coverage would have remained ≥ 95% without mid-treatment CBCT-guided patient repositioning. Conclusion: For 96% of fractions in patients immobilized with a customized BodyFIX dual vacuum bag, the IFM vector was within the 5 mm PTV margin used. Mid-treatment CBCT-guided couch repositioning did not significantly impact ITV coverage and prolonged treatment duration. Mid-treatment imaging may remain pertinent for selected patients with strict OAR dose constraints.

SU-F-J-91: Sparing Lung Function in Treatment Planning Using Dual Energy Tomography

PURPOSE To propose an alternate treatment plan that minimizes the dose to the functional lung tissues. In clinical situation, the evaluation of the lung functionality is typically derived from perfusion scintigraphy. However, such technique has spatial and temporal resolutions generally inferior to those of a CT scan. Alternatively, it is possible to evaluate pulmonary function by analysing the iodine concentration determined via contrast-enhanced dual energy CT (DECT) scan. METHODS Five lung cancer patients underwent a scintigraphy and a contrast-enhanced DECT scan (SOMATOM Definition Flash, Siemens). The iodine concentration was evaluated using the two-material decomposition method to produce a functional map of the lung. The validation of the approach is realized by comparison between the differential function computed by DECT and scintigraphy. The functional map is then used to redefine the V5 (volume of the organ that received more than 5 Gy during a radiotherapy treatment) to a novel functional parameter, the V5f. The V5f, that uses a volume weighted by its function level, can assist in evaluating optimal beam entry points for a specific treatment plan. RESULTS The results show that the differential functions obtained by scintigraphy and DECT are in good agreement with a mean difference of 6%. In specific cases, we are able to visually correlate low iodine concentration with abnormal pulmonary lung or cancerous tumors. The comparison between V5f and V5 has shown that some entry points can be better exploited and that new ones are now accessible, 2.34 times more in average, without increasing the V5f -- thus allowing easier optimization of other planning objectives. CONCLUSION In addition to the high-resolution DECT images, the iodine map provides local information used to detect potential functional heterogeneities in the 3D space. We propose that this information be used to calculate new functional dose parameters such as the V5f. The presenting author, Andreanne Lapointe, received a canadian scholarship from MITACS. Part of the funding is from the compagny Siemens.

SU-F-18C-04: A Combination of Monoenergetic Reconstruction and Stoichiometric Calibration for Tissue Characterization Using Dual Energy Computed Tomography

PURPOSE Dual energy computed tomography (DECT) pre-reconstruction methods require the prior knowledge of the X-ray source spectrum to allow extracting physical parameters needed for radiation therapy dose calculation, such as electron density (ED) and the effective atomic number (EAN). While DECT stoichiometric calibration may provide reliable performance for typical radiation therapy clinical conditions, it is yet to be adapted to prereconstruction methods. The presence of noise and inaccurate spectrum description may lead to systematic errors and artifacts which compromise the accuracy of treatment planning. METHODS A new technique is investigated which consists in applying a DECT stoichiometric calibration method to a set of monoenergetic images obtained with a DECT prereconstruction method. To evaluate the performance of this extended method, a simulation environment is developed to generate DECT scans under well controlled conditions, to reconstruct monoenergetic images of a tissue-equivalent phantom from transformed sinograms and to extract ED and EAN maps using a DECT formalism. RESULT Under simulated clinical conditions, the accuracy in determining ED with the extended method versus a pre-reconstruction method alone is shown to be better than 0.35% versus 0.5%, respectively. In the presence of a realistic noise level, EAN determination presents a relative mean error that drops from 2.5% to 0.5% once the calibration is applied. Considering a spectrum alteration by a 1 mm Cu layer, EAN errors are up to 30% for the pre-reconstruction method alone versus less than 3% for the extended method. CONCLUSION This study shows that combining pre-reconstruction DECT methods with a stoichiometric calibration considerably improves the accuracy and reliability of tissue characterization for radiation therapy in a clinical context. The presented method could potentially be adapted as gold standard for dose calculation methods based on DECT.