Purpose

Missing or discrepant imaging volume is a common challenge in deformable image registration (DIR). To minimize the adverse impact, we train a neural network to synthesize cropped portions of head and neck CT's and then test its use in DIR.

Methods

Using a training dataset of 409 head and neck CT's, we trained a generative adversarial network to take in a cropped 3D image and output an image with synthesized anatomy in the cropped region. The network used a 3D U-Net generator along with Visual Geometry Group (VGG) deep feature losses. To test our technique, for each of the 53 test volumes, we used Elastix to deformably register combinations of a randomly cropped, full, and synthetically full volume to a single cropped, full, and synthetically full target volume. We additionally tested our method's robustness to crop extent by progressively increasing the amount of cropping, synthesizing the missing anatomy using our network, and then performing the same registration combinations. Registration performance was measured using 95% Hausdorff distance across 16 contours.

Results

We successfully trained a network to synthesize missing anatomy in superiorly and inferiorly cropped images. The network can estimate large regions in an incomplete image, far from the cropping boundary. Registration using our estimated full images was not significantly different from registration using the original full images. The average contour matching error for full image registration was 9.9 mm, whereas our method was 11.6, 12.1, and 13.6 mm for synthesized-to-full, full-to-synthesized, and synthesized-to-synthesized registrations, respectively. In comparison, registration using the cropped images had errors of 31.7 mm and higher. Plotting the registered image contour error as a function of initial preregistered error shows that our method is robust to registration difficulty. Synthesized-to-full registration was statistically independent of cropping extent up to 18.7 cm superiorly cropped. Synthesized-to-synthesized registration was nearly independent, with a -0.04 mm of change in average contour error for every additional millimeter of cropping.

Conclusions

Different or inadequate in scan extent is a major cause of DIR inaccuracies. We address this challenge by training a neural network to complete cropped 3D images. We show that with image completion, the source of DIR inaccuracy is eliminated, and the method is robust to varying crop extent.

Central neurocytoma (CN) typically presents as an intraventricular mass causing obstructive hydrocephalus. The first line of treatment is surgical resection with adjuvant conventional radiotherapy. Stereotactic radiosurgery (SRS) was proposed as an alternative therapy for CN because of its lower risk profile. The objective of this systematic analysis is to assess the efficacy of SRS for CN. A systematic analysis for CN treated with SRS was conducted in PubMed. Baseline patient characteristics and outcomes data were extracted. Heterogeneity and publication bias were also assessed. Univariate and multivariate linear regressions were used to test for correlations to the primary outcome: local control (LC). The estimated cumulative rate of LC was 92.2% (95% confidence interval: 86.5-95.7%, p<0.001). Mean follow-up time was 62.4 months (range 3-149 months). Heterogeneity and publication bias were insignificant. The univariate linear regression models for both mean tumor volume and mean dose were significantly correlated with improved LC (p<0.001). Our data suggests that SRS may be an effective and safe therapy for CN. However, the rarity of CN still limits the efficacy of a quantitative analysis. Future multi-institutional, randomized trials of CN patients should be considered to further elucidate this therapy.

Purpose: To analyze geometric discrepancy and dosimetric impact in using contours generated by auto-segmentation (AS) against manually segmented (MS) clinical contours. Methods: A 48-subject prostate atlas was created and another 15 patients were used for testing. Contours were generated using a commercial atlas-based segmentation tool and compared to their clinical MS counterparts. The geometric correlation was evaluated using the Dice similarity coefficient (DSC) and Hausdorff distance (HD). Dosimetric relevance was evaluated for a subset of patients by assessing the DVH differences derived by optimizing plan dose using the AS and MS contours, respectively, and evaluating with respect to each. A paired t-test was employed for statistical comparison. The discrepancy in plan quality with respect to clinical dosimetric endpoints was evaluated. The analysis was repeated for head/neck (HN) with a 31-subject atlas and 15 test cases. Results: Dice agreement between AS and MS differed significantly across structures: from (L:0.92/R: 0.91) for the femoral heads to seminal vesical of 0.38 in the prostate cohort, and from 0.98 for the brain, to 0.36 for the chiasm of the HN group. Despite the geometric disagreement, the paired t-tests showed the lack of statistical evidence for systematic differences in dosimetric plan quality yielded by the AS and MS approach for the prostate cohort. In HN cases, statistically significant differences in dosimetric endpoints were observed in structures with small volumes or elongated shapes such as cord (p = 0.01) and esophagus (p = 0.04). The largest absolute dose difference of 11 Gy was seen in the mean pharynx dose. Conclusion: Varying AS performance among structures suggests a differential approach of using AS on a subset of structures and focus MS on the rest. The discrepancy between geometric and dosimetric-end-point driven evaluation also indicates the clinical utility of AS contours in optimization and evaluating plan quality despite of suboptimal geometrical accuracy.

This study evaluates the potential for tumor dose escalation in recurrent head and neck cancer (rHNC) patients with automated non-coplanar volumetric modulated arc therapy (VMAT) stereotactic body radiation therapy (SBRT) planning (HyperArc). Twenty rHNC patients are planned with conventional VMAT SBRT to 40 Gy while minimizing organ-at-risk (OAR) doses. They are then re-planned with the HyperArc technique to match these minimal OAR doses while escalating the target dose as high as possible. Then, we compare the dosimetry, tumor control probability (TCP), and normal tissue complication probability (NTCP) for the two plan types. Our results show that the HyperArc technique significantly increases the mean planning target volume (PTV) and gross tumor volume (GTV) doses by 10.8 ± 4.4 Gy (25%) and 11.5 ± 5.1 Gy (26%) on average, respectively. There are no clinically significant differences in OAR doses, with maximum dose differences of <2 Gy on average. The average TCP is 23% (± 21%) higher for HyperArc than conventional plans, with no significant differences in NTCP for the brainstem, cord, mandible, or larynx. HyperArc can achieve significant tumor dose escalation while maintaining minimal OAR doses in the head and neck-potentially enabling improved local control for rHNC SBRT patients without increased risk of treatment-related toxicities.

Purpose

To estimate the overall spatial distortion on clinical patient images for a 0.35 T MR-guided radiotherapy system.

Methods

Ten patients with head-and-neck cancer underwent CT and MR simulations with identical immobilization. The MR images underwent the standard systematic distortion correction post-processing. The images were rigidly registered and landmark-based analysis was performed by an anatomical expert. Distortion was quantified using Euclidean distance between each landmark pair and tagged by tissue interface: bone-tissue, soft tissue, or air-tissue. For baseline comparisons, an anthropomorphic phantom was imaged and analyzed.

Results

The average spatial discrepancy between CT and MR landmarks was 1.15 ± 1.14 mm for the phantom and 1.46 ± 1.78 mm for patients. The error histogram peaked at 0-1 mm. 66% of the discrepancies were <2 mm and 51% <1 mm. In the patient data, statistically significant differences (p-values < 0.0001) were found between the different tissue interfaces with averages of 0.88 ± 1.24 mm, 2.01 ± 2.20 mm, and 1.41 ± 1.56 mm for the air/tissue, bone/tissue, and soft tissue, respectively. The distortion generally correlated with the in-plane radial distance from the image center along the longitudinal axis of the MR.

Conclusion

Spatial distortion remains in the MR images after systematic distortion corrections. Although the average errors were relatively small, large distortions observed at bone/tissue interfaces emphasize the need for quantitative methods for assessing and correcting patient-specific spatial distortions.

This study reports the initial results for the first 15 patients on a prospective phase II clinical trial exploring the safety, feasibility, and efficacy of the HyperArc technique for recurrent head and neck cancer treatment. Eligible patients were simulated and planned with both conventional VMAT and HyperArc techniques and the plan with superior dosimetry was selected for treatment. Dosimetry, delivery feasibility and safety, treatment-related toxicity, and patient-reported quality of life (QOL) were all evaluated. HyperArc was chosen over conventional VMAT for all 15 patients and enabled statistically significant increases in dose conformity (R50% reduced by 1.2 ± 2.1, p < 0.05) and mean PTV and GTV doses (by 15.7 ± 4.9 Gy, p < 0.01 and 17.1 ± 6.0 Gy, p < 0.01, respectively). The average HyperArc delivery was 2.8 min longer than conventional VMAT (p < 0.01), and the mean intrafraction motion was ≤ 0.5 ± 0.4 mm and ≤0.3 ± 0.1°. With a median follow-up of 12 months, treatment-related toxicity was minimal (only one grade 3 acute toxicity above baseline) and patient-reported QOL metrics were favorable. HyperArc enabled superior dosimetry and significant target dose escalation compared to conventional VMAT planning, and treatment delivery was feasible, safe, and well-tolerated by patients.

Purpose

To develop and evaluate a fast patient localization tool using megavoltage (MV)-topogram on helical tomotherapy.

Methods and materials

Eighty-one MV-topogram pairs for 18 pelvis patients undergoing radiation were acquired weekly under an institutional review board-approved clinical trial. The MV-topogram imaging protocol requires 2 orthogonal acquisitions at static gantry angles of 0 degrees and 90 degrees for a programed scan length. A MATLAB based in-house software was developed to reconstruct the MV-topograms offline. Reference images (digitally reconstructed topograms, digitally reconstructed topograms) were generated using the planning computed tomography and tomotherapy geometry. The MV-topogram based alignment was determined by registering the MV-topograms to the digitally reconstructed topogram using bony landmark on commercial MIM software. The daily shifts in 3 translational directions determined from MV-topograms were compared with the megavoltage computed tomography (MVCT) based patient shifts. Linear-regression and two one-sided tests equivalence tests were performed to investigate the relation and equivalence between the 2 techniques. Seventy-eight MV-topogram pairs for 19 head and neck patients were included to validate the finding.

Results

The magnitudes of alignment differences of (MVCT - MV-topogram) (and standard deviations) were -0.3 ± 2.1, -0.8 ± 2.4, and 1.6 ± 1.7 mm for pelvis and 0.6 ± 1.2, 0.8 ± 4.2, 1.6 ± 2.6 mm for head and neck; the linear-regression coefficients between 2 imaging techniques were 1.18, 1.10, 0.94, and 0.86, 0.63, 0.38 in the lateral, longitudinal, vertical directions for pelvis and head and neck, respectively. The acquisition time for a pair of MV-topograms was up to 12.7 times less than MVCT scans (coarse scan mode) while covering longer longitudinal length.

Conclusions

MV-topograms showed equivalent clinical performance to the standard MVCT with significantly less acquisition time for pelvis and H&N patients. The MV-topogram can be used as an alternative or complimentary tool for bony landmark-based patient alignment on tomotherapy.

Background

There is great interest in finding ways to identify patients who will develop toxicity to cancer therapies. This has become especially pressing in the era of immune therapy, where toxicity can be long-lasting and life-altering, and primarily comes in the form of immune-related adverse effects (irAEs). Treatment with the first drugs in this class, anti-programmed death 1 (anti-PD1)/programmed death-ligand 1 (PDL1) checkpoint therapies, results in grade 2 or higher irAEs in up to 25%-30% of patients, which occur most commonly within the first 6 months of treatment and can include arthralgias, rash, pruritus, pneumonitis, diarrhea and/or colitis, hepatitis, and endocrinopathies. We tested the hypothesis that germline microRNA pathway functional variants, known to predict altered systemic stress responses to cancer therapies, would predict irAEs in patients across cancer types.

Methods

MicroRNA pathway variants were evaluated for an association with grade 2 or higher toxicity using four classifiers on 62 patients with melanoma, and then the panel's performance was validated on 99 patients with other cancer types. Trained classifiers included classification trees, LASSO-regularized logistic regression, boosted trees, and random forests. Final performance measures were reported on the training set using leave-one-out cross validation and validated on held-out samples. The predicted probability of toxicity was evaluated for its association, if any, with response categories to anti-PD1/PDL1 therapy in the melanoma cohort.

Results

A biomarker panel was identified that predicts toxicity with 80% accuracy (F1=0.76, area under the curve (AUC)=0.82) in the melanoma training cohort and 77.6% accuracy (F1=0.621, AUC=0.778) in the pan-cancer validation cohort. In the melanoma cohort, the predictive probability of toxicity was not associated with response categories to anti-PD1/PDL1 therapy (p=0.70). In the same cohort, the most significant biomarker of toxicity in RAC1, predicting a greater than ninefold increased risk of toxicity (p<0.001), was also not associated with response to anti-PD1/PDL1 therapy (p=0.151).

Conclusions

A germline microRNA-based biomarker signature predicts grade 2 and higher irAEs to anti-PD1/PDL1 therapy, regardless of tumor type, in a pan-cancer manner. These findings represent an important step toward personalizing checkpoint therapy, the use of which is growing rapidly.