Quality assessment of medical decision making in radiation oncology: variability in target volume delineation for brain tumours

Emergence of MR-Linac in Radiation Oncology: Successes and Challenges of Riding on the MRgRT Bandwagon

The special issue of JCM on “Advances of MRI in Radiation Oncology” provides a unique forum for scientific literature related to MR imaging in radiation oncology. This issue covered many aspects, such as MR technology, motion management, economics, soft-tissue–air interface issues, and disease sites such as the pancreas, spine, sarcoma, prostate, head and neck, and rectum from both camps—the Unity and MRIdian systems. This paper provides additional information on the success and challenges of the two systems. A challenging aspect of this technology is low throughput and the monumental task of education and training that hinders its use for the majority of therapy centers. Additionally, the cost of this technology is too high for most institutions, and hence widespread use is still limited. This article highlights some of the difficulties and how to resolve them.

Impact of Interobserver Variability in Manual Segmentation of Non-Small Cell Lung Cancer (NSCLC) Applying Low-Rank Radiomic Representation on Computed Tomography

Simple Summary

Discovery of predictive and prognostic radiomic features in cancer is currently of great interest to the radiologic and oncologic community. Tumor phenotypic and prognostic information can be obtained by extracting features on tumor segmentations, and it is typically imaging analysts, physician trainees, and attending physicians who provide these labeled datasets for analysis. The potential impact of level and type of specialty training on interobserver variability in manual segmentation of NSCLC was examined. Although there was some variability in segmentation between readers, the subsequently extracted radiomic features were overall well correlated. High fidelity radiomic feature extraction relies on accurate feature extraction from imaging that produce robust prognostic and predictive radiomic NSCLC biomarkers. This study concludes that this goal can be obtained using segmenters of different levels of training and clinical experience.

Abstract

This study tackles interobserver variability with respect to specialty training in manual segmentation of non-small cell lung cancer (NSCLC). Four readers included for segmentation are: a data scientist (BY), a medical student (LS), a radiology trainee (MH), and a specialty-trained radiologist (SK) for a total of 293 patients from two publicly available databases. Sørensen–Dice (SD) coefficients and low rank Pearson correlation coefficients (CC) of 429 radiomics were calculated to assess interobserver variability. Cox proportional hazard (CPH) models and Kaplan-Meier (KM) curves of overall survival (OS) prediction for each dataset were also generated. SD and CC for segmentations demonstrated high similarities, yielding, SD: 0.79 and CC: 0.92 (BY-SK), SD: 0.81 and CC: 0.83 (LS-SK), and SD: 0.84 and CC: 0.91 (MH-SK) in average for both databases, respectively. OS through the maximal CPH model for the two datasets yielded c-statistics of 0.7 (95% CI) and 0.69 (95% CI), while adding radiomic and clinical variables (sex, stage/morphological status, and histology) together. KM curves also showed significant discrimination between high- and low-risk patients (p-value < 0.005). This supports that readers’ level of training and clinical experience may not significantly influence the ability to extract accurate radiomic features for NSCLC on CT. This potentially allows flexibility in the training required to produce robust prognostic imaging biomarkers for potential clinical translation.

Intra- and inter-physician variability in target volume delineation in radiation therapy

Reduction in setup errors is advocated through daily imaging and adaptive therapy, where the target volume is drawn daily. Previous studies suggest that inter-physician volume variation is significant (1.5 cm standard deviation [SD]); however, there are limited data for intra-physician consistency in daily target volume delineation, which is investigated in this study. Seven patients with lung cancer were chosen based on the perceived difficulty of contouring their disease, varying from simple parenchymal lung nodules to lesions with extensive adjacent atelectasis. Four physicians delineated the gross tumor volume (GTV) for each patient on 10 separate days to see the intra- and inter-physician contouring. Isocenter coordinates (x, y and z), target volume (cm3), and largest dimensions on anterior-posterior (AP) and lateral views were recorded for each GTV. Our results show that the variability among the physicians was reflected by target volumes ranging from +109% to -86% from the mean while isocenter coordinate changes were minimal; 3.8, 1.7 and 1.9 mm for x, y and z coordinates, respectively. The orthogonal image (AP and lateral) change varied 16.3 mm and 15.0 mm respectively among days and physicians. We conclude than when performing daily imaging, random variability in contouring resulted in isocenter changes up to ±3.8 mm in our study. The shape of the target varied within ±16 mm. This study suggests that when using daily imaging to track isocenter, target volume, or treatment parameters, physicians should be aware of personal variability when considering margins added to the target volume in daily decision making especially for difficult cases.

New target volume delineation and PTV strategies to further personalise radiotherapy

Target volume delineation uncertainty (DU) is arguably one of the largest geometric uncertainties in radiotherapy that are accounted for using planning target volume (PTV) margins. Geometrical uncertainties are typically derived from a limited sample of patients. Consequently, the resultant margins are not tailored to individual patients. Furthermore, standard PTVs cannot account for arbitrary anisotropic extensions of the target volume originating from DU. We address these limitations by developing a method to measure DU for each patient by a single clinician. This information is then used to produce PTVs that account for each patient's unique DU, including any required anisotropic component. We do so using a two-step uncertainty evaluation strategy that does not rely on multiple samples of data to capture the DU of a patient's gross tumour volume (GTV) or clinical target volume. For simplicity, we will just refer to the GTV in the following. First, the clinician delineates two contour sets; one which bounds all voxels believed to have a probability of belonging to the GTV of 1, while the second includes all voxels with a probability greater than 0. Next, one specifies a probability density function for the true GTV boundary position within the boundaries of the two contours. Finally, a patient-specific PTV, designed to account for all systematic errors, is created using this information along with measurements of the other systematic errors. Clinical examples indicate that our margin strategy can produce significantly smaller PTVs than the van Herk margin recipe. Our new radiotherapy target delineation concept allows DUs to be quantified by the clinician for each patient, leading to PTV margins that are tailored to each unique patient, thus paving the way to a greater personalisation of radiotherapy.

Clinical Target Volume Auto-Segmentation of Esophageal Cancer for Radiotherapy After Radical Surgery Based on Deep Learning

Radiotherapy plays an important role in controlling the local recurrence of esophageal cancer after radical surgery. Segmentation of the clinical target volume is a key step in radiotherapy treatment planning, but it is time-consuming and operator-dependent. This paper introduces a deep dilated convolutional U-network to achieve fast and accurate clinical target volume auto-segmentation of esophageal cancer after radical surgery. The deep dilated convolutional U-network, which integrates the advantages of dilated convolution and the U-network, is an end-to-end architecture that enables rapid training and testing. A dilated convolution module for extracting multiscale context features containing the original information on fine texture and boundaries is integrated into the U-network architecture to avoid information loss due to down-sampling and improve the segmentation accuracy. In addition, batch normalization is added to the deep dilated convolutional U-network for fast and stable convergence. In the present study, the training and validation loss tended to be stable after 40 training epochs. This deep dilated convolutional U-network model was able to segment the clinical target volume with an overall mean Dice similarity coefficient of 86.7% and a respective 95% Hausdorff distance of 37.4 mm, indicating reasonable volume overlap of the auto-segmented and manual contours. The mean Cohen kappa coefficient was 0.863, indicating that the deep dilated convolutional U-network was robust. Comparisons with the U-network and attention U-network showed that the overall performance of the deep dilated convolutional U-network was best for the Dice similarity coefficient, 95% Hausdorff distance, and Cohen kappa coefficient. The test time for segmentation of the clinical target volume was approximately 25 seconds per patient. This deep dilated convolutional U-network could be applied in the clinical setting to save time in delineation and improve the consistency of contouring.

The effects of volume of interest delineation on MRI-based radiomics analysis: evaluation with two disease groups

BACKGROUND

Manual delineation of volume of interest (VOI) is widely used in current radiomics analysis, suffering from high variability. The tolerance of delineation differences and possible influence on each step of radiomics analysis are not clear, requiring quantitative assessment. The purpose of our study was to investigate the effects of delineation of VOIs on radiomics analysis for the preoperative prediction of metastasis in nasopharyngeal carcinoma (NPC) and sentinel lymph node (SLN) metastasis in breast cancer.

METHODS

This study retrospectively enrolled two datasets (NPC group: 238 cases; SLN group: 146 cases). Three operations, namely, erosion, smoothing, and dilation, were implemented on the VOIs accurately delineated by radiologists to generate diverse VOI variations. Then, we extracted 2068 radiomics features and evaluated the effects of VOI differences on feature values by the intra-class correlation coefficient (ICC). Feature selection was conducted by Maximum Relevance Minimum Redundancy combined with 0.632+ bootstrap algorithms. The prediction performance of radiomics models with random forest classifier were tested on an independent validation cohort by the area under the receive operating characteristic curve (AUC).

RESULTS

The larger the VOIs changed, the fewer features with high ICCs. Under any variation, SLN group showed fewer features with ICC ≥ 0.9 compared with NPC group. Not more than 15% top-predictive features identical to the accurate VOIs were observed across feature selection. The differences of AUCs of models derived from VOIs across smoothing or dilation with 3 pixels were not statistically significant compared with the accurate VOIs (p > 0.05) except for T2-weighted fat suppression images (smoothing: 0.845 vs. 0.725, p = 0.001; dilation: 0.800 vs. 0.725, p = 0.042). Dilation with 5 and 7 pixels contributed to remarkable AUCs in SLN group but the opposite in NPC group. The radiomics models did not perform well when tested by data from other delineations.

CONCLUSIONS

Differences in delineation of VOIs affected radiomics analysis, related to specific disease and MRI sequences. Differences from smooth delineation or expansion with 3 pixels width around the tumors or lesions were acceptable. The delineation for radiomics analysis should follow a predefined and unified standard.

The Impact of Transitioning to Prospective Contouring and Planning Rounds as Peer Review

Purpose

Our peer-review program previously consisted of weekly chart rounds performed before the end of the first week of treatment. In order to perform peer review before the start of treatment when possible, we implemented daily prospective contouring and planning rounds (CPR).

Methods and materials

At the time of computed tomography simulation, patients were categorized by the treating physician into 5 treatment groups based on urgency and complexity (ie, standard, urgent, palliative nonemergent, emergent, and special procedures). A scoring system was developed to record the outcome of case presentations, and the results of the CPR case presentations were compared with the time period 2.5 years before CPR implementation, for which peer review was performed retrospectively.

Results

CPR was implemented on October 1, 2015, and a total of 4759 patients presented for care through May 31, 2018. The majority were in the standard care path (n = 3154; 66.3%). Among the remainder of the charts, 358 (7.5%), 430 (9.0%), and 179 (3.8%) cases were in the urgent, nonemergent palliative, and emergent care paths, respectively. The remaining patients were in the special procedures group, representing brachytherapy and stereotactic radiosurgery. A total of 125 patients (2.6%) required major changes and were re-presented after the suggested modifications, 102 patients (2.1%) had minor recommendations that did not require a repeat presentation, and 247 cases (5.2%) had minor documentation-related recommendations that did not require editing of the contours. In the 2.5 years before the implementation, records of a total of 1623 patients were reviewed, and only 9 patients (0.6%) had minor recommendation for change. The remainder was noted as complete agreement.

Conclusions

Contouring and planning rounds were successfully implemented at our clinic. Pretreatment and, most often, preplanning review of contours and directives allows for a more detailed review and changes to be made early on in the treatment planning process. When compared with historical case presentations, the CPR method made our peer review more thorough and improved standardization.

Stereotactic Shifts During Frame-Based Image-Guided Stereotactic Radiosurgery: Clinical Measurements

PURPOSE

To determine the magnitude and reason for discrepancies between frame- and cone beam computed tomography (CBCT)-determined stereotactic coordinates, we reviewed frame-based Gamma Knife radiosurgery procedures in which CBCT was performed before treatment.

METHODS AND MATERIALS

Clinical and treatment documentation was reviewed for 150 frame placements for which stereotactic coordinates were defined via both frame and fiducials on computed tomography imaging and CBCT. Treatment planning system-reported rotational and translational differences and standard deviations (SDs) between frame-based and CBCT-based stereotactic coordinates were recorded. Potential clinical predictors for increased differences were collected. Multiple linear regressions were performed to evaluate for associations with increased translations and rotations.

RESULTS

The absolute mean of the measured pitch, yaw, and roll shifts was 0.14 degrees (range -0.71-0.63 degrees, SD 0.19 degrees), 0.16 degrees (range -0.50 to 0.83 degrees, SD 0.21 degrees), and 0.12 degrees (range 0.37-0.51 degrees, SD 0.15 degrees), respectively. The absolute mean of the measured shifts in the left-right, anteroposterior, and superior-inferior direction was 0.29 mm (range -1.29 to 0.82 mm, SD 0.35 mm), 0.24 mm (range -0.59 to 0.33 mm, SD 0.19 mm), and 0.24 mm (range -0.69 to 0.91 mm, SD 0.27 mm), respectively. Three cases (2.0%) exceeded 1 mm in translational difference, all in the left-right direction (1.05, 1.13, and 1.29 mm). Lower Karnofsky Performance Scale status was associated with greater translational differences (vector magnitude, P = .023) and rotation (pitch, P = .044; yaw, P = .002). Usage of longer total pin length (sum of all 4 fixation pin lengths) was associated with increased rotation but not with translation (P < .001 and P = .56, respectively).

CONCLUSIONS

CBCT imaging in this cohort of frame-based cases suggests that the discrepancy in stereotactic coordinates is less than 1 mm or degree in most cases. Low Karnofsky Performance Scale status and longer total pin length correlate with larger differences between frame-defined and CBCT-defined stereotactic coordinates.

Multi-institutional study of the variability in target delineation for six targets commonly treated with radiosurgery

BACKGROUND

Although accurate delineation of the target is a key factor of success in radiosurgery there are no consensus guidelines for target contouring.

AIM

The aim of the present study was therefore to quantify the variability in target delineation and discuss the potential clinical implications, for six targets regarded as common in stereotactic radiosurgery.

MATERIAL AND METHODS

Twelve Gamma Knife centers participated in the study by contouring the targets and organs at risks and performing the treatment plans. Analysis of target delineation variability was based on metrics defined based on agreement volumes derived from overlapping structures following a previously developed method. The 50% agreement volume (AV₅₀), the common and the encompassing volumes as well as the Agreement Volume Index (AVI) were determined.

RESULTS

Results showed that the lowest AVI (0.16) was found for one of the analyzed metastases (range of delineated volumes 1.27-3.33 cm³). AVI for the other two metastases was 0.62 and 0.37, respectively. Corresponding AVIs for the cavernous sinus meningioma, pituitary adenoma and vestibular schwannoma were 0.22, 0.37 and 0.50.

CONCLUSIONS

This study showed that the variability in the contouring was much higher than expected and therefore further work in standardizing the contouring practice in radiosurgery is warranted.

How Thick Should CT/MR Slices be to Plan Conformal Radiotherapy? A Study on the Accuracy of Three-Dimensional Volume Reconstruction

Aim and background

The aim of this experimental study was to correlate the thickness of acquired CT slices (2, 4 and 8 mm) or MR slices (4 and 7 mm) with the accuracy of three-dimensional volume reconstruction as performed by a commercially available radiation therapy planning system.

Methods

We used a cylindrical phantom, with a 15-cm diameter and 20-cm height, containing 5 spheres (12.7-31.8 mm diameter) of solid Plexiglas sunk in a 3% agar jelly solution. The phantom was scanned by the CT scan with 3 different slice thicknesses (2, 4 and 8 mm and a distance of 0 mm between the slices). Two different acquisition techniques (slice thickness of 4 and 7 mm with 0.8 and 1.4 mm slice distance, respectively) were compared in the MR study. The volume values calculated from measurements were compared with the known true volume values of the spheres.

Results

The average percentage volume difference between calculated and true values for the smaller spheres reconstructed with CT images 2 and 4 mm thick was generally less than 8%, whereas the error for volumes reconstructed with 8-mm-thick CT slices was more than 20%. For the larger spheres, the error was generally less than 5%. The data produced by MR acquisition agreed with those obtained using CT sections.

Conclusions

For targets less than 1.5 cm in diameter on our system it is reasonable to acquire CT images with the smallest thickness available. For targets between 1.5 and 3 cm, it seems sufficient to acquire the localization images with a slice thickness of 4 mm. For targets more than 4 cm in diameter, considering that with our radiation therapy planning system the time spent for manual contouring and for isodose calculation highly increased with the number of acquired images, we suggest that the acquisition of CT-MR slices 8-10-mm thick is totally adequate even for Conformal radiotherapy treatments.

Deep Deconvolutional Neural Network for Target Segmentation of Nasopharyngeal Cancer in Planning Computed Tomography Images

BACKGROUND

Radiotherapy is one of the main treatment methods for nasopharyngeal carcinoma (NPC). It requires exact delineation of the nasopharynx gross tumor volume (GTVnx), the metastatic lymph node gross tumor volume (GTVnd), the clinical target volume (CTV), and organs at risk in the planning computed tomography images. However, this task is time-consuming and operator dependent. In the present study, we developed an end-to-end deep deconvolutional neural network (DDNN) for segmentation of these targets.

METHODS

The proposed DDNN is an end-to-end architecture enabling fast training and testing. It consists of two important components: an encoder network and a decoder network. The encoder network was used to extract the visual features of a medical image and the decoder network was used to recover the original resolution by deploying deconvolution. A total of 230 patients diagnosed with NPC stage I or stage II were included in this study. Data from 184 patients were chosen randomly as a training set to adjust the parameters of DDNN, and the remaining 46 patients were the test set to assess the performance of the model. The Dice similarity coefficient (DSC) was used to quantify the segmentation results of the GTVnx, GTVnd, and CTV. In addition, the performance of DDNN was compared with the VGG-16 model.

RESULTS

The proposed DDNN method outperformed the VGG-16 in all the segmentation. The mean DSC values of DDNN were 80.9% for GTVnx, 62.3% for the GTVnd, and 82.6% for CTV, whereas VGG-16 obtained 72.3, 33.7, and 73.7% for the DSC values, respectively.

CONCLUSION

DDNN can be used to segment the GTVnx and CTV accurately. The accuracy for the GTVnd segmentation was relatively low due to the considerable differences in its shape, volume, and location among patients. The accuracy is expected to increase with more training data and combination of MR images. In conclusion, DDNN has the potential to improve the consistency of contouring and streamline radiotherapy workflows, but careful human review and a considerable amount of editing will be required.

Expertise Affects Inter-Observer Agreement at Peripheral Locations within a Brain Tumor

Magnetic resonance imaging (MRI) is a crucial tool for clinical brain tumor detection and delineation. Since the process of gross tumor volume delineation resides with clinicians, a better understanding of how they perform this task is required if improvements in life expectancy are to be made. Novice-expert comparison studies have been used to examine the effect of expertise on abnormality detection, but little research has investigated expertise-related differences in brain tumor delineation. In this study, undergraduate students (novices) and radiologists (experts) inspected a combination of T1 and T2 single and whole brain MRI scans, each containing a tumor. Using a tablet and stylus to provide an interactive environment, participants had an unlimited amount of time to scroll freely through the MRI slices and were instructed to delineate (i.e., draw a boundary) around any tumorous tissue. There was no reliable evidence for a difference in the gross tumor volume or total number of slices delineated between experts and novices. Agreement was low across both expertise groups and significantly lower at peripheral locations within a tumor than central locations. There was an interaction between expertise level and location within a tumor with experts displaying higher agreement at the peripheral slices than novices. An effect of brain image set on the order in which participants inspected the slices was also observed. The implications of these results for the training undertaken by early career radiologists and current practices in hospitals are discussed.

Reducing radiation dose and enhancing imaging quality of 4DCT for radiation therapy using iterative reconstruction algorithms

Purpose

Four-dimensional computed tomography (CT) images are typically used to quantify the necessary internal target volumes for thoracic and abdominal tumors. However, 4-dimensional CT is typically associated with excessive imaging dose to patients and the situation is exacerbated when using repeat 4-dimensional CT imaging on a weekly or daily basis throughout fractionated therapy. The aim of this work is to evaluate an iterative reconstruction (IR) algorithm that helps reduce the imaging dose to the patient while maintaining imaging quality as quantified by point spread function and contrast-to-noise ratios (CNRs).

Methods and materials

An IR algorithm, SAFIRE, was applied to CT data of a phantom and patients with varying CT doses and reconstruction kernels. Phantom data enable measurements of spatial resolution, contrast, and noise. The impact of SAFIRE on 4-dimensional CT was assessed with patient data acquired at 2 different dose levels during image guided radiation therapy with an in-room CT.

Results

Phantom data demonstrate that IR reduces noise approximately in proportion to the number of iterations indicated by the strength (SAFIRE 1 to SAFIRE 5). Spatial resolution and contrast are conserved independent of dose and reconstruction parameters. The CNR increases with an increase of imaging dose or an increase in the number of iterations. The use of IR on CT sets confirms the results that were derived from phantom scans. The IR significantly enhances single breathing phase CTs in 4-dimensional CT sets as assessed by CT number discrimination. Furthermore, the IR of the low dose 4-dimensional CT features a 45% increase in the CNR in comparison with the standard dose 4-dimensional CT.

Conclusions

The use of IR algorithms reduces noise while preserving spatial resolution and contrast, as evaluated from both phantom and patient CT data sets. For 4-dimensional CT, the IR can significantly improve image quality and reduce imaging dose without compromising image quality.

Quantitative assessment of target delineation variability for thymic cancers: Agreement evaluation of a prospective segmentation challenge

OBJECTIVES

We sought to quantitatively determine the inter-observer variability of expert radiotherapy target-volume delineation for thymic cancers, as part of a larger effort to develop an expert-consensus contouring atlas.

METHODS

A pilot dataset was created consisting of a standardized case presentation with pre- and post-operative DICOM CT image sets from a single patient with Masaoka-Koga Stage III thymoma. Expert thoracic radiation oncologists delineated tumor targets on the pre- and post-operative scans as they would for a definitive and adjuvant case, respectively. Respondents completed a survey including recommended dose prescription and target volume margins for definitive and post-operative scenarios. Inter-observer variability was analyzed quantitatively with Warfield's simultaneous truth, performance level estimation (STAPLE) algorithm and Dice similarity coefficient (DSC).

RESULTS

Seven users completed contouring for definitive and adjuvant cases; of these, 5 completed online surveys. Segmentation performance was assessed, with high mean±SD STAPLE-estimated segmentation sensitivity for definitive case GTV and CTV at 0.77 and 0.80, respectively, and post-operative CTV sensitivity of 0.55; all volumes had specificity of ≥0.99. Inter-observer agreement was markedly higher for the definitive target volumes, with mean±SD DSC of 0.88±0.03 and 0.89±0.04 for GTV and CTV respectively, compared to post-op CTV DSC of 0.69±0.06 (Kruskal-Wallis p<0.01.

CONCLUSION

Expert agreement for definitive case volumes was exceptionally high, though significantly lower agreement was noted post-operatively. Technique and dose prescription between experts was substantively consistent, and these preliminary results will be utilized to create an expert-consensus contouring atlas to aid the non-expert radiation oncologist in the planning of these challenging, rare tumors.

SINGLE INSTITUTION VARIABILITY IN INTENSITY MODULATED RADIATION TARGET DELINEATION FOR CANINE NASAL NEOPLASIA

Contouring variability is a significant barrier to the accurate delivery and reporting of radiation therapy. The aim of this descriptive study was to determine the variation in contouring radiation targets and organs at risk by participants within our institution. Further, we also aimed to determine if all individuals contoured the same normal tissues. Two canine nasal tumor datasets were selected and contoured by two ACVR-certified radiation oncologists and two radiation oncology residents from the same institution. Eight structures were consistently contoured including the right and left eye, the right and left lens, brain, the gross tumor volume (GTV), clinical target volume (CTV), and planning target volume (PTV). Spinal cord, hard and soft palate, and bulla were contoured on 50% of datasets. Variation in contouring occurred in both targets and normal tissues at risk and was particularly significant for the GTV, CTV, and PTV. The mean metric score and dice similarity coefficient were below the threshold criteria in 37.5-50% and 12.5-50% of structures, respectively, quantitatively indicating contouring variation. This study refutes our hypothesis that minimal variation in target and normal tissue delineation occurs. The variation in contouring may contribute to different tumor response and toxicity for any given patient. Our results also highlight the difficulty associated with replication of published radiation protocols or treatments, as even with replete contouring description the outcome of treatment is still fundamentally influenced by the individual contouring the patient.

A longitudinal magnetic resonance elastography study of murine brain tumors following radiation therapy

An accurate and noninvasive method for assessing treatment response following radiotherapy is needed for both treatment monitoring and planning. Measurement of solid tumor volume alone is not sufficient for reliable early detection of therapeutic response, since changes in physiological and/or biomechanical properties can precede tumor volume change following therapy. In this study, we use magnetic resonance elastography to evaluate the treatment effect after radiotherapy in a murine brain tumor model. Shear modulus was calculated and compared between the delineated tumor region of interest (ROI) and its contralateral, mirrored counterpart. We also compared the shear modulus from both the irradiated and non-irradiated tumor and mirror ROIs longitudinally, sampling four time points spanning 9-19 d post tumor implant. Results showed that the tumor ROI had a lower shear modulus than that of the mirror ROI, independent of radiation. The shear modulus of the tumor ROI decreased over time for both the treated and untreated groups. By contrast, the shear modulus of the mirror ROI appeared to be relatively constant for the treated group, while an increasing trend was observed for the untreated group. The results provide insights into the tumor properties after radiation treatment and demonstrate the potential of using the mechanical properties of the tumor as a biomarker. In future studies, more closely spaced time points will be employed for detailed analysis of the radiation effect.

Impact of microscopic disease extension, extra-CTV tumour islets, incidental dose and dose conformity on tumour control probability

The impact of microscopic disease extension (MDE), extra-CTV tumour islets (TIs), incidental dose and dose conformity on tumour control probability (TCP) is analyzed using insilico simulations in this study. MDE in the region in between GTV and CTV is simulated inclusive of geometric uncertainties (GE) using spherical targets and spherical dose distribution. To study the effect of incidental dose on TIs and the effect of dose-response curve (DRC) on tumour control, islets were randomly distributed and TCP was calculated for various dose levels by rescaling the dose. Further, the impact of dose conformity on required PTV margins is also studied. The required PTV margins are ~2 mm lesser than assuming a uniform clonogen density if an exponential clonogen density fall off in the GTV-CTV is assumed. However, margins are almost equal if GE is higher in both cases. This shows that GE has a profound impact on margins. The effect of TIs showed a bi-phasic relation with increasing dose, indicating that patients with islets not in the beam paths do not benefit from dose escalation. Increasing dose conformity is also found to have considerable effect on TCP loss especially for larger GE. Further, smaller margins in IGRT should be used with caution where uncertainty in CTV definition is of concern.

Uncertainties in target volume delineation in radiotherapy - are they relevant and what can we do about them?

Background

Modern radiotherapy techniques enable delivery of high doses to the target volume without escalating dose to organs at risk, offering the possibility of better local control while preserving good quality of life. Uncertainties in target volume delineation have been demonstrated for most tumour sites, and various studies indicate that inconsistencies in target volume delineation may be larger than errors in all other steps of the treatment planning and delivery process. The aim of this paper is to summarize the degree of delineation uncertainties for different tumour sites reported in the literature and review the effect of strategies to minimize them.

Conclusions

Our review confirmed that interobserver variability in target volume contouring represents the largest uncertainty in the process for most tumour sites, potentially resulting in a systematic error in dose delivery, which could influence local control in individual patients. For most tumour sites the optimal combination of imaging modalities for target delineation still needs to be determined. Strict use of delineation guidelines and protocols is advisable both in every day clinical practice and in clinical studies to diminish interobserver variability. Continuing medical education of radiation oncologists cannot be overemphasized, intensive formal training on interpretation of sectional imaging should be included in the program for radiation oncology residents.

Postoperative radiotherapy for glioma: improved delineation of the clinical target volume using the geodesic distance calculation

OBJECTS

To introduce a new method for generating the clinical target volume (CTV) from gross tumor volume (GTV) using the geodesic distance calculation for glioma.

METHODS

One glioblastoma patient was enrolled. The GTV and natural barriers were contoured on each slice of the computer tomography (CT) simulation images. Then, a graphic processing unit based on a parallel Euclidean distance transform was used to generate the CTV considering natural barriers. Three-dimensional (3D) visualization technique was applied to show the delineation results. Speed of operation and precision were compared between this new delineation method and the traditional method.

RESULTS

In considering spatial barriers, the shortest distance from the point sheltered from these barriers equals the sum of the distance along the shortest path between the two points; this consists of several segments and evades the spatial barriers, rather than being the direct Euclidean distance between two points. The CTV was generated irregularly rather than as a spherical shape. The time required to generate the CTV was greatly reduced. Moreover, this new method improved inter- and intra-observer variability in defining the CTV.

CONCLUSIONS

Compared with the traditional CTV delineation, this new method using geodesic distance calculation not only greatly shortens the time to modify the CTV, but also has better reproducibility.

An automatic method for accurate volume delineation of heterogeneous tumors in PET

PURPOSE

Accurate volumetric tumor delineation is of increasing importance in radiation treatment planning. Many tumors exhibit only moderate tracer uptake heterogeneity and delineation methods using an adaptive threshold lead to robust results. These methods use a tumor reference value R (e.g., ROI maximum) and the tumor background Bg to compute the volume reproducing threshold. This threshold corresponds to an isocontour which defines the tumor boundary. However, the boundaries of strongly heterogeneous tumors can not be described by an isocontour anymore and therefore conventional threshold methods are not suitable for accurate delineation. The aim of this work is the development and validation of a delineation method for heterogeneous tumors.

METHODS

The new method (voxel-specific threshold method, VTM) can be considered as an extension of an adaptive threshold method (lesion-specific threshold method, LTM), where instead of a lesion-specific threshold for the whole ROI, a voxel-specific threshold is computed by determining for each voxel Bg and R in the close vicinity of the voxel. The absolute threshold for the considered voxel is then given by Tabs=T×(R-Bg)+Bg, where T=0.39 was determined with phantom measurements.

VALIDATION

30 clinical datasets from patients with non-small-cell lung cancer were used to generate 30 realistic anthropomorphic software phantoms of tumors with different heterogeneities and well-known volumes and boundaries. Volume delineation was performed with VTM and LTM and compared with the known lesion volumes and boundaries.

RESULTS

In contrast to LTM, VTM was able to reproduce the true tumor boundaries accurately, independent of the heterogeneity. The deviation of the determined volume from the true volume was (0.8±4.2)% for VTM and (11.0±16.4)% for LTM.

CONCLUSIONS

In anthropomorphic software phantoms, the new method leads to promising results and to a clear improvement of volume delineation in comparison to conventional background-corrected thresholding. In the next step, the suitability for clinical routine will be further investigated.

Quality indicators in radiation oncology

Oncologic specialty societies and multidisciplinary collaborative groups have dedicated considerable effort to developing evidence-based quality indicators (QIs) to facilitate quality improvement, accreditation, benchmarking, reimbursement, maintenance of certification, and regulatory reporting. In particular, the field of radiation oncology has a long history of organized quality assessment efforts and continues to work toward developing consensus quality standards in the face of continually evolving technologies and standards of care. This report provides a comprehensive review of the current state of quality assessment in radiation oncology. Specifically, this report highlights implications of the healthcare quality movement for radiation oncology and reviews existing efforts to define and measure quality in the field, with focus on dimensions of quality specific to radiation oncology within the "big picture" of oncologic quality assessment efforts.

Extracting fuzzy classification rules from texture segmented HRCT lung images

Automatic tools for detection and identification of lung and lesion from high-resolution CT (HRCT) are becoming increasingly important both for diagnosis and for delivering high-precision radiation therapy. However, development of robust and interpretable classifiers still presents a challenge especially in case of non-small cell lung carcinoma (NSCLC) patients. In this paper, we have attempted to devise such a classifier by extracting fuzzy rules from texture segmented regions from HRCT images of NSCLC patients. A fuzzy inference system (FIS) has been constructed starting from a feature extraction procedure applied on overlapping regions from the same organs and deriving simple if-then rules so that more linguistically interpretable decisions can be implemented. The proposed method has been tested on 138 regions extracted from CT scan images acquired from patients with lung cancer. Assuming two classes of tissues C1 (healthy tissues) and C2 (lesion) as negative and positive, respectively; preliminary results report an AUC = 0.98 for lesions and AUC = 0.93 for healthy tissue, with an optimal operating condition related to sensitivity = 0.96, and specificity = 0.98 for lesions and sensitivity 0.99, and specificity = 0.94 for healthy tissue. Finally, the following results have been obtained: false-negative rate (FNR) = 6 % (C1), FNR = 2 % (C2), false-positive rate (FPR) = 4 % (C1), FPR = 3 % (C2), true-positive rate (TPR) = 94 %, (C1) and TPR = 98 % (C2).

Assessment of intra-observer variability in measurement of high-grade brain tumors

A 25% increase in bidimensional products (BPs) of tumor diameter has been used as a criterion for brain tumor progression. We studied intra-observer variability in measurements of BPs. Ten patients with contrast-enhancing glioblastoma multiforme underwent baseline and follow-up MR imaging. Seven observers measured BPs in various planes. Differences in BPs between scans were expressed as a percentage of baseline. This calculation was performed for both readings of the baseline and follow-up scans. Differences between change from baseline to follow-up on each reading (termed D values) were calculated for each reader (total of 196 D values). Median D value in each plane was calculated for each reader. Range of D values was 12.36-33.64% in axial plane (average 10.63%), 12.18-38.62% in coronal plane (average 26.84%) and 15.12-35.48% in sagittal plane (average 26.11%). Across all planes, 88 (45%) D values were >25%. When all imaging planes for any single observation were combined, in 76% of cases, at least one D value of >25% was seen. Based on the high degree of intra-observer variability, tumor measurements producing an increase in BP of >25% can routinely be obtained solely by chance.

Human-computer interaction in radiotherapy target volume delineation: a prospective, multi-institutional comparison of user input devices

The purpose of this study was the prospective comparison of objective and subjective effects of target volume region of interest (ROI) delineation using mouse-keyboard and pen-tablet user input devices (UIDs). The study was designed as a prospective test/retest sequence, with Wilcoxon signed rank test for matched-pair comparison. Twenty-one physician-observers contoured target volume ROIs on four standardized cases (representative of brain, prostate, lung, and head and neck malignancies) twice: once using QWERTY keyboard/scroll-wheel mouse UID and once with pen-tablet UID (DTX2100, Wacom Technology Corporation, Vancouver, WA, USA). Active task time, ROI manipulation task data, and subjective survey data were collected. One hundred twenty-nine target volume ROI sets were collected, with 62 paired pen-tablet/mouse-keyboard sessions. Active contouring time was reduced using the pen-tablet UID, with mean ± SD active contouring time of 26 ± 23 min, compared with 32 ± 25 with the mouse (p ≤ 0.01). Subjective estimation of time spent was also reduced from 31 ± 26 with mouse to 27 ± 22 min with the pen (p = 0.02). Task analysis showed ROI correction task reduction (p = 0.045) and decreased panning and scrolling tasks (p < 0.01) with the pen-tablet; drawing, window/level changes, and zoom commands were unchanged (p = n.s.) Volumetric analysis demonstrated no detectable differences in ROI volume nor intra- or inter-observer volumetric coverage. Fifty-two of 62 (84%) users preferred the tablet for each contouring task; 5 of 62 (8%) denoted no preference, and 5 of 62 (8%) chose the mouse interface. The pen-tablet UID reduced active contouring time and reduced correction of ROIs, without substantially altering ROI volume/coverage.

Critical discussion of evaluation parameters for inter-observer variability in target definition for radiation therapy

BACKGROUND AND PURPOSE

Inter-observer studies represent a valid method for the evaluation of target definition uncertainties and contouring guidelines. However, data from the literature do not yet give clear guidelines for reporting contouring variability. Thus, the purpose of this work was to compare and discuss various methods to determine variability on the basis of clinical cases and a literature review.

PATIENTS AND METHODS

In this study, 7 prostate and 8 lung cases were contoured on CT images by 8 experienced observers. Analysis of variability included descriptive statistics, calculation of overlap measures, and statistical measures of agreement. Cross tables with ratios and correlations were established for overlap parameters.

RESULTS

It was shown that the minimal set of parameters to be reported should include at least one of three volume overlap measures (i.e., generalized conformity index, Jaccard coefficient, or conformation number). High correlation between these parameters and scatter of the results was observed.

CONCLUSION

A combination of descriptive statistics, overlap measure, and statistical measure of agreement or reliability analysis is required to fully report the interrater variability in delineation.

Automatic volume delineation in oncological PET. Evaluation of a dedicated software tool and comparison with manual delineation in clinical data sets

AIM

Evaluation of a dedicated software tool for automatic delineation of 3D regions of interest in oncological PET.

PATIENTS, METHODS

The applied procedure encompasses segmentation of user-specified subvolumes within the tomographic data set into separate 3D ROIs, automatic background determination, and local adaptive thresholding of the background corrected data. Background correction and adaptive thresholding are combined in an iterative algorithm. Nine experienced observers used this algorithm for automatic delineation of a total of 37 ROIs in 14 patients. Additionally, the observers delineated the same ROIs also manually (using a freely chosen threshold for each ROI) and the results of automatic and manual ROI delineation were compared.

RESULTS

For the investigated 37 ROIs the manual delineation shows a strong interobserver variability of (26.8±6.3)% (range: 15% to 45%) while the corresponding value for automatic delineation is (1.1±1.0)% (range: <0.1% to 3.6%). The fractional deviation of the automatic volumes from the observer-averaged manual ones is (3.7±12.7)%.

CONCLUSION

The evaluated software provides results in very good agreement with observer-averaged manual evaluations, facilitates and accelerates the volumetric evaluation, eliminates the problem of interobserver variability and appears to be a useful tool for volumetric evaluation of oncological PET in clinical routine.

Evaluation of tomotherapy MVCT image enhancement program for tumor volume delineation

The aims of this study were to investigate the variability between physicians in delineation of head and neck tumors on original tomotherapy megavoltage CT (MVCT) studies and corresponding software enhanced MVCT images, and to establish an optimal approach for evaluation of image improvement. Five physicians contoured the gross tumor volume (GTV) for three head and neck cancer patients on 34 original and enhanced MVCT studies. Variation between original and enhanced MVCT studies was quantified by DICE coefficient and the coefficient of variance. Based on volume of agreement between physicians, higher correlation in terms of average DICE coefficients was observed in GTV delineation for enhanced MVCT for patients 1, 2, and 3 by 15%, 3%, and 7%, respectively, while delineation variance among physicians was reduced using enhanced MVCT for 12 of 17 weekly image studies. Enhanced MVCT provides advantages in reduction of variance among physicians in delineation of the GTV. Agreement on contouring by the same physician on both original and enhanced MVCT was equally high.

PACS numbers: 87.57.N‐, 87.57.np, 87.57.nt

A review of methods of analysis in contouring studies for radiation oncology

Inter-observer variability in anatomical contouring is the biggest contributor to uncertainty in radiation treatment planning. Contouring studies are frequently performed to investigate the differences between multiple contours on common datasets. There is, however, no widely accepted method for contour comparisons. The purpose of this study is to review the literature on contouring studies in the context of radiation oncology, with particular consideration of the contouring comparison methods they employ. A literature search, not limited by date, was conducted using Medline and Google Scholar with key words: contour, variation, delineation, inter/intra observer, uncertainty and trial dummy-run. This review includes a description of the contouring processes and contour comparison metrics used. The use of different processes and metrics according to tumour site and other factors were also investigated with limitations described. A total of 69 relevant studies were identified. The most common tumour sites were prostate (26), lung (10), head and neck cancers (8) and breast (7).The most common metric of comparison was volume used 59 times, followed by dimension and shape used 36 times, and centre of volume used 19 times. Of all 69 publications, 67 used a combination of metrics and two used only one metric for comparison. No clear relationships between tumour site or any other factors that may influence the contouring process and the metrics used to compare contours were observed from the literature. Further studies are needed to assess the advantages and disadvantages of each metric in various situations.

Effects of cold sphere walls in PET phantom measurements on the volume reproducing threshold

We studied quantitatively the effects of the discontinuity introduced in an otherwise homogeneous background by the cold walls of the standard spherical glass inserts commonly used in phantom measurements for calibration of threshold-based approaches to volumetric evaluation of PET investigations. We concentrated especially on the question of threshold-based volume determination. We computed analytically the convolution of an isotropic Gaussian point-spread function with the insert geometry (hot sphere + cold wall + warm background) and derived the theoretical background dependence of the volume reproducing threshold. This analysis shows a clear wall-related reduction of the optimal threshold with increasing background. The predictions of our theoretical analysis were verified in phantom measurements at background fractions between 0 and 0.29. Defining the background-corrected relative threshold [formula: see text] (T(abs): absolute volume reproducing threshold, A: measured activity at centre, B: background), we find that for a wall-less sphere T is independent of the background level. In the presence of cold walls, T drops (for not too small spheres, where recovery at the centre approaches 100%) from about 43% at B/A = 0 to about 25% at B/A = 0.5. Applying these thresholds to wall-less spheres leads to sizeable overestimates of the true volumes (43% at B/A = 0.5 for a sphere of 6 ml volume). We conclude that phantom measurements with standard sphere inserts for calibration of optimal thresholding algorithms introduce a systematic bias if performed at finite background levels. The observed background dependence is an artefact of the measurement procedure and does not reflect the conditions present in actual patient investigations.

Conformal radiotherapy for lung cancer: interobservers' variability in the definition of gross tumor volume between radiologists and radiotherapists

Background

Conformal external radiotherapy aims to improve tumor control by boosting tumor dose, reducing morbidity and sparing healthy tissues. To meet this objective careful visualization of the tumor and adjacent areas is required. However, one of the major issues to be solved in this context is the volumetric definition of the targets. This study proposes to compare the gross volume of lung tumors as delineated by specialized radiologists and radiotherapists of a cancer center.

Methods

Chest CT scans of a total of 23 patients all with non-small cell lung cancer, not submitted to surgery, eligible and referred to conformal radiotherapy on the Hospital A. C. Camargo (São Paulo, Brazil), during the year 2004 were analyzed. All cases were delineated by 2 radiologists and 2 radiotherapists. Only the gross tumor volume and the enlarged lymph nodes were delineated. As such, four gross tumor volumes were achieved for each one of the 23 patients.

Results

There was a significant positive correlation between the 2 measurements (among the radiotherapists, radiologists and intra-class) and there was randomness in the distribution of data within the constructed confidence interval.

Conclusion

There were no significant differences in the definition of gross tumor volume between radiologists and radiotherapists.

Physical considerations on discrepancies in target volume delineation

BACKGROUND AND PURPOSE

To compare the delineations and interpretations of target volumes by physicians in different radio-oncology centers.

MATERIALS AND METHODS

Eleven Swiss radio-oncology centers delineated volumes according to ICRU 50 recommendations for one prostate and one head and neck case. In order to evaluate the consistency of the volume delineations, the following parameters were determined: 1) the target volumes (GTV, CTV and manually expanded PTV) and their extensions in the three main axes and 2) the correlation of the volume delineated by each pair of centers using the ratio of the intersection to the union (called proximity index).

RESULTS

The delineated prostate volume was 105+/-55 cm(3) for the CTV and 218+/-44 cm(3) for the PTV. The delineated head and neck volume was 46+/-15 cm(3) for the GTV, 327+/-154 cm(3) for the CTV and 528+/-106 cm(3) for the PTV. The mean proximity index for the prostate case was 0.50+/-0.13 for the CTV and 0.57+/-0.11 for the PTV. The proximity index for the head and neck case was 0.45+/-0.09 for the GTV, 0.42+/-0.13 for the CTV and 0.59+/-0.06 for the PTV.

CONCLUSIONS

Large discrepancies between all the delineated target volumes were observed. There was an inverse relationship between the CTV volume and the margin between CTV and PTV, leading to less discrepancies in the PTV than is the CTV delineations. There was more spread in the sagittal and frontal planes due to CT pixel anisotropy, which suggests that radiation oncologists should delineate the target volumes not only in the transverse plane, but also in the sagittal and frontal planes to improve the delineation by allowing a consistency check.

Variability of target and normal structure delineation for breast cancer radiotherapy: an RTOG Multi-Institutional and Multiobserver Study

PURPOSE

To quantify the multi-institutional and multiobserver variability of target and organ-at-risk (OAR) delineation for breast-cancer radiotherapy (RT) and its dosimetric impact as the first step of a Radiation Therapy Oncology Group effort to establish a breast cancer atlas.

METHODS AND MATERIALS

Nine radiation oncologists specializing in breast RT from eight institutions independently delineated targets (e.g., lumpectomy cavity, boost planning target volume, breast, supraclavicular, axillary and internal mammary nodes, chest wall) and OARs (e.g., heart, lung) on the same CT images of three representative breast cancer patients. Interobserver differences in structure delineation were quantified regarding volume, distance between centers of mass, percent overlap, and average surface distance. Mean, median, and standard deviation for these quantities were calculated for all possible combinations. To assess the impact of these variations on treatment planning, representative dosimetric plans based on observer-specific contours were generated.

RESULTS

Variability in contouring the targets and OARs between the institutions and observers was substantial. Structure overlaps were as low as 10%, and volume variations had standard deviations up to 60%. The large variability was related both to differences in opinion regarding target and OAR boundaries and approach to incorporation of setup uncertainty and dosimetric limitations in target delineation. These interobserver differences result in substantial variations in dosimetric planning for breast RT.

CONCLUSIONS

Differences in target and OAR delineation for breast irradiation between institutions/observers appear to be clinically and dosimetrically significant. A systematic consensus is highly desirable, particularly in the era of intensity-modulated and image-guided RT.

Quality assurance of intensity-modulated radiation therapy

The current paradigm for the quality assurance (QA) program for intensity-modulated radiation therapy (IMRT) includes QA of the treatment planning system, QA of the delivery system, and patient-specific QA. Although the IMRT treatment planning and delivery system is the same as for conventional three-dimensional conformal radiation therapy, it has more parameters to coordinate and verify. Because of complex beam intensity modulation, each IMRT field often includes many small irregular off-axis fields, resulting in isodose distributions for each IMRT plan that are more conformal than those from conventional treatment plans. Therefore, these features impose a new and more stringent set of QA requirements for IMRT planning and delivery. The generic test procedures to validate dose calculation and delivery accuracy for both treatment planning and IMRT delivery have to be customized for each type of IMRT planning and delivery strategy. The rationale for such an approach is that the overall accuracy of IMRT delivery is incumbent on the piecewise uncertainties in both the planning and delivery processes. The end user must have well-defined evaluation criteria for each element of the planning and delivery process. Such information can potentially be used to determine a priori the accuracy of IMRT planning and delivery.

Conventionally-fractionated image-guided intensity modulated radiotherapy (IG-IMRT): a safe and effective treatment for cancer spinal metastasis

BACKGROUND

Treatments for cancer spinal metastasis were always palliative. This study was conducted to investigate the safety and effectiveness of IG-IMRT for these patients.

METHODS

10 metastatic lesions were treated with conventionally-fractionated IG-IMRT. Daily kilovoltage cone-beam computed tomography (kV-CBCT) scan was applied to ensure accurate positioning. Plans were evaluated by the dose-volume histogram (DVH) analysis.

RESULTS

Before set-up correction, the positioning errors in the left-right (LR), superior-inferior (SI) and anterior-posterior (AP) axes were 0.3 +/- 3.2, 0.4 +/- 4.5 and -0.2 +/- 3.9 mm, respectively. After repositioning, those errors were 0.1 +/- 0.7, 0 +/- 0.8 and 0 +/- 0.7 mm, respectively. The systematic/random uncertainties ranged 1.4-2.3/3.0-4.1 before and 0.1-0.2/0.7-0.8 mm after online set-up correction. In the original IMRT plans, the average dose of the planning target volume (PTV) was 61.9 Gy, with the spinal cord dose less than 49 Gy. Compared to the simulated PTVs based on the pre-correction CBCT, the average volume reduction of PTVs was 42.3% after online correction. Also, organ at risk (OAR) all benefited from CBCT-based set-up correction and had significant dose reduction with IGRT technique. Clinically, most patients had prompt pain relief within one month of treatment. There was no radiation-induced toxicity detected clinically during a median follow-up of 15.6 months.

CONCLUSION

IG-IMRT provides a new approach to treat cancer spinal metastasis. The precise positioning ensures the implementation of optimal IMRT plan, satisfying both the dose escalation of tumor targets and the radiation tolerance of spinal cord. It might benefit the cancer patient with spinal metastasis.

Early clinical evaluation of a novel three-dimensional structure delineation software tool (SCULPTER) for radiotherapy treatment planning

Modern radiotherapy treatment planning (RTP) necessitates increased delineation of target volumes and organs at risk. Conventional manual delineation is a laborious, time-consuming and subjective process. It is prone to inconsistency and variability, but has the potential to be improved using automated segmentation algorithms. We carried out a pilot clinical evaluation of SCULPTER (Structure Creation Using Limited Point Topology Evidence in Radiotherapy) - a novel prototype software tool designed to improve structure delineation for RTP. Anonymized MR and CT image datasets from patients who underwent radiotherapy for bladder or prostate cancer were studied. An experienced radiation oncologist used manual and SCULPTER-assisted methods to create clinically acceptable organ delineations. SCULPTER was also tested by four other RTP professionals. Resulting contours were compared by qualitative inspection and quantitatively by using the volumes of the structures delineated and the time taken for completion. The SCULPTER tool was easy to apply to both MR and CT images and diverse anatomical sites. SCULPTER delineations closely reproduced manual contours with no significant volume differences detected, but SCULPTER delineations were significantly quicker (p<0.05) in most cases. In conclusion, clinical application of SCULPTER resulted in rapid and simple organ delineations with equivalent accuracy to manual methods, demonstrating proof-of-principle of the SCULPTER system and supporting its potential utility in RTP.

Open low-field magnetic resonance imaging for target definition, dose calculations and set-up verification during three-dimensional CRT for glioblastoma multiforme

AIMS

To assess the effect on target delineation of using magnetic resonance simulation for planning of glioblastoma multiforme (GBM). Dose calculations derived from computed tomography- and magnetic resonance-derived plans were computed. The accuracy of set-up verification using magnetic resonance imaging (MRI)-based digital reconstructed radiographs (DRRs) was assessed.

MATERIALS AND METHODS

Ten patients with GBM were simulated using computed tomography and MRI. MRI was acquired with a low-field (0.23 T) MRI unit (SimMRI). Gross tumour volumes (GTVs) were delineated by two radiation oncologists on computed tomography and MRI. In total, 30 plans were generated using both the computed tomography, with (planbathoCT) and without (planCT) heterogeneity correction, and MRI data sets (planSimMRI). The minimum dose delivered (Dmin) to the GTV between computed tomography- and MRI-based plans was compared. The accuracy of set-up positioning using MRI DRRs was assessed by four radiation oncologists.

RESULTS

The mean GTVs delineated on computed tomography were significantly (P<0.001) larger than those contoured on MRI. The mean (+/-standard deviation) Dmin difference percentage was 0.3+/-0.8, 0.1+/-0.6 and -0.2+/-1.0% for the planCT/planbathoCT-, planCT/planSimMRI- and planbathoCT/planSimMRI-derived plans, respectively. The set-up differences observed with the computed tomography and MRI DRRs ranged from 1.0 to 4.0 mm (mean 1.5 mm; standard deviation+/-1.4).

CONCLUSIONS

GTVs defined on computed tomography were significantly larger than those delineated on MRI. Compared with computed tomography-derived plans, MRI-based dose calculations were accurate. The precision of set-up verifications based on computed tomography- and MRI-derived DRRs seemed similar. The use of MRI only for the planning of GBM should be further assessed.

A study of prostate delineation referenced against a gold standard created from the visible human data

PURPOSE

To measure inter- and intra-observer variation and systematic error in CT based prostate delineation, where individual delineations are referenced against a gold standard produced from photographic anatomical images from the Visible Human Project (VHP).

MATERIALS AND METHODS

The CT and anatomical images of the VHP male form the basic data set for this study. The gold standard was established based on 1mm thick anatomical photographic images. These were registered against the 3mm thick CT images that were used for target delineation. A total of 120 organ delineations were performed by six radiation oncologists.

RESULTS

The physician delineated prostate volume was on average 30% larger than the "true" prostate volume, but on average included only 84% of the gold standard volume. Our study found a systematic delineation error such that posterior portions of the prostate were always missed while anteriorly some normal tissue was always defined as target.

CONCLUSIONS

Our data suggest that radiation oncologists are more concerned with the unintentional inclusion of rectal tissue than they are in missing prostate volume. In contrast, they are likely to overextend the anterior boundary of the prostate to encompass normal tissue such as the bladder.

What's new in target volume definition for radiologists in ICRU Report 71? How can the ICRU volume definitions be integrated in clinical practice?

The optimal definition of the size, shape and location of gross tumour volume is one of the most important steps in the planning of radiation therapy, and necessitates a proper understanding of the procedure from both the oncologic radiologist and the radiation oncologist. This overview reports on the different terms and concepts that have been recommended in the ICRU Reports for this purpose; the latest Report 71 focuses on both previously given recommendations, and especially on electron beam therapy. This paper also highlights some of the problems that are encountered in the use of the International Commission on Radiation Units and Measurements (ICRU) recommendations in clinical practice, and at the interface between the radiation oncologist and the diagnostic oncologist.

Comparison and Evaluation of PET/CT Image Registration

Diagnosis, treatment planning and prognosis may be improved by combining functional data with anatomic location via PET/CT dual-modality image registration. The aim of this work is to reveal the spatial accuracy, reliability and speed of registration techniques used in GeminiTM PET/CT Imaging System, as a part of quality assurance program for clinical and research applications. A multi-layer alignment bearing source device was used to validate PET/CT intrinsic alignment, as well as a Rando a Man phantom. A series of known transformations were performed to simulate different misalignment between PET and CT images. Two semi-automated registration techniques used to correct misalignment were assessed and compared quantitatively by measuring absolute distances between the centroids of corresponding fiducial markers in the registered volumes. Comparisons between independent and cross fiducial localization indicate that fused PET/CT is superior to visual correlation of PET and CT individually in identifying registration errors. The experimental and theoretical findings confirm that PET/CT hybrid system can produce robust intrinsic image alignment and accurate PET/CT image registration with careful user interaction. The maximum spatial error in our experiments is below 4 mm, and is better than the spatial resolution of the PET scanner used. This is considered to be sufficient for most PET/CT applications.

Variability of gross tumor volume delineation in head-and-neck cancer using CT and PET/CT fusion

PURPOSE

To assess the need for gross tumor volume (GTV) delineation protocols in head-and-neck cancer (HNC) treatment planning by use of positron emission tomography (PET)/computed tomography (CT) fusion imaging. Assessment will consist of interobserver and intermodality variation analysis.

METHODS AND MATERIALS

Sixteen HNC patients were accrued for the study. Four physicians (2 neuroradiologists and 2 radiation oncologists) contoured GTV on 16 patients. Physicians were asked to contour GTV on the basis of the CT alone, and then on PET/CT fusion. Statistical analysis included analysis of variance for interobserver variability and Student's paired sample t test for intermodality and interdisciplinary variability. A Boolean pairwise analysis was included to measure degree of overlap.

RESULTS

Near-significant variation occurred across physicians' CT volumes (p = 0.09) and significant variation occurred across physicians' PET/CT volumes (p = 0.0002). The Boolean comparison correlates with statistical findings. One radiation oncologist's PET/CT fusion volumes were significantly larger than his CT volumes (p < 0.01). Conversely, the other radiation oncologist's CT volumes tended to be larger than his fusion volumes (p = 0.06). No significant interdisciplinary variation was seen. Significant disagreement occurred between radiation oncologists.

CONCLUSION

Significant differences in GTV delineation were found between multiple observers contouring on PET/CT fusion. The need for delineation protocol has been confirmed.