Variabilities of Magnetic Resonance Imaging-, Computed Tomography-, and Positron Emission Tomography-Computed Tomography-Based Tumor and Lymph Node Delineations for Lung Cancer Radiation Therapy Planning

Automated segmentation of lesions and organs at risk on [68Ga]Ga-PSMA-11 PET/CT images using self-supervised learning with Swin UNETR

BACKGROUND

Prostate-specific membrane antigen (PSMA) PET/CT imaging is widely used for quantitative image analysis, especially in radioligand therapy (RLT) for metastatic castration-resistant prostate cancer (mCRPC). Unknown features influencing PSMA biodistribution can be explored by analyzing segmented organs at risk (OAR) and lesions. Manual segmentation is time-consuming and labor-intensive, so automated segmentation methods are desirable. Training deep-learning segmentation models is challenging due to the scarcity of high-quality annotated images. Addressing this, we developed shifted windows UNEt TRansformers (Swin UNETR) for fully automated segmentation. Within a self-supervised framework, the model's encoder was pre-trained on unlabeled data. The entire model was fine-tuned, including its decoder, using labeled data.

METHODS

In this work, 752 whole-body [68Ga]Ga-PSMA-11 PET/CT images were collected from two centers. For self-supervised model pre-training, 652 unlabeled images were employed. The remaining 100 images were manually labeled for supervised training. In the supervised training phase, 5-fold cross-validation was used with 64 images for model training and 16 for validation, from one center. For testing, 20 hold-out images, evenly distributed between two centers, were used. Image segmentation and quantification metrics were evaluated on the test set compared to the ground-truth segmentation conducted by a nuclear medicine physician.

RESULTS

The model generates high-quality OARs and lesion segmentation in lesion-positive cases, including mCRPC. The results show that self-supervised pre-training significantly improved the average dice similarity coefficient (DSC) for all classes by about 3%. Compared to nnU-Net, a well-established model in medical image segmentation, our approach outperformed with a 5% higher DSC. This improvement was attributed to our model's combined use of self-supervised pre-training and supervised fine-tuning, specifically when applied to PET/CT input. Our best model had the lowest DSC for lesions at 0.68 and the highest for liver at 0.95.

CONCLUSIONS

We developed a state-of-the-art neural network using self-supervised pre-training on whole-body [68Ga]Ga-PSMA-11 PET/CT images, followed by fine-tuning on a limited set of annotated images. The model generates high-quality OARs and lesion segmentation for PSMA image analysis. The generalizable model holds potential for various clinical applications, including enhanced RLT and patient-specific internal dosimetry.

[Research progress on the identification of central lung cancer and atelectasis using multimodal imaging]

Central lung cancer is a common disease in clinic which usually occurs above the segmental bronchus. It is commonly accompanied by bronchial stenosis or obstruction, which can easily lead to atelectasis. Accurately distinguishing lung cancer from atelectasis is important for tumor staging, delineating the radiotherapy target area, and evaluating treatment efficacy. This article reviews domestic and foreign literatures on how to define the boundary between central lung cancer and atelectasis based on multimodal images, aiming to summarize the experiences and propose the prospects.

Automated Lung Cancer Segmentation Using a PET and CT Dual-Modality Deep Learning Neural Network

PURPOSE

To develop an automated lung tumor segmentation method for radiation therapy planning based on deep learning and dual-modality positron emission tomography (PET) and computed tomography (CT) images.

METHODS AND MATERIALS

A 3-dimensional (3D) convolutional neural network using inputs from diagnostic PETs and simulation CTs was constructed with 2 parallel convolution paths for independent feature extraction at multiple resolution levels and a single deconvolution path. At each resolution level, the extracted features from the convolution arms were concatenated and fed through the skip connections into the deconvolution path that produced the tumor segmentation. Our network was trained/validated/tested by a 3:1:1 split on 290 pairs of PET and CT images from patients with lung cancer treated at our clinic, with manual physician contours as the ground truth. A stratified training strategy based on the magnitude of the gross tumor volume (GTV) was investigated to improve performance, especially for small tumors. Multiple radiation oncologists assessed the clinical acceptability of the network-produced segmentations.

RESULTS

The mean Dice similarity coefficient, Hausdorff distance, and bidirectional local distance comparing manual versus automated contours were 0.79 ± 0.10, 5.8 ± 3.2 mm, and 2.8 ± 1.5 mm for the unstratified 3D dual-modality model. Stratification delivered the best results when the model for the large GTVs (>25 mL) was trained with all-size GTVs and the model for the small GTVs (<25 mL) was trained with small GTVs only. The best combined Dice similarity coefficient, Hausdorff distance, and bidirectional local distance from the 2 stratified models on their corresponding test data sets were 0.83 ± 0.07, 5.9 ± 2.5 mm, and 2.8 ± 1.4 mm, respectively. In the multiobserver review, 91.25% manual versus 88.75% automatic contours were accepted or accepted with modifications.

CONCLUSIONS

By using an expansive clinical PET and CT image database and a dual-modality architecture, the proposed 3D network with a novel GTVbased stratification strategy generated clinically useful lung cancer contours that were highly acceptable on physician review.

Reduction of inter-observer variability using MRI and CT fusion in delineating of primary tumor for radiotherapy in lung cancer with atelectasis

Purpose

To compare the difference between magnetic resonance imaging (MRI) and computed tomography (CT) in delineating the target area of lung cancer with atelectasis.

Method

A retrospective analysis was performed on 15 patients with lung cancer accompanied by atelectasis. All positioning images were transferred to Eclipse treatment planning systems (TPSs). Six MRI sequences (T1WI, T1WI+C, T1WI+C Delay, T1WI+C 10 minutes, T2WI, DWI) were registered with positioning CT. Five radiation oncologists delineated the tumor boundary to obtain the gross tumor volume (GTV). Conformity index (CI) and dice coefficient (DC) were used to measure differences among observers.

Results

The differences in delineation mean volumes, CI, and DC among CT and MRIs were significant. Multiple comparisons were made between MRI sequences and CT. Among them, DWI, T2WI, and T1WI+C 10 minutes sequences were statistically significant with CT in mean volumes, DC, and CI. The mean volume of DWI, T2WI, and T1WI+C 10 minutes sequence in the target area is significantly smaller than that on the CT sequence, but the consistency is higher than that of CT sequences.

Conclusions

The recognition of atelectasis by MRI was better than that by CT, which could reduce interobserver variability of primary tumor delineation in lung cancer with atelectasis. Among them, DWI, T2WI, T1WI+C 10 minutes may be a better choice to improve the GTV delineation of lung cancer patients with atelectasis.

Unpaired Cross-Modality Educed Distillation (CMEDL) for Medical Image Segmentation

Accurate and robust segmentation of lung cancers from CT, even those located close to mediastinum, is needed to more accurately plan and deliver radiotherapy and to measure treatment response. Therefore, we developed a new cross-modality educed distillation (CMEDL) approach, using unpaired CT and MRI scans, whereby an informative teacher MRI network guides a student CT network to extract features that signal the difference between foreground and background. Our contribution eliminates two requirements of distillation methods: (i) paired image sets by using an image to image (I2I) translation and (ii) pre-training of the teacher network with a large training set by using concurrent training of all networks. Our framework uses an end-to-end trained unpaired I2I translation, teacher, and student segmentation networks. Architectural flexibility of our framework is demonstrated using 3 segmentation and 2 I2I networks. Networks were trained with 377 CT and 82 T2w MRI from different sets of patients, with independent validation (N = 209 tumors) and testing (N = 609 tumors) datasets. Network design, methods to combine MRI with CT information, distillation learning under informative (MRI to CT), weak (CT to MRI) and equal teacher (MRI to MRI), and ablation tests were performed. Accuracy was measured using Dice similarity (DSC), surface Dice (sDSC), and Hausdorff distance at the 95th percentile (HD95). The CMEDL approach was significantly (p < 0.001) more accurate (DSC of 0.77 vs. 0.73) than non-CMEDL methods with an informative teacher for CT lung tumor, with a weak teacher (DSC of 0.84 vs. 0.81) for MRI lung tumor, and with equal teacher (DSC of 0.90 vs. 0.88) for MRI multi-organ segmentation. CMEDL also reduced inter-rater lung tumor segmentation variabilities.

Variability of gross tumour volume delineation: MRI and CT based tumour and lymph node delineation for Lung radiotherapy

PURPOSE

To compare gross tumour volume (GTV) delineation of lung cancer on magnetic resonance imaging (MRI) and positron emission tomography (PET) versus computed tomography (CT) and PET.

METHODS

Three experienced thoracic radiation oncologists delineated GTVs on twenty-six patients with lung cancer, based on CT registered to PET, T2-weighted MRI registered to PET and T1-weighted MRI registered with PET. All observers underwent education on reviewing T1 and T2 images along with guidance on window and level setup. Interobserver and intermodality variation was performed based ondice similarity coefficient (DSC), Hausdorff distance (HD), and average Hausdorff distance (AvgHD) metrics. To compute interobserver variability (IOV) a simultaneous truth and performance level estimation (STAPLE) volume for each image modality was used as reference volume. For intermodality analysis, each observers CT based primary and nodal GTV was used as reference volume.

RESULTS

A mean DSC of 0.9 across all observers for primary GTV (GTVp) and a DSC of > 0.7 for nodal GTV (GTVn) was demonstrated for IOV. Mean T2 and T1 GTVp and GTVn were smaller than CT GTVp and GTVn but the difference in volume between modalities was not statistically significant. Significant difference (p<0.01) for GTVp and GTVn was found between T2 and T1 GTVp and GTVn compared to CT GTVp and GTVn based on DSC metrics. Large variation in volume similarity was noted based on HD of up-to 5.4cm for observer volumes compared to STAPLE volume.

CONCLUSION

Interobserver variability in GTV delineation was similar for MRI and PET versus CT and PET. The significant difference between MRI compared to CT delineated volumes needs to be further explored.

MRI and CBCT for lymph node identification and registration in patients with NSCLC undergoing radical radiotherapy

PURPOSE

This study compared MRI to CBCT for the identification and registration of lymph nodes (LN) in patients with locally advanced (LA)-NSCLC, to assess the suitability of targeting LNs in future MR-image guided radiotherapy (MRgRT) workflows.

METHOD

Radiotherapy radiographers carried out Visual Grading Analysis (VGA) assessment of image quality, LN registration and graded their confidence in registration for each of the 24 LNs on CBCT and two MR sequences, MR1 (T2w Turbo Spin Echo) and MR2 (T1w DIXON water only image).

RESULTS

Pre-registration image quality assessment revealed MR1 and MR2 as significantly superior to CBCT in terms of image quality (p ≤ 0.01). No significant differences were noted in interobserver variability for LN registration between CBCT, MR1 and MR2. Observers were more confident in their MR registrations compared to their CBCT based LN registrations (p ≤ 0.02).

SUMMARY

Interobserver setup correction variability was not found to be significantly different between CBCT and MR. Image quality and registration confidence were found to be superior for MRI sequences. This is a promising step towards MR-guided radiotherapy for the treatment of LA-NSCLC.

[Uncertainties in the current concept of radiotherapy planning target volume]

The planning target volume is an essential notion in radiotherapy, that requires a new conceptualization. Indeed, the variability and diversity of the uncertainties involved or improved with the development of the new modern technologies and devices in radiotherapy suggest that random and systematic errors cannot be currently generalized. This article attempts to discuss these various uncertainties and tries to demonstrate that a redefinition of the concept of planning target volume toward its personalization for each patient and the robustness notion are likely an improvement basis to take into account the radiotherapy uncertainties.

Comparison of gross tumor volumes of pulmonary metastasis defined by CT and MRI in 0.345 T MRI-guided radiotherapy

OBJECTIVE

To assess the difference in gross tumor volumes (GTVs) defined by CT (GTV-CT) and by low magnetic field strength (0.345 T) MRI (GTV-MRI) in patients simulated for MRI-guided radiotherapy forlung metastasis.

METHODS

28 patients (148 lesions) who underwent CT and MRI simulation with the tri-60Co MRI-guided radiotherapy system (MRIdian, ViewRay) were included in this study. GTV-CT and GTV-MRI were compared using the paired t-test. The equivalence of variance between GTV-CT and GTV-MRI of small lesions (GTV-CT <1 ml) and large ones (GTV-CT >= 1 ml) was evaluated using F-test. The correlation between GTV-CT and GTV-MRI was evaluated by the correlation coefficient.

RESULTS

GTV-MRI was 120% larger than GTV-CT (p < 0.001) for small lesions, whereas GTV-MRI was 40% larger than GTV-CT (p < 0.001) for large lesions. In small lesions, the variation in GTV-MRI was significantly larger than that of GTV-CT (p < 0.001). There was no significant difference in the variation of GTV-MRI and GTV-CT in large lesions (p = 0.121). The correlation coefficient for small lesions was 0.93, whereas that for large lesions was 0.99, with large lesions having better correlation.

CONCLUSIONS

GTV-MRI was larger than GTV-CT and the correlation between GTV-MRI and GTV-CT was better in large lesions. If the tumor volume is 1 ml or larger, the lesion can be accurately monitored even with a low magnetic field strength MRI.

ADVANCES IN KNOWLEDGE

This study is the first clinical report to evaluate the tolerability of MRI images in 0.345 T MRI-guided radiotherapy for lung metastasis. GTV contoured by MRI was larger than GTV by CT, and this tendency was more pronounced in small tumors of less than 1 ml.

Comparison of gross target volumes based on four-dimensional CT, positron emission tomography-computed tomography, and magnetic resonance imaging in thoracic esophageal cancer

PURPOSE

The application value of ¹⁸ F-FDG PET-CT combined with MRI in the radiotherapy of esophageal carcinoma was discussed by comparing the differences in position, volume, and the length of GTVs delineated on the end-expiration (EE) phase of 4DCT, ¹⁸ F-FDG PET-CT, and T₂ W-MRI.

METHODS

A total of 26 patients with thoracic esophageal cancer sequentially performed 3DCT, 4DCT, ¹⁸ F-FDG PET-CT, and MRI simulation for thoracic localization. All images were fused with the 3DCT images by deformable registration. GTV_CT and GTV_50% were delineated on 3DCT and the EE phase of 4DCT images, respectively. The GTV based on PET-CT images was determined by thresholds of SUV ≥ 2.5 and designated as GTV_PET2.5 . The images of T₂ -weighted sequence and diffusion-weighted sequence were referred as GTV_MRI and GTV_DWI , respectively. The length of the abnormality seen on the 4DCT, PET-CT, and DWI was compared.

RESULTS

GTV_PET2.5 was significantly larger than GTV_50% and GTV_MRI (P = .000 and 0.008, respectively), and the volume of GTV_MRI was similar to that of GTV_50% (P = .439). Significant differences were observed between the CI of GTV_MRI to GTV_50% and GTV_PET2.5 to GTV_50% (P = .004). The CI of GTV_MRI to GTV_CT and GTV_PET2.5 to GTV_CT were statistically significant (P = .039). The CI of GTV_MRI to GTV_PET2.5 was significantly lower than that of GTV_MRI to GTV_50% , GTV_MRI to GTV_CT , GTV_PET2.5 to GTV_50% , and GTV_PET2.5 to GTV_CT (P = .000-0.021). Tumor length measurements by endoscopy were similar to the tumor length as measured by PET and DWI scan (P > .05), and there was no significant difference between the longitudinal length of GTV_PET2.5 and GTV_DWI (P = .072).

CONCLUSION

The volumes of GTV_MRI and GTV_50% were similar. However, GTV_MRI has different volumes and poor spatial matching compared with GTV_PET2.5 .The MRI imaging could not include entire respiration. It may be a good choice to guide target delineation and construction of esophageal carcinoma by combining 4DCT with MRI imaging. Utilization of DWI in treatment planning for esophageal cancer may provide further information to assist with target delineation. Further studies are needed to determine if this technology will translate into meaningful differences in clinical outcome.

Performing clinical 18F-FDG-PET/MRI of the mediastinum optimising a dedicated, patient-friendly protocol

OBJECTIVE

To construct a mediastinal-specific fluorine-18-fluorodeoxyglucose (F-FDG)-PET/MR protocol with high-quality MRI of minimal acquisition-time and comparable diagnostic value to F-FDG-PET/computed tomography (CT).

MATERIALS AND METHODS

Fifteen healthy participants received PET/MRI and 10 patients with mediastinal tumours (eight non-small-cell lung, two oesophageal cancer) received F-FDG-PET/MRI immediately after F-FDG-PET/CT. Sequences volume interpolated breath-hold examination (T1-VIBE) and Half-Fourier acquisition single-shot turbo spin echo (T2-HASTE) were optimised by varying the parameters: breath-hold (BH, end-expiration), fat suppression (spectral adiabatic inversion recovery), and ECG-triggering (ECG, end-diastole). Image quality (IQ) of each sequence-variation was qualitatively scored by medical experts and quantitatively assessed by calculating signal-to-noise ratios, contrast relative to muscle, standardized-uptake-value, and tumour-to-blood ratios. Patient comfort was evaluated on patients' experience. Diagnostic accuracy of F-FDG-PET/MRI was compared to F-FDG-PET/CT, in reference to histopathology/cytopathology.

RESULTS

ECG-triggered T1-VIBE images showed the highest signal-to-noise ratio (P < 0.01) and the largest contrast between mediastinal soft-tissues, regardless of BH or free-breathing acquisition. IQ of ECG-triggered T1-VIBE scans in BH were scored qualitatively highest with good reader agreement (κ = 0.62). IQ of T2-HASTE was not significantly affected by BH acquisition (P > 0.9). Qualitative IQ of T1-VIBE and T2-HASTE declined after spectral adiabatic inversion recovery fat-suppression. All patients could maintain BH at end-expiration and reported no discomfort. Diagnostic performance of F-FDG-PET/MR was not significantly different from F-FDG-PET/CT with comparable staging, standardized-uptake-values, and tumour-to-blood ratios. However, T-status was more often over-staged on F-FDG-PET/CT, while N-status was more frequently under-staged on F-FDG-PET/MR.

CONCLUSION

ECG-triggered T1-VIBE sequences acquired during short, multiple BHs are recommended for mediastinal imaging using F-FDG-PET/MR. With dedicated protocols, F-FDG-PET/MRI will be useful in thoracic oncology and aid in diagnostic evaluation and tailored treatment decision-making.

Deep neural network for automatic characterization of lesions on 68Ga-PSMA-11 PET/CT

PURPOSE

This study proposes an automated prostate cancer (PC) lesion characterization method based on the deep neural network to determine tumor burden on ⁶⁸Ga-PSMA-11 PET/CT to potentially facilitate the optimization of PSMA-directed radionuclide therapy.

METHODS

We collected ⁶⁸Ga-PSMA-11 PET/CT images from 193 patients with metastatic PC at three medical centers. For proof-of-concept, we focused on the detection of pelvis bone and lymph node lesions. A deep neural network (triple-combining 2.5D U-Net) was developed for the automated characterization of these lesions. The proposed method simultaneously extracts features from axial, coronal, and sagittal planes, which mimics the workflow of physicians and reduces computational and memory requirements.

RESULTS

Among all the labeled lesions, the network achieved 99% precision, 99% recall, and an F1 score of 99% on bone lesion detection and 94%, precision 89% recall, and an F1 score of 92% on lymph node lesion detection. The segmentation accuracy is lower than the detection. The performance of the network was correlated with the amount of training data.

CONCLUSION

We developed a deep neural network to characterize automatically the PC lesions on ⁶⁸Ga-PSMA-11 PET/CT. The preliminary test within the pelvic area confirms the potential of deep learning methods. Increasing the amount of training data should further enhance the performance of the proposed method and may ultimately allow whole-body assessments.

Inter-observer variability in target delineation increases during adaptive treatment of head-and-neck and lung cancer

Introduction: Inter-observer variability (IOV) in target volume delineation is a well-documented source of geometric uncertainty in radiotherapy. Such variability has not yet been explored in the context of adaptive re-delineation based on imaging data acquired during treatment. We compared IOV in the pre- and mid-treatment setting using expert primary gross tumour volume (GTV) and clinical target volume (CTV) delineations in locoregionally advanced head-and-neck squamous cell carcinoma (HNSCC) and (non-)small cell lung cancer [(N)SCLC]. Material and methods: Five and six observers participated in the HNSCC and (N)SCLC arm, respectively, and provided delineations for five cases each. Imaging data consisted of CT studies partly complemented by FDG-PET and was provided in two separate phases for pre- and mid-treatment. Global delineation compatibility was assessed with a volume overlap metric (the Generalised Conformity Index), while local extremes of IOV were identified through the standard deviation of surface distances from observer delineations to a median consensus delineation. Details of delineation procedures, in particular, GTV to CTV expansion and adaptation strategies, were collected through a questionnaire. Results: Volume overlap analysis revealed a worsening of IOV in all but one case per disease site, which failed to reach significance in this small sample (p-value range .063-.125). Changes in agreement were propagated from GTV to CTV delineations, but correlation could not be formally demonstrated. Surface distance based analysis identified longitudinal target extent as a pervasive source of disagreement for HNSCC. High variability in (N)SCLC was often associated with tumours abutting consolidated lung tissue or potentially invading the mediastinum. Adaptation practices were variable between observers with fewer than half stating that they consistently adapted pre-treatment delineations during treatment. Conclusion: IOV in target volume delineation increases during treatment, where a disparity in institutional adaptation practices adds to the conventional causes of IOV. Consensus guidelines are urgently needed.

Chest Magnetic Resonance Imaging Decreases Inter-observer Variability of Gross Target Volume for Lung Tumors

Purpose: PET/CT is a standard medical imaging used in the delineation of gross tumor volume (GTV) in case of radiation therapy for lung tumors. However, PET/CT could present some limitations such as resolution and standardized uptake value threshold. Moreover, chest MRI has shown good potential in diagnosis for thoracic oncology. Therefore, we investigated the influence of chest MRI on inter-observer variability of GTV delineation.

Methods and Materials: Five observers contoured the GTV on CT for 14 poorly defined lung tumors during three contouring phases based on true daily clinical routine and acquisition: CT phase, with only CT images; PET phase, with PET/CT; and MRI phase, with both PET/CT and MRI. Observers waited at least 1 week between each phases to decrease memory bias. Contours were compared using descriptive statistics of volume, coefficient of variation (COV), and Dice similarity coefficient (DSC).

Results: MRI phase volumes (median 4.8 cm³) were significantly smaller than PET phase volumes (median 6.4 cm³, p = 0.015), but not different from CT phase volumes (median 5.7 cm³, p = 0.30). The mean COV was improved for the MRI phase (0.38) compared to the CT (0.58, p = 0.024) and PET (0.53, p = 0.060) phases. The mean DSC of the MRI phase (0.67) was superior to those of the CT and PET phases (0.56 and 0.60, respectively; p < 0.001 for both).

Conclusions: The addition of chest MRI seems to decrease inter-observer variability of GTV delineation for poorly defined lung tumors compared to PET/CT alone and should be explored in further prospective studies.

Does magnetic resonance imaging improve soft tissue sarcoma contouring for radiotherapy?

Objective:

Soft tissue sarcomas (STS) are a rare, heterogeneous tumour group. Radiotherapy improves local control. CT is used to plan radiotherapy, but has poor soft tissue definition. MRI has superior soft tissue definition. Contour variation amongst oncologists is an important factor in treatment failure. This study is the first to directly compare STS tumour contouring using CT vs MRI.

Methods:

Planning CT and T₂ weighted MR images of eight patients with STS were distributed to four oncologists. Gross tumour volume was contoured on both imaging modalities using in-house software. Images were recontoured 6 weeks later. The mean distance to agreement (DTA), standard deviation of the DTA, dice similarity coefficient (DSC) and contour volume were calculated for each oncologist and compared to a median contour volume. Results for CT and MRI were compared using a pairwise Student's t-test.

Results:

When comparing MRI to CT, tumour volumes were significantly smaller, with a difference of 21.4 cm³ across all patients (p = 0.008). There was not a statistically significant difference in the mean distance to agreement or dice similarity coefficient, but the standard deviation of the DTA showed a statistically significant improvement ( p = 0.04). For intraobserver variation, there was no statistically significant improvement using MRI vs CT.

Conclusion:

Oncologists contour smaller tumour volumes using MRI, with reduced interobserver variation. Improving the reliability and consistency of contouring is needed for improved quality assurance.

Advances in knowledge:

With further experience, the use of MRI in STS radiotherapy planning may reduce variation between oncologists and contribute to improved local control and reduced treatment toxicities.

The impact of a radiologist-led workshop on MRI target volume delineation for radiotherapy

INTRODUCTION

Magnetic resonance imaging (MRI) is increasingly used for target volume delineation in radiotherapy due to its superior soft tissue visualisation compared to computed tomography (CT). The aim of this study was to assess the impact of a radiologist-led workshop on inter-observer variability in volume delineation on MRI.

METHODS

Data from three separate studies evaluating the impact of MRI in lung, breast and cervix were collated. At pre-workshop evaluation, observers involved in each clinical site were instructed to delineate specified volumes. Radiologists specialising in each cancer site conducted an interactive workshop on interpretation of images and anatomy for each clinical site. At post-workshop evaluation, observers repeated delineation a minimum of 2 weeks after the workshops. Inter-observer variability was evaluated using dice similarity coefficient (DSC) and volume similarity (VOLSIM) index comparing reference and observer volumes.

RESULTS

Post-workshop primary gross tumour volumes (GTV) were smaller than pre-workshop volumes for lung with a mean percentage reduction of 10.4%. Breast clinical target volumes (CTV) were similar but seroma volumes were smaller post-workshop on both supine (65% reduction) and prone MRI (73% reduction). Based on DSC scores, improvement in inter-observer variability was seen for the seroma cavity volume on prone MRI with a reduction in DSC score range from 0.4-0.8 to 0.7-0.9. Breast CTV demonstrated good inter-observer variability scores (mean DSC 0.9) for both pre- and post-workshop. Post-workshop observer delineated cervix GTV was smaller than pre-workshop by 26.9%.

CONCLUSION

A radiologist-led workshop did not significantly reduce inter-observer variability in volume delineation for the three clinical sites. However, some improvement was noted in delineation of breast CTV, seroma volumes and cervix GTV.

Variability of Gross Tumor Volume Delineation for Stereotactic Body Radiotherapy of the Lung With Tri-60Co Magnetic Resonance Image-Guided Radiotherapy System (ViewRay): A Comparative Study With Magnetic Resonance- and Computed Tomography-Based Target Delineation

Introduction:

To evaluate the intra-/interobserver variability of gross target volumes between delineation based on magnetic resonance imaging and computed tomography in patients simulated for stereotactic body radiotherapy for primary lung cancer and lung metastasis.

Materials and Methods:

Twenty-five patients (27 lesions) who underwent computed tomography and magnetic resonance simulation with the MR-⁶⁰Co system (ViewRay) were included in the study. Gross target volumes were delineated on the magnetic resonance imaging (GTV_MR) and computed tomography (GTV_CT) images by 2 radiation oncologists (RO1 and RO2). Volumes of all contours were measured. Levels of intraobserver (GTV_MR_RO vs GTV_CT_RO) and interobserver (GTV_MR_RO1 vs GTV_MR_RO2; GTV_CT_RO1 vs GTV_CT_RO2) agreement were evaluated using the generalized κ statistics and the paired t test.

Results:

No significant volumetric difference was observed between all 4 comparisons (GTV_MR_RO1 vs GTV_CT_RO1, GTV_MR_RO2 vs GTV_CT_RO2, GTV_MR_RO1 vs GTV_MR_RO2, and GTV_CT_RO1 vs GTV_CT_RO2; P > .05), with mean volumes of GTVs ranging 5 to 6 cm³. The levels of agreement between those 4 comparisons were all substantial with mean κ values of 0.64, 0.66, 0.74, and 0.63, respectively. However, the interobserver agreement level was significantly higher for GTV_CT compared to GTV_MR (P <.001). The mean κ values significantly increased in all 4 comparisons for tumors >5 cm³ compared to tumors ≤5 cm³ (all P < .05).

Conclusion:

No significant differences in volumes between magnetic resonance- and computed tomograpghy-based Gross target volumes were found among 2 ROs. Magnetic resonance-based GTV delineation for lung stereotactic body radiotherapy also demonstrated acceptable interobserver agreement. Tumors >5 cm³ show higher intra-/interobserver agreement compared to tumors <5 cm³. More experience should be accumulated to reduce variability in magnetic resonance-based Gross target volumes delineation in lung stereotactic body radiotherapy.

Magnetic resonance imaging in precision radiation therapy for lung cancer

Radiotherapy remains the cornerstone of curative treatment for inoperable locally advanced lung cancer, given concomitantly with platinum-based chemotherapy. With poor overall survival, research efforts continue to explore whether integration of advanced radiation techniques will assist safe treatment intensification with the potential for improving outcomes. One advance is the integration of magnetic resonance imaging (MRI) in the treatment pathway, providing anatomical and functional information with excellent soft tissue contrast without exposure of the patient to radiation. MRI may complement or improve the diagnostic staging accuracy of F-18 fluorodeoxyglucose position emission tomography and computerized tomography imaging, particularly in assessing local tumour invasion and is also effective for identification of nodal and distant metastatic disease. Incorporating anatomical MRI sequences into lung radiotherapy treatment planning is a novel application and may improve target volume and organs at risk delineation reproducibility. Furthermore, functional MRI may facilitate dose painting for heterogeneous target volumes and prediction of normal tissue toxicity to guide adaptive strategies. MRI sequences are rapidly developing and although the issue of intra-thoracic motion has historically hindered the quality of MRI due to the effect of motion, progress is being made in this field. Four-dimensional MRI has the potential to complement or supersede 4D CT and 4D F-18-FDG PET, by providing superior spatial resolution. A number of MR-guided radiotherapy delivery units are now available, combining a radiotherapy delivery machine (linear accelerator or cobalt-60 unit) with MRI at varying magnetic field strengths. This novel hybrid technology is evolving with many technical challenges to overcome. It is anticipated that the clinical benefits of MR-guided radiotherapy will be derived from the ability to adapt treatment on the fly for each fraction and in real-time, using 'beam-on' imaging. The lung tumour site group of the Atlantic MR-Linac consortium is working to generate a challenging MR-guided adaptive workflow for multi-institution treatment intensification trials in this patient group.