Identification of reproducible radiomic features from on-board volumetric images: A multi-institutional phantom study

BACKGROUND

Radiomics analysis using on-board volumetric images has attracted research attention as a method for predicting prognosis during treatment; however, the lack of standardization is still one of the main concerns.

PURPOSE

This study investigated the factors that influence the reproducibility of radiomic features extracted from on-board volumetric images using an anthropomorphic radiomics phantom. Furthermore, a phantom experiment was conducted with different treatment machines from multiple institutions as external validation to identify reproducible radiomic features.

METHODS

The phantom was designed to be 35 × 20 × 20 cm with eight types of heterogeneous spheres (⌀ = 1, 2, and 3 cm). On-board volumetric images were acquired using 15 treatment machines from eight institutions. Of these, kilovoltage cone-beam computed tomography (kV-CBCT) image data acquired from four treatment machines at one institution were used as an internal evaluation dataset to explore the reproducibility of radiomic features. The remaining image data, including kV-CBCT, megavoltage-CBCT (MV-CBCT), and megavoltage computed tomography (MV-CT) provided by seven different institutions (11 treatment machines), were used as an external validation dataset. A total of 1,302 radiomic features, including 18 first-order, 75 texture, 465 (i.e., 93 × 5) Laplacian of Gaussian (LoG) filter-based, and 744 (i.e., 93 × 8) wavelet filter-based features, were extracted within the spheres. The intraclass correlation coefficient (ICC) was calculated to explore feature repeatability and reproducibility using an internal evaluation dataset. Subsequently, the coefficient of variation (COV) was calculated to validate the feature variability of external institutions. An absolute ICC exceeding 0.85 or COV under 5% was considered indicative of a highly reproducible feature.

RESULTS

For internal evaluation, ICC analysis showed that the median percentage of radiomic features with high repeatability was 95.2%. The ICC analysis indicated that the median percentages of highly reproducible features for inter-tube current, reconstruction algorithm, and treatment machine were decreased by 20.8%, 29.2%, and 33.3%, respectively. For external validation, the COV analysis showed that the median percentage of reproducible features was 31.5%. A total of 16 features, including nine LoG filter-based and seven wavelet filter-based features, were indicated as highly reproducible features. The gray-level run-length matrix (GLRLM) was classified as containing the most frequent features (N = 8), followed by the gray-level dependence matrix (N = 7) and gray-level co-occurrence matrix (N = 1) features.

CONCLUSIONS

We developed the standard phantom for radiomics analysis of kV-CBCT, MV-CBCT, and MV-CT images. With this phantom, we revealed that the differences in the treatment machine and image reconstruction algorithm reduce the reproducibility of radiomic features from on-board volumetric images. Specifically, the most reproducible features for external validation were LoG or wavelet filter-based GLRLM features. However, the acceptability of the identified features should be examined in advance at each institution before applying the findings to prognosis prediction.

Impact of a novel multilayer imager on metal artifacts in MV-CBCT

OBJECTIVE

Megavoltage cone-beam CT (MV-CBCT) imaging offers several advantages including reduced metal artifacts and accurate electron density mapping for adaptive or emergent situations. However, MV-CBCT imaging is limited by the poor efficiency of current detectors. Here we examine a new MV imager and compare CBCT reconstructions under clinically relevant scenarios. 
Approach: A multilayer imager (MLI), consisting of four vertically stacked standard flat-panel imagers, was mounted to a clinical linear accelerator. A custom anthropomorphic pelvis phantom with replaceable femoral heads was imaged using MV-CBCT and kV-CBCT. Bone, aluminum, and titanium were used as femoral head inserts. 8MU 2.5MV scans were acquired for all four layers and (as reference) the top layer. Prostate and bladder were contoured on a reference CT and transferred to the other scans after rigid registration, from which the structural similarity index measure (SSIM) was calculated. Prostate and bladder were also contoured on CBCT scans without guidance, and Dice coefficients were compared to CT contours.
Main results: kV-CBCT demonstrated the highest SSIMs with bone inserts (prostate:0.86, bladder:0.94) and lowest with titanium inserts (0.32, 0.37). 4-layer MV-CBCT SSIMs were preserved with bone (0.75, 0.80) as compared to titanium (0.67, 0.74), outperforming kV-CBCT when metal is present. 1-layer MV-CBCT consistently underperformed 4-layer results across all phantom configurations. Unilateral titanium inserts and bilateral aluminum insert results fell between the bone and bilateral titanium results. Dice coefficients trended similarly, with 4-layer MV-CBCT reducing metal artifact impact relative to KV-CBCT to provide better soft-tissue identification. 
Significance: MV-CBCT with a 4-layer MLI showed improvement over single-layer MV scans, approaching kV-CBCT quality for soft-tissue contrast. In the presence of artifact-producing metal implants, four-layer MV-CBCT scans outperformed kV-CBCT by eliminating artifacts and single-layer MV-CBCT by reducing noise. MV-CBCT with a novel multi-layer imager may be a valuable alternative to kV-CBCT, particularly in the presence of metal.

Characterization of the Megavoltage Cone-Beam Computed Tomography (MV-CBCT) System on HalcyonTM for IGRT: Image Quality Benchmark, Clinical Performance, and Organ Doses

Purpose: The Varian Halcyon includes an ultrafast 6 MV flattening filter free (FFF) cone-beam computed tomography (MV-CBCT). Although a kV-CBCT add-on is available, in the basic configuration MV is used for image guided radiotherapy (IGRT). We characterized the MV-CBCT imager in terms of reproducibility, linearity, field of view (FOV) dependence, detectability of soft-tissue, and the effect of metal implants. The performance of the MV-CBCT in the clinic, including resulting dose to organs, is also discussed herein.

Methods: A Gammex phantom was scanned using a Halcyon MV-CBCT and a 120 kVp Siemens Definition Edge CT. Mean and standard deviation of Hounsfield Units (HUs) for different electron density relative to water (ρeW) inserts were extracted. Doses to clinical patients due to MV-CBCT are calculated within Eclipse during treatment planning.

Results: A stable and near-linear HU-to-ρeW curve was obtained using the MV-CBCT. As the scan length increased from 10 to 28cm, the linearity of curve improved while the mean HUs decreased by 30%. All soft tissue inserts in the Gammex phantom were distinguishable. A crescent artifact affected HU measurements by up to 40 HUs. Soft-tissue contrast was sufficient for clinical online image-guidance in the low dose (5 MU) mode. Mean doses per fraction to organs-at-risk (OARs) were as high as 6 cGy for head and neck, 5 cGy for breast, and 4 cGy for pelvis patients. Metal rods did not affect HU values or introduce noticeable artifacts.

Conclusions: Halcyon's MV-CBCT has sufficient soft tissue contrast for IGRT and lacks metal-induced artifacts. Even though the absolute HU values vary with phantom size and scanning length, the HU-to-ρeW conversions are linear and stable day-to-day. In clinical cases, highest tissue doses from MV-CBCT ranged from 2-7cGy per fraction for various treatment sites, which could be significant for some organs at risk. Dose to out-of-treatment-field organs can be limited by reducing the scan length definition during planning and using the low dose mode. The high quality imaging mode did not provide material advantages over the low dose mode. Adequate IGRT was successfully delivered to multiple tumor sites using MV-CBCT.

Characterizing a novel scintillating glass for application to megavoltage cone-beam computed tomography

PURPOSE

The purpose of this study was to evaluate the performance of a prototype electric portal imaging device (EPID) with a high detective quantum efficiency (DQE) scintillator, LKH-5. Specifically, image quality in context of both planar and megavoltage (MV) cone-beam computed tomography (CBCT) is analyzed.

METHODS

Planar image quality in terms of modulation transfer function (MTF), noise power spectrum (NPS), and DQE are measured and compared to an existing EPID (AS-1200) using the 6 MV beamline for a Varian TrueBeam linac. Imager performance is contextualized for three-dimensional (3D), MV-CBCT performance by measuring imager lag and analyzing the expected degradation of the DQE as a function of dose. Finally, comparisons between reconstructed images of the Catphan phantom in terms of qualitative quality and signal-difference-to-noise ratio (SDNR) are made for 6 MV images using both conventional and LKH-5 EPIDs as well as for the kilovoltage (kV) on-board imager (OBI).

RESULTS

Analysis of the NPS reveals linearity at all measured doses using the prototype LKH-5 detector. While the first zero of the MTF is much lower for the LKH-5 detector than the conventional EPID (0.6 cycles/mm vs 1.6 cycles/mm), the normalized NPS (NNPS) multiplied by total quanta (qNNPS) of the LKH-5 detector is roughly a factor of seven to eight times lower, yielding a DQE(0) of approximately 8%. First, second, and third frame lag were measured at approximately 23%, 5%, and 1%, respectively, although no noticeable image artifacts were apparent in reconstructed volumes. Analysis of low-dose performance reveals that DQE(0) remains at 80% of its maximum value at a dose as low as 7.5 × 10^-6  MU. For a 400 projection technique, this represents a total scan dose of 0.0030 MU, suggesting that if imaging doses are increased to a value typical of kV-CBCT scans (~2.7 cGy), the LKH-5 detector will retain quantum noise limited performance. Finally, comparing Catphan scans, the prototype detector exhibits much lower image noise than the conventional EPID, resulting in improved small object representation. Furthermore, SDNR of H₂ O and polystyrene cylinders improved from -1.95 and 2.94 to -15 and 18.7, respectively.

CONCLUSIONS

Imaging performance of the prototype LKH-5 detector was measured and analyzed for both planar and 3D contexts. Improving noise transfer of the detector results in concurrent improvement of DQE(0). For 3D imaging, temporal characteristics were adequate for artifact-free performance and at relevant doses, the detector retained quantum noise limited performance. Although quantitative MTF measurements suggest poorer resolution, small object representation of the prototype imager is qualitatively improved over the conventional detector due to the measured reduction in noise.

Super-resolution imaging in a multiple layer EPID

A new portal imager consisting of four vertically stacked conventional electronic portal imaging device (EPID) layers has been constructed in pursuit of improved detective quantum efficiency (DQE). We hypothesize that super-resolution (SR) imaging can also be achieved in such a system by shifting each layer laterally by half a pixel relative to the layer above. Super-resolution imaging will improve resolution and contrast-to-noise ratio (CNR) in megavoltage (MV) planar and cone beam computed tomography (MV-CBCT) applications. Simulations are carried out to test this hypothesis with digital phantoms. To assess planar resolution, 2 mm long iron rods with 0.3 × 0.3 mm² square cross-section are arranged in a grid pattern at the center of a 1 cm thick solid water. For measuring CNR in MV-CBCT, a 20 cm diameter digital phantom with 8 inserts of different electron densities is used. For measuring resolution in MV-CBCT, a digital phantom featuring a bar pattern similar to the Gammex™ phantom is used. A 6 MV beam is attenuated through each phantom and detected by each of the four detector layers. Fill factor of the detector is explicitly considered. Projections are blurred with an estimated point spread function (PSF) before super-resolution reconstruction. When projections from multiple shifted layers are used in SR reconstruction, even a simple shift-add fusion can significantly improve the resolution in reconstructed images. In the reconstructed planar image, the grid pattern becomes visually clearer. In MV-CBCT, combining projections from multiple layers results in increased CNR and resolution. The inclusion of two, three and four layers increases CNR by 40%, 70% and 99%, respectively. Shifting adjacent layers by half a pixel almost doubles resolution. In comparison, using four perfectly aligned layers does not improve resolution relative to a single layer.

The risk of secondary cancer in nasopharyngeal carcinoma paediatric patients due to intensity modulated radiotherapy and mega-voltage cone beam computed tomography

INTRODUCTION

There is a growing interest in the study of radiation-induced secondary cancer. The aim of this work is (i) to estimate the peripheral doses attributable to intensity modulated radiotherapy (IMRT) and mega-voltage cone beam computed tomography (MV-CBCT) for some organs at risk (OARs) which surround the target being treated (Nasopharynx) in paediatric patients. (ii) To estimate the risk of radiation-induced secondary cancers attributable to patient setup verification imaging dose using MV-CBCT for Nasopharyngeal Carcinoma (NPC) in paediatric patients and comparing it with that attributable to the therapeutic dose using IMRT.

METHODS

Intensity modulated radiotherapy treatment planning of 10 NPC paediatric patients was carried out on KonRad release 2.2.23. The additional radiation doses to the patients attributable to MV-CBCT were calculated also using Xio Version 4.4. A paediatric phantom and thermoluminescent dosimeters (TLDs) were used to measure the patient doses attributable to IMRT. These doses were then compared with the calculated doses. The risk of induced secondary cancers attributable to IMRT and MV-CBCT was calculated and compared to each other.

RESULTS

The absorbed doses (mean dose) for the OARs (Brain, Brain stem, spinal cord, thyroid, oesophagus, mandible, heart, optic nerve, lung and eye) were higher for the therapeutic dose than for the imaging dose used in the verification of patient position before and during the treatment. The risk of induced secondary cancers in thyroid, oesophagus and lung (the only organs from the OARs which have tabulated values for risk calculations) was higher for therapeutic dose (7.29 ± 0.73%, 2.62 ± 0.17% and 6.76 ± 0.87%, respectively) than for verification imaging dose (0.14 ± 0.00%, 0.06 ± 0.00%, 0.10 ± 0.03% respectively).

CONCLUSION

The risk of secondary cancers attributable to verification imaging dose using MV-CBCT is very small compared to therapeutic dose using IMRT. Therefore, it is important to focus on the risk of secondary cancers attributable to therapeutic dose especially when using IMRT, where the produced leakage radiation is considerably high compared to some other techniques (such as conformal radiotherapy).

Evaluating positional accuracy using megavoltage cone-beam computed tomography for IMRT with head-and-neck cancer

Accurate dose delivery is essential for the success of intensity-modulated radiation therapy (IMRT) for patients with head-and-neck (HN) cancer. Reproducibility of IMRT dose delivery to HN regions can be critically influenced by treatment-related changes in body contours. Moreover, some set-up margins may not be adaptable to positional uncertainties of HN structures at every treatment. To obtain evidence for appropriate set-up margins in various head and neck areas, we prospectively evaluated positional deviation (δ values) of four bony landmarks (i.e. the clivus and occipital protuberance for the head region, and the mental protuberance and C5 for the neck region) using megavoltage cone-beam computed tomography during a treatment course. Over 800 δ values were analyzed in each translational direction. Positional uncertainties for HN cancer patients undergoing IMRT were evaluated relative to the body mass index. Low positional accuracy was observed for the neck region compared with the head region. For the head region, most of the δ was distributed within ± 5 mm, and use of the current set-up margin was appropriate. However, the δ values for the neck region were within ± 8 mm. Especially for overweight patients, a few millimeters needed to be added to give an adequate set-up margin. For accurate dose delivery to targets and to avoid excess exposure to normal tissues, we recommend that the positional verification process be performed before every treatment.

Megavoltage cone beam computed tomography: Commissioning and evaluation of patient dose

The improvement in conformal radiotherapy techniques enables us to achieve steep dose gradients around the target which allows the delivery of higher doses to a tumor volume while maintaining the sparing of surrounding normal tissue. One of the reasons for this improvement was the implementation of intensity-modulated radio therapy (IMRT) by using linear accelerators fitted with multi-leaf collimator (MLC), Tomo therapy and Rapid arc. In this situation, verification of patient set-up and evaluation of internal organ motion just prior to radiation delivery become important. To this end, several volumetric image-guided techniques have been developed for patient localization, such as Siemens OPTIVUE/MVCB and MVision megavoltage cone beam CT (MV-CBCT) system. Quality assurance for MV-CBCT is important to insure that the performance of the Electronic portal image device (EPID) and MV-CBCT is suitable for the required treatment accuracy. In this work, the commissioning and clinical implementation of the OPTIVUE/MVCB system was presented. The geometry and gain calibration procedures for the system were described. The image quality characteristics of the OPTIVUE/MVCB system were measured and assessed qualitatively and quantitatively, including the image noise and uniformity, low-contrast resolution, and spatial resolution. The image reconstruction and registration software were evaluated. Dose at isocenter from CBCT and the EPID were evaluated using ionization chamber and thermo-luminescent dosimeters; then compared with that calculated by the treatment planning system (TPS- XiO 4.4). The results showed that there are no offsets greater than 1 mm in the flat panel alignment in the lateral and longitudinal direction over 18 months of the study. The image quality tests showed that the image noise and uniformity were within the acceptable range, and that a 2 cm large object with 1% electron density contrast can be detected with the OPTIVUE/MVCB system with 5 monitor units (MU) protocol. The registration software was accurate within 2 mm in the anterior-posterior, left-right, and superior-inferior directions. The additional dose to the patient from MV-CBCT study set with 5 MU at the isocenter of the treatment plan was 5 cGy. For Electronic portal image device (EPID) verification using two orthogonal images with 2 MU per image the additional dose to the patient was 3.8 cGy. These measured dose values were matched with that calculated by the TPS-XiO, where the calculated doses were 5.2 cGy and 3.9 cGy for MVCT and EPID respectively.