High-resolution 3T to 7T ADC map synthesis with a hybrid CNN-transformer model

BACKGROUND

7 Tesla (7T) apparent diffusion coefficient (ADC) maps derived from diffusion-weighted imaging (DWI) demonstrate improved image quality and spatial resolution over 3 Tesla (3T) ADC maps. However, 7T magnetic resonance imaging (MRI) currently suffers from limited clinical unavailability, higher cost, and increased susceptibility to artifacts.

PURPOSE

To address these issues, we propose a hybrid CNN-transformer model to synthesize high-resolution 7T ADC maps from multimodal 3T MRI.

METHODS

The Vision CNN-Transformer (VCT), composed of both Vision Transformer (ViT) blocks and convolutional layers, is proposed to produce high-resolution synthetic 7T ADC maps from 3T ADC maps and 3T T1-weighted (T1w) MRI. ViT blocks enabled global image context while convolutional layers efficiently captured fine detail. The VCT model was validated on the publicly available Human Connectome Project Young Adult dataset, comprising 3T T1w, 3T DWI, and 7T DWI brain scans. The Diffusion Imaging in Python library was used to compute ADC maps from the DWI scans. A total of 171 patient cases were randomly divided into 130 training cases, 20 validation cases, and 21 test cases. The synthetic ADC maps were evaluated by comparing their similarity to the ground truth volumes with the following metrics: peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), and mean squared error (MSE). In addition, RESULTS: The results are as follows: PSNR: 27.0 ± 0.9 dB, SSIM: 0.945 ± 0.010, and MSE: 2.0E-3 ± 0.4E-3. Both qualitative and quantitative results demonstrate that VCT performs favorably against other state-of-the-art methods. We have introduced various efficiency improvements, including the implementation of flash attention and training on 176×208 resolution images. These enhancements have resulted in the reduction of parameters and training time per epoch by 50% in comparison to ResViT. Specifically, the training time per epoch has been shortened from 7.67 min to 3.86 min.

CONCLUSION

We propose a novel method to predict high-resolution 7T ADC maps from low-resolution 3T ADC maps and T1w MRI. Our predicted images demonstrate better spatial resolution and contrast compared to 3T MRI and prediction results made by ResViT and pix2pix. These high-quality synthetic 7T MR images could be beneficial for disease diagnosis and intervention, producing higher resolution and conformal contours, and as an intermediate step in generating synthetic CT for radiation therapy, especially when 7T MRI scanners are unavailable.

High-resolution MRI synthesis using a data-driven framework with denoising diffusion probabilistic modeling

Objective. High-resolution magnetic resonance imaging (MRI) can enhance lesion diagnosis, prognosis, and delineation. However, gradient power and hardware limitations prohibit recording thin slices or sub-1 mm resolution. Furthermore, long scan time is not clinically acceptable. Conventional high-resolution images generated using statistical or analytical methods include the limitation of capturing complex, high-dimensional image data with intricate patterns and structures. This study aims to harness cutting-edge diffusion probabilistic deep learning techniques to create a framework for generating high-resolution MRI from low-resolution counterparts, improving the uncertainty of denoising diffusion probabilistic models (DDPM).Approach. DDPM includes two processes. The forward process employs a Markov chain to systematically introduce Gaussian noise to low-resolution MRI images. In the reverse process, a U-Net model is trained to denoise the forward process images and produce high-resolution images conditioned on the features of their low-resolution counterparts. The proposed framework was demonstrated using T2-weighted MRI images from institutional prostate patients and brain patients collected in the Brain Tumor Segmentation Challenge 2020 (BraTS2020).Main results. For the prostate dataset, the bicubic interpolation model (Bicubic), conditional generative-adversarial network (CGAN), and our proposed DDPM framework improved the noise quality measure from low-resolution images by 4.4%, 5.7%, and 12.8%, respectively. Our method enhanced the signal-to-noise ratios by 11.7%, surpassing Bicubic (9.8%) and CGAN (8.1%). In the BraTS2020 dataset, the proposed framework and Bicubic enhanced peak signal-to-noise ratio from resolution-degraded images by 9.1% and 5.8%. The multi-scale structural similarity indexes were 0.970 ± 0.019, 0.968 ± 0.022, and 0.967 ± 0.023 for the proposed method, CGAN, and Bicubic, respectively.Significance. This study explores a deep learning-based diffusion probabilistic framework for improving MR image resolution. Such a framework can be used to improve clinical workflow by obtaining high-resolution images without penalty of the long scan time. Future investigation will likely focus on prospectively testing the efficacy of this framework with different clinical indications.

MGMT promoter methylation prediction based on multiparametric MRI via vision graph neural network

Purpose

Glioblastoma (GBM) is aggressive and malignant. The methylation status of the O 6 -methylguanine-DNA methyltransferase (MGMT) promoter in GBM tissue is considered an important biomarker for developing the most effective treatment plan. Although the standard method for assessing the MGMT promoter methylation status is via bisulfite modification and deoxyribonucleic acid (DNA) sequencing of biopsy or surgical specimens, a secondary automated method based on medical imaging may improve the efficiency and accuracy of those tests.

Approach

We propose a deep vision graph neural network (ViG) using multiparametric magnetic resonance imaging (MRI) to predict the MGMT promoter methylation status noninvasively. Our model was compared to the RSNA radiogenomic classification winners. The dataset includes 583 usable patient cases. Combinations of MRI sequences were compared. Our multi-sequence fusion strategy was compared with those using single MR sequences.

Results

Our best model [Fluid Attenuated Inversion Recovery (FLAIR), T1-weighted pre-contrast (T1w), T2-weighted (T2)] outperformed the winning models with a test area under the curve (AUC) of 0.628, an accuracy of 0.632, a precision of 0.646, a recall of 0.677, a specificity of 0.581, and an F1 score of 0.661. Compared to the winning models with single MR sequences, our ViG utilizing fused-MRI showed a significant improvement statistically in AUC scores, which are FLAIR (p = 0.042 ), T1w (p = 0.017 ), T1wCE (p = 0.001 ), and T2 (p = 0.018 ).

Conclusions

Our model is superior to challenge champions. A graph representation of the medical images enabled good handling of complexity and irregularity. Our work provides an automatic secondary check pipeline to ensure the correctness of MGMT methylation status prediction.

Hippocampus substructure segmentation using morphological vision transformer learning

The hippocampus plays a crucial role in memory and cognition. Because of the associated toxicity from whole brain radiotherapy, more advanced treatment planning techniques prioritize hippocampal avoidance, which depends on an accurate segmentation of the small and complexly shaped hippocampus. To achieve accurate segmentation of the anterior and posterior regions of the hippocampus from T1 weighted (T1w) MR images, we developed a novel model, Hippo-Net, which uses a cascaded model strategy. The proposed model consists of two major parts: (1) a localization model is used to detect the volume-of-interest (VOI) of hippocampus. (2) An end-to-end morphological vision transformer network (Franchietal2020Pattern Recognit.102107246, Ranemetal2022 IEEE/CVF Conf. on Computer Vision and Pattern Recognition Workshops (CVPRW) pp 3710-3719) is used to perform substructures segmentation within the hippocampus VOI. The substructures include the anterior and posterior regions of the hippocampus, which are defined as the hippocampus proper and parts of the subiculum. The vision transformer incorporates the dominant features extracted from MR images, which are further improved by learning-based morphological operators. The integration of these morphological operators into the vision transformer increases the accuracy and ability to separate hippocampus structure into its two distinct substructures. A total of 260 T1w MRI datasets from medical segmentation decathlon dataset were used in this study. We conducted a five-fold cross-validation on the first 200 T1w MR images and then performed a hold-out test on the remaining 60 T1w MR images with the model trained on the first 200 images. In five-fold cross-validation, the Dice similarity coefficients were 0.900 ± 0.029 and 0.886 ± 0.031 for the hippocampus proper and parts of the subiculum, respectively. The mean surface distances (MSDs) were 0.426 ± 0.115 mm and 0.401 ± 0.100 mm for the hippocampus proper and parts of the subiculum, respectively. The proposed method showed great promise in automatically delineating hippocampus substructures on T1w MR images. It may facilitate the current clinical workflow and reduce the physicians' effort.

A multi-site phase I trial of Veliparib with standard radiation and temozolomide in patients with newly diagnosed glioblastoma multiforme (GBM)

PURPOSE

A multi-site Phase I trial was conducted to determine the safety, maximum tolerated dose, and pharmacokinetics (PK) of Veliparib, a Poly (ADP-ribose) polymerase [PARP] enzyme inhibitor, when administered with temozolomide (TMZ) alone and then with temozolomide and radiation (RT) in patients with newly diagnosed glioblastoma.

METHODS

Given the potential for myelosuppression when a PARP inhibitor is combined with chemotherapy, the first 6 patients accrued were given Veliparib 10 mg bid and TMZ 75 mg/m2/d daily for six weeks. If this was well tolerated, the same doses of Veliparib and TMZ would be tested along with standard radiation with plans to dose escalate the Veliparib in subsequent patient cohorts. Once a maximal tolerated dose was determined, a 78 patient phase II study was planned. Peripheral blood pharmacokinetics were assessed.

RESULTS

Twenty-four patients were enrolled. In the first 6 patients who received 6 weeks of TMZ with Veliparib only one dose limiting toxicity (DLT) occurred. The next 12 patients received 6 weeks of RT + TMZ + veliparib and 4/12 (33%) had dose limiting hematologic toxicities. As a result, Veliparib was reduced by 50% to 10 mg BID every other week, but again 3/3 patients had dose limiting hematologic toxicities. The trial was then terminated. The mean clearance (± SD) CL/F of Veliparib for the initial dose (27.0 ± 9.0 L/h, n = 16) and at steady-state for 10 mg BID (23.5 ± 10.4 L/h, n = 18) were similar. Accumulation for BID dosing was 56% (± 33%).

CONCLUSIONS

Although Veliparib 10 mg BID administered with TMZ 75 mg/m2 for six weeks was well tolerated, when this regimen was combined with standard partial brain irradiation it was severely myelosuppressive even when the dose was reduced by 50%. This study again highlights the potential of localized cranial radiotherapy to significantly increase hematologic toxicity of marginally myelosuppressive systemic therapies.

A Multi-Site Phase I Trial of Veliparib with Standard Radiation and Temozolomide in Patients with Newly Diagnosed Glioblastoma Multiforme (GBM)

Purpose

A multi-site Phase I trial was conducted to determine the safety, maximum tolerated dose, and pharmacokinetics (PK) of Veliparib, a Poly (ADP-ribose) polymerase [PARP] enzyme inhibitor, when administered with temozolomide (TMZ) alone and then with temozolomide and radiation (RT) in patients with newly diagnosed glioblastoma.

Methods

Given the potential for myelosuppression when a PARP inhibitor is combined with chemotherapy, the first 6 patients accrued were given Veliparib 10 mg bid and TMZ 75 mg/m2/d daily for six weeks. If this was well tolerated, the same doses of Veliparib and TMZ would be tested along with standard radiation with plans to dose escalate the Veliparib in subsequent patient cohorts. Once a maximal tolerated dose was determined, a 78 patient phase II study was planned. Peripheral blood pharmacokinetics were assessed.

Results

Twenty-four patients were enrolled. In the first 6 patients who received 6 weeks of TMZ with Veliparib only one dose limiting toxicity (DLT) occurred. The next 12 patients received 6 weeks of RT + TMZ + veliparib and 4/12 (33%) had dose limiting hematologic toxicities. As a result, Veliparib was reduced by 50% to 10 mg BID every other week, but again 3/3 patients had dose limiting hematologic toxicities. The trial was then terminated. The mean clearance (± SD) CL/F of Veliparib for the initial dose (27.0 ± 9.0 L/h, n = 16) and at steady-state for 10 mg BID (23.5 ± 10.4 L/h, n = 18) were similar. Accumulation for BID dosing was 56% (± 33%).

Conclusions

Although Veliparib 10 mg BID administered with TMZ 75 mg/m2 for six weeks was well tolerated, when this regimen was combined with standard partial brain irradiation it was severely myelosuppressive even when the dose was reduced by 50%. This study again highlights the potential of localized cranial radiotherapy to significantly increase hematologic toxicity of marginally myelosuppressive systemic therapies.

Examining the Effect of ALK and EGFR Mutations on Survival Outcomes in Surgical Lung Brain Metastasis Patients

In the context of the post-genomic era, where targeted oncological therapies like monoclonal antibodies (mAbs) and tyrosine-kinase inhibitors (TKIs) are gaining prominence, this study investigates whether these therapies can enhance survival for lung carcinoma patients with specific genetic mutations-EGFR-amplified and ALK-rearranged mutations. Prior to this study, no research series had explored how these mutations influence patient survival in cases of surgical lung brain metastases (BMs). Through a multi-site retrospective analysis, the study examined patients who underwent surgical resection for BM arising from primary lung cancer at Emory University Hospital from January 2012 to May 2022. The mutational statuses were determined from brain tissue biopsies, and survival analyses were conducted. Results from 95 patients (average age: 65.8 ± 10.6) showed that while 6.3% had anaplastic lymphoma kinase (ALK)-rearranged mutations and 20.0% had epidermal growth factor receptor (EGFR)-amplified mutations-with 9.5% receiving second-line therapies-these mutations did not significantly correlate with overall survival. Although the sample size of patients receiving targeted therapies was limited, the study highlighted improved overall survival and progression-free survival rates compared to earlier trials, suggesting advancements in systemic lung metastasis treatment. The study suggests that as more targeted therapies emerge, the prospects for increased overall survival and progression-free survival in lung brain metastasis patients will likely improve.

Hippocampus Substructure Segmentation Using Morphological Vision Transformer Learning

Background

The hippocampus plays a crucial role in memory and cognition. Because of the associated toxicity from whole brain radiotherapy, more advanced treatment planning techniques prioritize hippocampal avoidance, which depends on an accurate segmentation of the small and complexly shaped hippocampus. Purpose: To achieve accurate segmentation of the anterior and posterior regions of the hippocampus from T1 weighted (T1w) MRI images, we developed a novel model, Hippo-Net, which uses a mutually enhanced strategy.

Methods

The proposed model consists of two major parts: 1) a localization model is used to detect the volume-of-interest (VOI) of hippocampus. 2) An end-to-end morphological vision transformer network is used to perform substructures segmentation within the hippocampus VOI. The substructures include the anterior and posterior regions of the hippocampus, which are defined as the hippocampus proper and parts of the subiculum. The vision transformer incorporates the dominant features extracted from MRI images, which are further improved by learning-based morphological operators. The integration of these morphological operators into the vision transformer increases the accuracy and ability to separate hippocampus structure into its two distinct substructures.A total of 260 T1w MRI datasets from Medical Segmentation Decathlon dataset were used in this study. We conducted a five-fold cross-validation on the first 200 T1w MR images and then performed a hold-out test on the remaining 60 T1w MR images with the model trained on the first 200 images. The segmentations were evaluated with two indicators, 1) multiple metrics including the Dice similarity coefficient (DSC), 95th percentile Hausdorff distance (HD95), mean surface distance (MSD), volume difference (VD) and center-of-mass distance (COMD); 2) Volumetric Pearson correlation analysis.

Results

In five-fold cross-validation, the DSCs were 0.900±0.029 and 0.886±0.031 for the hippocampus proper and parts of the subiculum, respectively. The MSD were 0.426±0.115mm and 0.401±0.100 mm for the hippocampus proper and parts of the subiculum, respectively. Conclusions: The proposed method showed great promise in automatically delineating hippocampus substructures on T1w MRI images. It may facilitate the current clinical workflow and reduce the physicians' effort.

Low-risk meningioma: Initial outcomes from NRG Oncology/RTOG 0539

BACKGROUND

Three- and five-year progression-free survival (PFS) for low-risk meningioma managed with surgery and observation reportedly exceeds 90%. Herewith we summarize outcomes for low-risk meningioma patients enrolled on NRG/RTOG 0539.

METHODS

This phase II trial allocated patients to one of three groups per World Health Organization grade, recurrence status, and resection extent. Low-risk patients had either gross total (GTR) or subtotal resection (STR) for a newly diagnosed grade 1 meningioma and were observed after surgery. The primary endpoint was 3-year PFS. Adverse events (AEs) were scored using Common Terminology Criteria for Adverse Events (CTCAE) version 3.

RESULTS

Among 60 evaluable patients, the median follow-up was 9.1 years. The 3-, 5-, and 10-year rates were 91.4% (95% CI, 84.2 to 98.6), 89.4% (95% CI, 81.3 to 97.5), 85.0% (95% CI, 75.3 to 94.7) for PFS and 98.3% (95% CI, 94.9 to 100), 98.3%, (95% CI, 94.9 to 100), 93.8% (95% CI, 87.0 to 100) for overall survival (OS), respectively. With centrally confirmed GTR, 3/5/10y PFS and OS rates were 94.3/94.3/87.6% and 97.1/97.1/90.4%. With STR, 3/5/10y PFS rates were 83.1/72.7/72.7% and 10y OS 100%. Five patients reported one grade 3, four grade 2, and five grade 1 AEs. There were no grade 4 or 5 AEs.

CONCLUSIONS

These results prospectively validate high PFS and OS for low-risk meningioma managed surgically but raise questions regarding optimal management following STR, a subcohort that could potentially benefit from adjuvant therapy.

Osteoradionecrosis of the craniotomy flap: a rare complication of stereotactic radiosurgery

Osteoradionecrosis (ORN), ischemic necrosis of irradiated bone without evidence of persisting or recurrent tumor, is a known complication of radiation therapy. ORN of the skull has not been reported following stereotactic radiosurgery (SRS). We report two cases of ORN of the skull following SRS for recurrent meningiomas post-resection. Both patients developed ORN in their craniotomy flaps in areas that received high doses of radiation due to their proximity to the recurrent tumors. In each case, the ORN was asymptomatic and was detected on surveillance magnetic resonance imaging. Both patients were followed closely with imaging that ultimately revealed either stability or improvement in the ORN, confirming the diagnosis without the need for biopsy. The cases reveal a role for close imaging surveillance instead of immediate biopsy in patients with new enhancement involving bone in high-dose radiation treatment regions.

Cascaded mutual enhancing networks for brain tumor subregion segmentation in multiparametric MRI

Accurate segmentation of glioma and its subregions plays an important role in radiotherapy treatment planning. Due to a very populated multiparameter magnetic resonance imaging image, manual segmentation tasks can be very time-consuming, meticulous, and prone to subjective errors. Here, we propose a novel deep learning framework based on mutual enhancing networks to automatically segment brain tumor subregions. The proposed framework is suitable for the segmentation of brain tumor subregions owing to the contribution of Retina U-Net followed by the implementation of a mutual enhancing strategy between the classification localization map (CLM) module and segmentation module. Retina U-Net is trained to accurately identify view-of-interest and feature maps of the whole tumor (WT), which are then transferred to the CLM module and segmentation module. Subsequently, CLM generated by the CLM module is integrated with the segmentation module to bring forth a mutual enhancing strategy. In this way, our proposed framework first focuses on WT through Retina U-Net, and since WT consists of subregions, a mutual enhancing strategy then further aims to classify and segment subregions embedded within WT. We implemented and evaluated our proposed framework on the BraTS 2020 dataset consisting of 369 cases. We performed a 5-fold cross-validation on 200 datasets and a hold-out test on the remaining 169 cases. To demonstrate the effectiveness of our network design, we compared our method against the networks without Retina U-Net, mutual enhancing strategy, and a recently published Cascaded U-Net architecture. Results of all four methods were compared to the ground truth for segmentation and localization accuracies. Our method yielded significantly (P < 0.01) better values of dice-similarity-coefficient, center-of-mass-distance, and volume difference compared to all three competing methods across all tumor labels (necrosis and non-enhancing, edema, enhancing tumor, WT, tumor core) on both validation and hold-out dataset. Overall quantitative and statistical results of this work demonstrate the ability of our method to both accurately and automatically segment brain tumor subregions.

Automatic Inverse Treatment Planning of Gamma Knife Radiosurgery via Deep Reinforcement Learning

PURPOSE

Several inverse planning algorithms have been developed for Gamma Knife (GK) radiosurgery to determine a large number of plan parameters via solving an optimization problem, which typically consists of multiple objectives. The priorities among these objectives need to be repetitively adjusted to achieve a clinically good plan for each patient. This study aimed to achieve automatic and intelligent priority-tuning, by developing a deep reinforcement learning (DRL) based method to model the tuning behaviors of human planners.

METHODS

We built a priority-tuning policy network using deep convolutional neural networks. Its input was a vector composed of multiple plan metrics that were used in our institution for GK plan evaluation. The network can determine which tuning action to take, based on the observed quality of the intermediate plan. We trained the network using an end-to-end DRL framework to approximate the optimal action-value function. A scoring function was designed to measure the plan quality to calculate the received reward of a tuning action.

RESULTS

Vestibular schwannoma was chosen as the test bed in this study. The number of training, validation and testing cases were 5, 5, and 16, respectively. For these three datasets, the average scores of the initial plans obtained with a same initial priority set were 3.63 ± 1.34, 3.83 ± 0.86 and 4.20 ± 0.78, respectively, while can be improved to 5.28 ± 0.23, 4.97 ± 0.44 and 5.22 ± 0.26 through manual priority tuning by human expert planners. Our network achieved competitive results with 5.42 ± 0.11, 5.10 ± 0. 42, 5.28 ± 0.20, respectively.

CONCLUSIONS

Our network can generate GK plans of comparable or slightly higher quality comparing with the plans generated by human planners via manual priority tuning for vestibular schwannoma cases. The network can potentially be incorporated into the clinical workflow as a planning assistance to improve GK planning efficiency and help to reduce plan quality variation caused by inter-planner variability. We also hope that our method can reduce the workload of GK planners and allow them to spend more time on more challenging cases. This article is protected by copyright. All rights reserved.

Race, Ethnicity, and Sex Among Senior Faculty in Radiation Oncology From 2000 to 2019

This cross-sectional study investigates intersections among race, ethnicity, and sex from 2000 to 2019 among senior faculty in radiation oncology.

MRI classification using semantic random forest with auto-context model

BACKGROUND

It is challenging to differentiate air and bone on MR images of conventional sequences due to their low contrast. We propose to combine semantic feature extraction under auto-context manner into random forest to improve reasonability of the MRI segmentation for MRI-based radiotherapy treatment planning or PET attention correction.

METHODS

We applied a semantic classification random forest (SCRF) method which consists of a training stage and a segmentation stage. In the training stage, patch-based MRI features were extracted from registered MRI-CT training images, and the most informative elements were selected via feature selection to train an initial random forest. The rest sequence of random forests was trained by a combination of MRI feature and semantic feature under an auto-context manner. During segmentation, the MRI patches were first fed into these random forests to derive patch-based segmentation. By using patch fusion, the final end-to-end segmentation was obtained.

RESULTS

The Dice similarity coefficient (DSC) for air, bone and soft tissue classes obtained via proposed method were 0.976±0.007, 0.819±0.050 and 0.932±0.031, compared to 0.916±0.099, 0.673±0.151 and 0.830±0.083 with random forest (RF), and 0.942±0.086, 0.791±0.046 and 0.917±0.033 with U-Net. SCRF also outperformed the competing methods in sensitivity and specificity for all three structure types.

CONCLUSIONS

The proposed method accurately segmented bone, air and soft tissue. It is promising in facilitating advanced MR application in diagnosis and therapy.

SYST-07. PILOT STUDY UTILIZING THE HDAC INHIBITOR BELINOSTAT WITH CHEMORADIATION FOR NEWLY-DIAGNOSED GLIOBLASTOMA

PURPOSE

Glioblastomas (GBMs) are highly aggressive brain tumors with poor prognosis. Belinostat is a histone deacetylase inhibitor with blood–brain barrier permeability, anti-GBM activity, and potential to enhance chemoradiation. This clinical trial sought to determine a tolerable dose of concurrent belinostat and assess the clinical efficacy of combining this drug with standard-of-care therapy.

METHODS

13 patients each were enrolled in control and belinostat cohorts. The belinostat cohort was given a belinostat regimen (500-750mg/m2 1x/day x 5 days) every 3 weeks (weeks 0, 3, and 6 of RT). All patients received standard temozolomide and radiation therapy (RT). Patient outcomes included progression-free survival, overall survival (OS), and analysis of recurrence pattern of the recurrent gross tumor volume (rGTV).

RESULTS

Belinostat at 750 mg/m2 produce dose-limiting toxicities (DLTs) in 2 of 3 patients while belinostat at 500 mg/m2 did not result in DLTs. Median OS was 18.5 months for the belinostat cohort and 15.8 months for the control cohort (p=0.53). The rGTVs in the control cohort occurred in areas that received higher radiation doses than that in the belinostat cohort. For those belinostat patients that experienced out-of-field recurrences, tumors were detectable by spectroscopic MRI (sMRI) before RT. In particular, one belinostat patient had an IDH-mutant GBM that had an extraordinary response to therapy with significant shrinkage of enhancing tumor much greater than expected.

CONCLUSION

Belinostat given concurrently at 500 mg/m2 is well-tolerated. While median OS was not significantly increased for the belinostat cohort, recurrence analysis suggests better in-field control with belinostat, suggesting a radio-sensitizing effect. This study suggests that belinostat can act as a synergistic therapeutic agent for GBMs that may be further enhanced by sMRI-guided RT and may be particularly effective against IDH mutant tumors. A trial is currently in development using belinostat with sMRI-guided RT for IDH-mutant high-grade gliomas.

CTNI-13. UPDATES ON CLINICAL OUTCOMES AND TUMOR RECURRENCE PATTERNS OF A HUMAN PILOT STUDY ASSESSING EFFICACY OF BELINOSTAT (PXD-101) COMBINING WITH CHEMORADIATION IN TREATING GLIOBLASTOMA

INTRODUCTION

Glioblastoma (GBM) is highly aggressive with poor prognosis. Belinostat is a histone deacetylase inhibitor with blood–brain barrier permeability that has anti-GBM activity and may enhance effects of chemoradiation. Our institution conducted a clinical trial evaluating clinical efficacy of belinostat with standard-of-care therapy for GBMs.

METHODS

13 and 14 patients were enrolled into cohort 1 (c1, control) or cohort 2 (c2, belinostat) with 12 in each group with sufficient follow-up MRIs for recurrence analysis. All patients received concurrent, adjuvant temozolomide and focal radiation therapy (RT). For c2 patients, the belinostat regimen (500-750mg/m2 1x/day x 5 days) was given over three cycles every 3 weeks (weeks -1, 2, and 5 of RT). RT margins of 5–10 mm and 3 mm were added to generate clinical tumor volumes and planning target volumes (PTVs). PTV1 (based on FLAIR MRI) and PTV2 (based on CE-T1w MRI) received 51 and 60 Gy, respectively, over 30 fractions. Volume at initial recurrence (rGTV) was contoured.

RESULTS

Mean age was 58.3 years for c1 and 51.1 years for c2. Patient/tumor characteristics were similar between cohorts. Median OS were 16.6 and 18.5 months for c1 and c2 (p=0.538), respectively. Average minimum, maximum and mean radiation dose to rGTV was 54.1 Gy, 64.2 Gy and 62 Gy, for c1, and 47.5 Gy, 57.6 Gy and 53.5 Gy, for c2 (p=0.322, 0.088 and 0.071), respectively. The mean overlap between rGTV and PTV1/PTV2 for c1 & c2 were 99.2% & 96.9%/99.8% & 78.7% (p=0.489/0.133), respectively.

CONCLUSION

Median OS was slightly longer for c2 though not statistically significant. rGTV in c1 received higher radiation doses and had more overlap with PTV2 than in c2. Out-of-field recurrence appears more likely in c2 suggesting better infield control with belinostat. This study highlights the potential of belinostat as a synergistic therapeutic agent for GBM treatment.

Brain tumor segmentation using 3D Mask R-CNN for dynamic susceptibility contrast enhanced perfusion imaging

The segmentation of neoplasms is an important part of radiotherapy treatment planning, monitoring disease progression, and predicting patient outcome. In the brain, functional magnetic resonance imaging (MRI) like dynamic susceptibility contrast enhanced (DSCE) or T1-weighted dynamic contrast enhanced (DCE) perfusion MRI are important tools for diagnosis. They play a crucial role in providing pre-operative assessment of tumor histology, grading, and biopsy guidance. However, the manual contouring of these neoplasms is tedious, expensive, time-consuming, and vulnerable to inter-observer variability. In this work, we propose a 3D mask region-based convolutional neural network (R-CNN) method to automatically segment brain tumors in DSCE MRI perfusion images. As our goal is to simultaneously localize and segment the tumor, our training process contained both a region-of-interest (ROI) localization and regression with voxel-wise segmentation. The combination of classification loss, ROI location and size regression loss, and segmentation loss were used to supervise the proposed network. We retrospectively investigated 21 patients' perfusion images, with between 50 and 70 perfusion time point volumes, a total of 1260 3D volumes. Tumor contours were automatically segmented by our proposed method and compared against other state-of-the-art methods and those delineated by physicians as the ground truth. The results of our method demonstrated good agreement with the ground truth contours. The average DSC, precision, recall, Hausdorff distance, mean surface distance (MSD), root MSD, and center of mass distance were 0.90 ± 0.04, 0.91 ± 0.04, 0.90 ± 0.06, 7.16 ± 5.78 mm, 0.45 ± 0.34 mm, 1.03 ± 0.72 mm, and 0.86 ± 0.91 mm, respectively. These results support the feasibility of our method in accurately localizing and segmenting brain tumors in DSCE perfusion MRI. Our 3D Mask R-CNN segmentation method in DSCE perfusion imaging has great promise for future clinical use.

Clinical Trial Design and Development Work Group Within the Quantitative Imaging Network

The Clinical Trial Design and Development Working Group within the Quantitative Imaging Network focuses on providing support for the development, validation, and harmonization of quantitative imaging (QI) methods and tools for use in cancer clinical trials. In the past 10 years, the Group has been working in several areas to identify challenges and opportunities in clinical trials involving QI and radiation oncology. The Group has been working with Quantitative Imaging Network members and the Quantitative Imaging Biomarkers Alliance leadership to develop guidelines for standardizing the reporting of quantitative imaging. As a validation platform, the Group led a multireader study to test a semi-automated positron emission tomography quantification software. Clinical translation of QI tools cannot be possible without a continuing dialogue with clinical users. This article also highlights the outreach activities extended to cooperative groups and other organizations that promote the use of QI tools to support clinical decisions.

The Brain Imaging Collaboration Suite (BrICS): A Cloud Platform for Integrating Whole-Brain Spectroscopic MRI into the Radiation Therapy Planning Workflow

Glioblastoma has poor prognosis with inevitable local recurrence despite aggressive treatment with surgery and chemoradiation. Radiation therapy (RT) is typically guided by contrast-enhanced T1-weighted magnetic resonance imaging (MRI) for defining the high-dose target and T2-weighted fluid-attenuation inversion recovery MRI for defining the moderate-dose target. There is an urgent need for improved imaging methods to better delineate tumors for focal RT. Spectroscopic MRI (sMRI) is a quantitative imaging technique that enables whole-brain analysis of endogenous metabolite levels, such as the ratio of choline-to-N-acetylaspartate. Previous work has shown that choline-to-N-acetylaspartate ratio accurately identifies tissue with high tumor burden beyond what is seen on standard imaging and can predict regions of metabolic abnormality that are at high risk for recurrence. To facilitate efficient clinical implementation of sMRI for RT planning, we developed the Brain Imaging Collaboration Suite (BrICS; https://brainimaging.emory.edu/brics-demo), a cloud platform that integrates sMRI with standard imaging and enables team members from multiple departments and institutions to work together in delineating RT targets. BrICS is being used in a multisite pilot study to assess feasibility and safety of dose-escalated RT based on metabolic abnormalities in patients with glioblastoma (Clinicaltrials.gov NCT03137888). The workflow of analyzing sMRI volumes and preparing RT plans is described. The pipeline achieved rapid turnaround time by enabling team members to perform their delegated tasks independently in BrICS when their clinical schedules allowed. To date, 18 patients have been treated using targets created in BrICS and no severe toxicities have been observed.

High-risk Meningioma: Initial Outcomes From NRG Oncology/RTOG 0539

BACKGROUND

Phase 2 cooperative group meningioma trial assessing the safety and efficacy of risk-adaptive management strategies. This is the initial analysis of the high-risk cohort.

METHODS AND MATERIALS

High-risk patients were those with a new or recurrent World Health Organization (WHO) grade III meningioma of any resection extent, recurrent WHO grade II of any resection extent, or new WHO grade II after subtotal resection. Patients received intensity-modulated radiotherapy (IMRT) using a simultaneous integrated boost technique (60 Gy high dose and 54 Gy low dose in 30 fractions). Three-year progression-free survival (PFS) was the primary endpoint. Adverse events (AEs) were scored per NCI Common Terminology Criteria for Adverse Events version 3.

RESULTS

Of 57 enrolled patients, 53 received protocol treatment. Median follow-up was 4.0 years (4.8 years for living patients). Two patients withdrew without progression before year 3; for the remaining 51 patients, 3-year PFS was 58.8%. Among all 53 protocol-treated patients, 3-year PFS was 59.2%. Three-year local control was 68.9%, and overall survival was 78.6%. Of 51 patients, 1 patient (1.9%) experienced a late grade-5 necrosis-related AE. All other acute (23 of 53 patients) and late (21 of 51 patients) AEs were grades 1 to 3.

CONCLUSIONS

Patients with high-risk meningioma treated with IMRT (60 Gy/30) experienced 3-year PFS of 58.8%. Combined acute and late AEs were limited to grades 1 to 3, except for a single necrosis-related grade 5 event. These results support postoperative IMRT for high-risk meningioma and invite ongoing investigations to improve outcomes further.

ACTR-70. A MULTISITE CLINICAL TRIAL OF SPECTROSCOPIC MRI-GUIDED RADIATION DOSE ESCALATION IN GLIOBLASTOMA PATIENTS

The standard of care for glioblastoma is neurosurgical resection followed by radiation therapy (RT) and temozolomide (TMZ) chemotherapy. Although RT is effective for glioblastoma, attempts to improve survival using RT dose escalation above 60 Gy have been largely unsuccessful. This may be because prior attempts have been targeted to resection cavity or enhancing tumor, which may not accurately predict areas at highest recurrence risk. To overcome the limitations of standard T1 and T2-weighted MRI in predicting tumor recurrence, we have shown that supplementation with 3D high-resolution spectroscopic MRI (sMRI) identifies actively proliferating tumor beyond areas of T1-enhancement by measuring endogenous metabolites and their ratios. Previously, we demonstrated that the choline to N-acetylaspartate ratio (Cho/NAA) best correlates with tumor cellularity in surgically resected tissue (ρ=0.82, p< 0.001) and, most importantly, areas of sMRI metabolic abnormalities predate disease recurrence in those same areas (Cordova et al, Neuro-Oncology 2016). Therefore, we seek to identify whether sMRI can be used by radiation oncologists to choose the optimal regions to target for RT dose escalation. To assess its feasibility and safety, we developed a web-based imaging platform designed specifically to incorporate sMRI into the RT planning clinical workflow and are using it in a multisite sMRI-guided dose escalation trial (NCT03137888; Emory, Johns Hopkins, U. Miami). Recently, we have completed full enrollment including 30 patients treated with sMRI-guided dose escalated RT across three institutions. We have demonstrated successful integration of sMRI into the RT planning workflow, and we have delivered sMRI-guided dose escalated RT plans to glioblastoma patients without severe adverse events to date. Follow-up data will be analyzed for overall and progression-free survival. Based on the feasibility and safety of this technique in the current trial, we plan to assess the efficacy of sMRI-guided dose-escalated RT on patient outcomes in a NCTN clinical trial.

RTHP-01. PENTOXIFYLLINE AND VITAMIN E FOR THE TREATMENT OF RADIATION NECROSIS AFTER STEREOTACTIC RADIOSURGERY

BACKGROUND

The combination of oral pentoxifylline (Ptx) and vitamin E (VitE) has been used to treat radiation-induced fibrosis and soft tissue injury. A 2008 study showed an average decrease in edema with the use of Ptx + VitE in 11 patients with suspected radiation necrosis after stereotactic radiosurgery (SRS). Here we review outcomes in patients prescribed Ptx + VitE at our institution for the treatment of radiation necrosis.

METHODS

48 patients were included in this analysis. All patients were treated with SRS, had evidence of radiation necrosis, and had MR imaging before and after starting Ptx + VitE. The radiation oncologist’s impression of the imaging in the electronic medical record was used to score response to treatment. Official radiology reports were also reviewed.

RESULTS

43.8% of patients showed evidence of improvement, 18.8% showed no change, and 25% showed worsening radiation necrosis on imaging after starting Ptx + VitE. One patient had a mixed response; 5 patients (10.4%) had disease progression. The median time to response assessment after starting Ptx + Vit E was 3.17 months. 9 patients progressed significantly on Ptx + VitE and required Bevacizumab, hyperbaric oxygen therapy or surgery. Patients who had multiple lesions treated with SRS were less likely to show improvement on imaging after Ptx + VitE (p=0.037). 34 patients were also prescribed dexamethasone, either before (7), with (16), or after starting (11) Ptx + VitE. Use of dexamethasone was not associated with improved response to Ptx + VitE (p=0.471). 3 patients stopped Ptx + VitE due to reported side effects.

CONCLUSION

Ptx + VitE appears safe for the treatment of radiation necrosis, but randomized data are needed to assess efficacy.

ACTR-51. REMARKABLE RESPONSE OF A PATIENT WITH SECONDARY GBM TO A HISTONE DEACETYLASE INHIBITOR

Glioblastomas are highly aggressive, grade IV tumors of glial cells that arise either as de novo primary tumors or as secondary tumors, which malignantly transformed from lower grade gliomas. Secondary glioblastomas have a relatively low incidence making up 5–10% of all glioblastoma diagnoses and tend to occur in younger patients. However, these tumors are still quite aggressive with survival outcomes that do not differ substantially from primary glioblastomas. Secondary glioblastomas also often harbor mutation of isocitrate dehydrogenase (IDH) enzyme, which produces the oncometabolite 2-hydroxyglutarate (2-HG) from alpha-ketoglutarate (α-KG). The accumulation of 2-HG has several downstream effects due to its competitive inhibition of α-KG-dependent enzymes, including epigenetic modification via hypermethylation of histones. Histone deacetylase inhibitors (HDACi) are a class of molecules that inhibit histone deacetylation and have been shown to have anti-tumor effect in part due to this epigenetic modification. Thus, because of their shared targets with regard to histone modification, it is plausible that HDACi could counter the oncometabolite effects of accumulated 2-HG in IDH1 mutant tumors. Since belinostat is a pan-HDACi that has improved blood-brain barrier penetration compared to many other HDACis, we conducted a pilot study that enrolled 15 patients examining its upfront use for the treatment of glioblastomas and found that it was well tolerated. One patient in particular, likely with secondary glioblastoma harboring the typical IDH1 mutation, underwent treatment with standard chemoradiation as well as belinostat on this study. For this case, a remarkable improvement in tumor burden with very significant decrease in enhancing residual tumor and restoration of magnetic resonance spectroscopy (MRS)-detectable metabolism was noted. Furthermore, improved neurocognitition and quality-of-life were also observed in this patient during the 18-month follow-up period. Collectively, these outcomes potentially support the use of belinostat as an adjuvant therapy for patients with secondary glioblastoma that harbor a mutant IDH enzyme.

SURG-31. SPECTROSCOPIC MRI TO GUIDE BIOPSY OF LOWER GRADE GLIOMAS

Primary brain tumors are serious and life-threatening; thus, accurate histopathologic diagnosis is critical for determining the proper clinical treatment regimen. Grade II/III gliomas (lower grade gliomas, or LGGs), including astrocytomas and oligodendrogliomas, are heterogeneous and potentially contain low- and high-grade areas within the same tumor. Therefore, it is critical to target biopsies to the most aggressive portion of the tumor to avoid tumor under-grading and under-treatment. While glioblastomas are typically targeted based on contrast-enhanced MRI, LGGs have little contrast enhancement to define targets for biopsy treatment guidance. Spectroscopic MRI (sMRI) is a high-resolution MRI imaging method which allows for detection of metabolic abnormalities such as choline and NAA in the entire brain without injection of a contrast agent. We have previously evaluated the relationship between sMRI Cho/NAA ratios and tumor infiltration in surgical specimens from high grade gliomas, demonstrating a strong correlation between sMRI results and glioma infiltration. We also used the location information to correlate sMRI data to genetic and histologic biomarkers (such as 1p19q, IDH, and MGMT). An IRB-approved pilot study to obtain sMRI prior to stereotactic biopsy has been done in 20 non-enhancing LGG cases. Patients with a suspected LGG diagnosis underwent sMRI at the time of their surgical planning MRI. sMRI images were then registered to the T1w-CE and T2/FLAIR images and imported into the Stealth neuronavigation system for biopsy planning. We found that all astrocytomas (regardless of grades) showed strongly elevated Cho/NAA, while the LGGs were hardly delineated on T1w and T2/FLAIR. We found that pathology-confirmed grade II oligodendroglioma do not have choline elevation; however, NAA was mildly decreased, myo-inositol was elevated, and creatine (Cr) was mildly elevated. sMRI is a useful tool to improve biopsy targeting in LGG patients by ensuring that the highest risk regions are sampled.

Pseudo CT Estimation using Patch-based Joint Dictionary Learning

Magnetic resonance (MR) simulators have recently gained popularity; it avoids the unnecessary radiation exposure associated with Computed Tomography (CT) when used for radiation therapy planning. We propose a method for pseudo CT estimation from MR images based on joint dictionary learning. Patient-specific anatomical features were extracted from the aligned training images and adopted as signatures for each voxel. The most relevant and informative features were identified to train the joint dictionary learning-based model. The well-trained dictionary was used to predict the pseudo CT of a new patient. This prediction technique was validated with a clinical study of 12 patients with MR and CT images of the brain. The mean absolute error (MAE), peak signal-to-noise ratio (PSNR), normalized cross correlation (NCC) indexes were used to quantify the prediction accuracy. We compared our proposed method with a state-of-the-art dictionary learning method. Overall our proposed method significantly improves the prediction accuracy over the state-of-the-art dictionary learning method. We have investigated a novel joint dictionary Iearning- based approach to predict CT images from routine MRIs and demonstrated its reliability. This CT prediction technique could be a useful tool for MRI-based radiation treatment planning or attenuation correction for quantifying PET images for PET/MR imaging.

Deep learning-based image quality improvement for low-dose computed tomography simulation in radiation therapy

Low-dose computed tomography (CT) is desirable for treatment planning and simulation in radiation therapy. Multiple rescanning and replanning during the treatment course with a smaller amount of dose than a single conventional full-dose CT simulation is a crucial step in adaptive radiation therapy. We developed a machine learning-based method to improve image quality of low-dose CT for radiation therapy treatment simulation. We used a residual block concept and a self-attention strategy with a cycle-consistent adversarial network framework. A fully convolution neural network with residual blocks and attention gates (AGs) was used in the generator to enable end-to-end transformation. We have collected CT images from 30 patients treated with frameless brain stereotactic radiosurgery (SRS) for this study. These full-dose images were used to generate projection data, which were then added with noise to simulate the low-mAs scanning scenario. Low-dose CT images were reconstructed from this noise-contaminated projection data and were fed into our network along with the original full-dose CT images for training. The performance of our network was evaluated by quantitatively comparing the high-quality CT images generated by our method with the original full-dose images. When mAs is reduced to 0.5% of the original CT scan, the mean square error of the CT images obtained by our method is ∼ 1.6 % , with respect to the original full-dose images. The proposed method successfully improved the noise, contract-to-noise ratio, and nonuniformity level to be close to those of full-dose CT images and outperforms a state-of-the-art iterative reconstruction method. Dosimetric studies show that the average differences of dose-volume histogram metrics are < 0.1    Gy ( p > 0.05 ). These quantitative results strongly indicate that the denoised low-dose CT images using our method maintains image accuracy and quality and are accurate enough for dose calculation in current CT simulation of brain SRS treatment. We also demonstrate the great potential for low-dose CT in the process of simulation and treatment planning.

MRI-Based Proton Treatment Planning for Base of Skull Tumors

Purpose

To introduce a novel, deep-learning method to generate synthetic computed tomography (SCT) scans for proton treatment planning and evaluate its efficacy.

Materials and Methods

50 Patients with base of skull tumors were divided into 2 nonoverlapping training and study cohorts. Computed tomography and magnetic resonance imaging pairs for patients in the training cohort were used for training our novel 3-dimensional generative adversarial network (cycleGAN) algorithm. Upon completion of the training phase, SCT scans for patients in the study cohort were predicted based on their magnetic resonance images only. The SCT scans obtained were compared against the corresponding original planning computed tomography scans as the ground truth, and mean absolute errors (in Hounsfield units [HU]) and normalized cross-correlations were calculated. Proton plans of 45 Gy in 25 fractions with 2 beams per plan were generated for the patients based on their planning computed tomographies and recalculated on SCT scans. Dose-volume histogram endpoints were compared. A γ-index analysis along 3 cardinal planes intercepting at the isocenter was performed. Proton distal range along each beam was calculated.

Results

Image quality metrics show agreement between the generated SCT scans and the ground truth with mean absolute error values ranging from 38.65 to 65.12 HU and an average of 54.55 ± 6.81 HU and a normalized cross-correlation average of 0.96 ± 0.01. The dosimetric evaluation showed no statistically significant differences (p > 0.05) within planning target volumes for dose-volume histogram endpoints and other metrics studied, with the exception of the dose covering 95% of the target volume, with a relative difference of 0.47%. The γ-index analysis showed an average passing rate of 98% with a 10% threshold and 2% and 2-mm criteria. Proton ranges of 48 of 50 beams (96%) in this study were within clinical tolerance adopted by 4 institutions.

Conclusions

This study shows our method is capable of generating SCT scans with acceptable image quality, dose distribution agreement, and proton distal range compared with the ground truth. Our results set a promising approach for magnetic resonance imaging-based proton treatment planning.

Abstract 3721: Improved fitting of HRMAS NMR spectra for ex vivo metabolomic analysis of glioma tissue

Gliomas are aggressive brain tumors with high rates of treatment resistance and low survival rates. High-resolution magic angle spinning (HRMAS) nuclear magnetic resonance (NMR) spectroscopy can quantify metabolite concentrations in glioma tissue, but most analysis techniques require manual peak selection and integration that prevents accurate and reliable quantitation of overlapping peaks and large macromolecule baselines. Our goal was to compare the effectiveness of operator-independent LCModel analysis of glioma spectra acquired from free induction decay (FID) and Carr-Purcell-Meiboom-Gill (CPMG) pulse sequences.

HRMAS NMR spectra were acquired using FID and CPMG sequences from 14 histologically-confirmed glioma tissue samples (WHO grade II (n=5), III (n=5), and IV (n=4)) collected during surgical resection from human brain tumor patients. Metabolite concentration ratios and Cramer-Rao lower bounds (CRLBs) were estimated using LCModel. Metabolite CRLBs were compared using a paired two sample t-test. Differences in concentrations as a function of WHO grade were determined with ANOVA and Tukey-Kramer post-hoc tests. Significance was determined by p&lt;.05.

Metabolite CRLBs for lactate and myo-inositol were significantly lower for CPMG compared to FID spectra (p&lt;.05). Most metabolites could be quantified with LCModel from spectra acquired with CPMG where many metabolites acquired with the FID sequence were not detected. For example, lactate was quantifiable from 3 of the spectra acquired with FID compared to 12 spectra with CPMG. The use of the CPMG sequence reduces quantification errors by eliminating confounding baseline signals. LCModel facilitates improved separation of overlapping resonances compared to manual peak integration, supporting the use of operator-independent methods for metabolic spectral analysis.

Comparison of metabolite concentrations as a function of WHO grade revealed significant differences in lactate and glutamine plus glutamate normalized to creatine (p&lt;.05). 2-hydroxyglutarate (2-HG) was also detected in 9 out of 13 isocitrate dehydrogenase (IDH)-mutated samples using both sequences. IDH-mutated tissue should produce 2-HG; however, these results are promising as 2-HG can be difficult to quantify in 1D NMR due to spectral overlap.

In conclusion, LCModel can reliably quantify HRMAS spectra acquired with the CPMG sequence but is less reliable with the FID sequence. Increases in lactate and glutamine plus glutamate concentrations as a function of tumor grade were consistent with previous results using HRMAS for glioma metabolic analysis, and 2-HG was detected in 1D HRMAS spectra acquired with both sequences. We expect that improved spectral fitting will contribute to future NMR-based metabolomics studies in glioma.

Note: This abstract was not presented at the meeting.

Citation Format: Selin Ekici, Ren Geryak, Stewart G. Neill, Hui-Kuo Shu, Candace C. Fleischer. Improved fitting of HRMAS NMR spectra for ex vivo metabolomic analysis of glioma tissue [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 3721.

MRI-only based synthetic CT generation using dense cycle consistent generative adversarial networks

PURPOSE

Automated synthetic computed tomography (sCT) generation based on magnetic resonance imaging (MRI) images would allow for MRI-only based treatment planning in radiation therapy, eliminating the need for CT simulation and simplifying the patient treatment workflow. In this work, the authors propose a novel method for generation of sCT based on dense cycle-consistent generative adversarial networks (cycle GAN), a deep-learning based model that trains two transformation mappings (MRI to CT and CT to MRI) simultaneously.

METHODS AND MATERIALS

The cycle GAN-based model was developed to generate sCT images in a patch-based framework. Cycle GAN was applied to this problem because it includes an inverse transformation from CT to MRI, which helps constrain the model to learn a one-to-one mapping. Dense block-based networks were used to construct generator of cycle GAN. The network weights and variables were optimized via a gradient difference (GD) loss and a novel distance loss metric between sCT and original CT.

RESULTS

Leave-one-out cross-validation was performed to validate the proposed model. The mean absolute error (MAE), peak signal-to-noise ratio (PSNR), and normalized cross correlation (NCC) indexes were used to quantify the differences between the sCT and original planning CT images. For the proposed method, the mean MAE between sCT and CT were 55.7 Hounsfield units (HU) for 24 brain cancer patients and 50.8 HU for 20 prostate cancer patients. The mean PSNR and NCC were 26.6 dB and 0.963 in the brain cases, and 24.5 dB and 0.929 in the pelvis.

CONCLUSION

We developed and validated a novel learning-based approach to generate CT images from routine MRIs based on dense cycle GAN model to effectively capture the relationship between the CT and MRIs. The proposed method can generate robust, high-quality sCT in minutes. The proposed method offers strong potential for supporting near real-time MRI-only treatment planning in the brain and pelvis.

Learning-based automatic segmentation of arteriovenous malformations on contrast CT images in brain stereotactic radiosurgery

PURPOSE

Stereotactic radiosurgery (SRS) is widely used to obliterate arteriovenous malformations (AVMs). Its performance relies on the accuracy of delineating the target AVM. Manual segmentation during a framed SRS procedure is time consuming and subject to inter- and intraobserver variation. To address these drawbacks, we proposed a deep learning-based method to automatically segment AVMs on CT simulation image sets.

METHODS

We developed a deep learning-based method using a deeply supervised three-dimensional (3D) V-Net with a compound loss function. A 3D supervision mechanism was integrated into a residual network, V-Net, to deal with the optimization difficulties when training deep networks with limited training data. The proposed compound loss function including logistic and Dice losses encouraged similarity and penalized discrepancy simultaneously between prediction and training dataset; this was utilized to supervise the 3D V-Net at different stages. To evaluate the accuracy of segmentation, we retrospectively investigated 80 AVM patients who had CT simulation and digital subtraction angiography (DSA) acquired prior to treatment. The AVM target volume was segmented by our proposed method. They were compared with clinical contours approved by physicians with regard to Dice overlapping, difference in volume and centroid, and dose coverage changes on original plan.

RESULTS

Contours created by the proposed method demonstrated very good visual agreement to the ground truth contours. The mean Dice similarity coefficient (DSC), sensitivity and specificity of the contours delineated by our method were >0.85 among all patients. The mean centroid distance between our results and ground truth was 0.675 ± 0.401 mm, and was not significantly different in any of the three orthogonal directions. The correlation coefficient between ground truth and AVM volume resulting from the proposed method was 0.992 with statistical significance. The mean volume difference among all patients was 0.076 ± 0.728 cc; there was no statistically significant difference. The average differences in dose metrics were all less than 0.2 Gy, with standard deviation less than 1 Gy. No statistically significant differences were observed in any of the dose metrics.

CONCLUSION

We developed a novel, deeply supervised, deep learning-based approach to automatically segment the AVM volume on CT images. We demonstrated its clinical feasibility by validating the shape and positional accuracy, and dose coverage of the automatic volume. These results demonstrate the potential of a learning-based segmentation method for delineating AVMs in the clinical setting.

MRI-based synthetic CT generation using semantic random forest with iterative refinement

Target delineation for radiation therapy treatment planning often benefits from magnetic resonance imaging (MRI) in addition to x-ray computed tomography (CT) due to MRI's superior soft tissue contrast. MRI-based treatment planning could reduce systematic MR-CT co-registration errors, medical cost, radiation exposure, and simplify clinical workflow. However, MRI-only based treatment planning is not widely used to date because treatment-planning systems rely on the electron density information provided by CTs to calculate dose. Additionally, air and bone regions are difficult to separategiven their similar intensities in MR imaging. The purpose of this work is to develop a learning-based method to generate patient-specific synthetic CT (sCT) from a routine anatomical MRI for use in MRI-only radiotherapy treatment planning. An auto-context model with patch-based anatomical features was integrated into a classification random forest to generate and improve semantic information. The semantic information along with anatomical features was then used to train a series of regression random forests based on the auto-context model. After training, the sCT of a new MRI can be generated by feeding anatomical features extracted from the MRI into the well-trained classification and regression random forests. The proposed algorithm was evaluated using 14 patient datasets withT1-weighted MR and corresponding CT images of the brain. The mean absolute error (MAE), peak signal-to-noise ratio (PSNR), and normalized cross correlation (NCC) were 57.45  ±  8.45 HU, 28.33  ±  1.68 dB, and 0.97  ±  0.01. We also compared the difference between dose maps calculated on the sCT and those on the original CT, using the same plan parameters. The average DVH differences among all patients are less than 0.2 Gy for PTVs, and less than 0.02 Gy for OARs. The sCT generation by the proposed method allows for dose calculation based MR imaging alone, and may be a useful tool for MRI-based radiation treatment planning.

Dosimetric study on learning-based cone-beam CT correction in adaptive radiation therapy

INTRODUCTION

Cone-beam CT (CBCT) image quality is important for its quantitative analysis in adaptive radiation therapy. However, due to severe artifacts, the CBCTs are primarily used for verifying patient setup only so far. We have developed a learning-based image quality improvement method which could provide CBCTs with image quality comparable to planning CTs (pCTs). The accuracy of dose calculations based on these CBCTs is unknown. In this study, we aim to investigate the dosimetric accuracy of our corrected CBCT (CCBCT) in brain stereotactic radiosurgery (SRS) and pelvic radiotherapy.

MATERIALS AND METHODS

We retrospectively investigated a total of 32 treatment plans from 22 patients, each of whom with both original treatment pCTs and CBCTs acquired during treatment setup. The CCBCT and original CBCT (OCBCT) were registered to the pCT for generating CCBCT-based and OCBCT-based treatment plans. The original pCT-based plans served as ground truth. Clinically-relevant dose volume histogram (DVH) metrics were extracted from the ground truth, OCBCT-based and CCBCT-based plans for comparison. Gamma analysis was also performed to compare the absorbed dose distributions between the pCT-based and OCBCT/CCBCT-based plans of each patient.

RESULTS

CCBCTs demonstrated better image contrast and more accurate HU ranges when compared side-by-side with OCBCTs. For pelvic radiotherapy plans, the mean dose error in DVH metrics for planning target volume (PTV), bladder and rectum was significantly reduced, from 1% to 0.3%, after CBCT correction. The gamma analysis showed the average pass rate increased from 94.5% before correction to 99.0% after correction. For brain SRS treatment plans, both original and corrected CBCT images were accurate enough for dose calculation, though CCBCT featured higher image quality.

CONCLUSION

CCBCTs can provide a level of dose accuracy comparable to traditional pCTs for brain and prostate radiotherapy planning and the correction method proposed here can be useful in CBCT-guided adaptive radiotherapy.

The role of erlotinib and the Optune device in a patient with an epidermal growth factor receptor viii amplified glioblastoma

The standard treatment for patients diagnosed with glioblastoma is surgical resection of tumor followed by high dose radiation and chemotherapy with temozolomide. For patients who experience allergic reactions to temozolomide despite desensitization protocols, alternative therapies must be considered. In this report, we present such a patient who then received treatment with an epidermal growth factor receptor inhibitor, erlotinib, concurrent with a tumor-treating field device, Optune. Through this combination of a targeted molecular therapy and the Optune device, the patient has been able to achieve stable disease 9 months after completing radiation.

MRI-based pseudo CT synthesis using anatomical signature and alternating random forest with iterative refinement model

We develop a learning-based method to generate patient-specific pseudo computed tomography (CT) from routinely acquired magnetic resonance imaging (MRI) for potential MRI-based radiotherapy treatment planning. The proposed pseudo CT (PCT) synthesis method consists of a training stage and a synthesizing stage. During the training stage, patch-based features are extracted from MRIs. Using a feature selection, the most informative features are identified as an anatomical signature to train a sequence of alternating random forests based on an iterative refinement model. During the synthesizing stage, we feed the anatomical signatures extracted from an MRI into the sequence of well-trained forests for a PCT synthesis. Our PCT was compared with original CT (ground truth) to quantitatively assess the synthesis accuracy. The mean absolute error, peak signal-to-noise ratio, and normalized cross-correlation indices were 60.87 ± 15.10    HU , 24.63 ± 1.73    dB , and 0.954 ± 0.013 for 14 patients' brain data and 29.86 ± 10.4    HU , 34.18 ± 3.31    dB , and 0.980 ± 0.025 for 12 patients' pelvic data, respectively. We have investigated a learning-based approach to synthesize CTs from routine MRIs and demonstrated its feasibility and reliability. The proposed PCT synthesis technique can be a useful tool for MRI-based radiation treatment planning.

Magnetic resonance imaging-based pseudo computed tomography using anatomic signature and joint dictionary learning

Magnetic resonance imaging (MRI) provides a number of advantages over computed tomography (CT) for radiation therapy treatment planning; however, MRI lacks the key electron density information necessary for accurate dose calculation. We propose a dictionary-learning-based method to derive electron density information from MRIs. Specifically, we first partition a given MR image into a set of patches, for which we used a joint dictionary learning method to directly predict a CT patch as a structured output. Then a feature selection method is used to ensure prediction robustness. Finally, we combine all the predicted CT patches to obtain the final prediction for the given MR image. This prediction technique was validated for a clinical application using 14 patients with brain MR and CT images. The peak signal-to-noise ratio (PSNR), mean absolute error (MAE), normalized cross-correlation (NCC) indices and similarity index (SI) for air, soft-tissue and bone region were used to quantify the prediction accuracy. The mean ± std of PSNR, MAE, and NCC were: 22.4±1.9  dB , 82.6±26.1 HU, and 0.91±0.03 for the 14 patients. The SIs for air, soft-tissue, and bone regions are 0.98±0.01 , 0.88±0.03 , and 0.69±0.08 . These indices demonstrate the CT prediction accuracy of the proposed learning-based method. This CT image prediction technique could be used as a tool for MRI-based radiation treatment planning, or for PET attenuation correction in a PET/MRI scanner.

MRI-based treatment planning for brain stereotactic radiosurgery: Dosimetric validation of a learning-based pseudo-CT generation method

Magnetic resonance imaging (MRI)-only radiotherapy treatment planning is attractive since MRI provides superior soft tissue contrast without ionizing radiation compared with computed tomography (CT). However, it requires the generation of pseudo CT from MRI images for patient setup and dose calculation. Our machine-learning-based method to generate pseudo CT images has been shown to provide pseudo CT images with excellent image quality, while its dose calculation accuracy remains an open question. In this study, we aim to investigate the accuracy of dose calculation in brain frameless stereotactic radiosurgery (SRS) using pseudo CT images which are generated from MRI images using the machine learning-based method developed by our group. We retrospectively investigated a total of 19 treatment plans from 14 patients, each of whom has CT simulation and MRI images acquired during pretreatment. The dose distributions of the same treatment plans were calculated on original CT simulation images as ground truth, as well as on pseudo CT images generated from MRI images. Clinically-relevant DVH metrics and gamma analysis were extracted from both ground truth and pseudo CT results for comparison and evaluation. The side-by-side comparisons on image quality and dose distributions demonstrated very good agreement of image contrast and calculated dose between pseudo CT and original CT. The average differences in Dose-volume histogram (DVH) metrics for Planning target volume (PTVs) were less than 0.6%, and no differences in those for organs at risk at a significance level of 0.05. The average pass rate of gamma analysis was 99%. These quantitative results strongly indicate that the pseudo CT images created from MRI images using our proposed machine learning method are accurate enough to replace current CT simulation images for dose calculation in brain SRS treatment. This study also demonstrates the great potential for MRI to completely replace CT scans in the process of simulation and treatment planning.

Progesterone improves neurocognitive outcomes following therapeutic cranial irradiation in mice

Despite improved therapeutic methods, CNS toxicity resulting from cancer treatment remains a major cause of post-treatment morbidity. More than half of adult patients with cranial irradiation for brain cancer develop neurobehavioral/cognitive deficits that severely impact quality of life. We examined the neuroprotective effects of the neurosteroid progesterone (PROG) against ionizing radiation (IR)-induced neurobehavioral/cognitive deficits in mice. Male C57/BL mice were exposed to one of two fractionated dose regimens of IR (3Gy×3 or 3Gy×5). PROG (16mg/kg; 0.16mg/g) was given as a pre-, concurrent or post-IR treatment for 14days. Mice were tested for short- and long-term effects of IR and PROG on neurobehavioral/cognitive function on days 10 and 30 after IR treatment. We evaluated both hippocampus-dependent and -independent memory functions. Locomotor activity, elevated plus maze, novel object recognition and Morris water maze tests revealed behavioral deficits following IR. PROG treatment produced improvement in behavioral performance at both time points in the mice given IR. Western blot analysis of hippocampal and cortical tissue showed that IR at both doses induced astrocytic activation (glial fibrillary acidic protein), reactive macrophages/microglia (CD68) and apoptosis (cleaved caspase-3) and PROG treatment inhibited these markers of brain injury. There was no significant difference in the degree of deficit in any test between the two dose regimens of IR at either time point. These findings could be important in the context of patients with brain tumors who may undergo radiotherapy and eventually develop cognitive deficits.

Pseudo CT Estimation from MRI Using Patch-based Random Forest

Recently, MR simulators gain popularity because of unnecessary radiation exposure of CT simulators being used in radiation therapy planning. We propose a method for pseudo CT estimation from MR images based on a patch-based random forest. Patient-specific anatomical features are extracted from the aligned training images and adopted as signatures for each voxel. The most robust and informative features are identified using feature selection to train the random forest. The well-trained random forest is used to predict the pseudo CT of a new patient. This prediction technique was tested with human brain images and the prediction accuracy was assessed using the original CT images. Peak signal-to-noise ratio (PSNR) and feature similarity (FSIM) indexes were used to quantify the differences between the pseudo and original CT images. The experimental results showed the proposed method could accurately generate pseudo CT images from MR images. In summary, we have developed a new pseudo CT prediction method based on patch-based random forest, demonstrated its clinical feasibility, and validated its prediction accuracy. This pseudo CT prediction technique could be a useful tool for MRI-based radiation treatment planning and attenuation correction in a PET/MRI scanner.

Simulating the Effect of Spectroscopic MRI as a Metric for Radiation Therapy Planning in Patients with Glioblastoma

Due to glioblastoma's infiltrative nature, an optimal radiation therapy (RT) plan requires targeting infiltration not identified by anatomical magnetic resonance imaging (MRI). Here, high-resolution, whole-brain spectroscopic MRI (sMRI) is used to describe tumor infiltration alongside anatomical MRI and simulate the degree to which it modifies RT target planning. In 11 patients with glioblastoma, data from preRT sMRI scans were processed to give high-resolution, whole-brain metabolite maps normalized by contralateral white matter. Maps depicting choline to N-Acetylaspartate (Cho/NAA) ratios were registered to contrast-enhanced T1-weighted RT planning MRI for each patient. Volumes depicting metabolic abnormalities (1.5-, 1.75-, and 2.0-fold increases in Cho/NAA ratios) were compared with conventional target volumes and contrast-enhancing tumor at recurrence. sMRI-modified RT plans were generated to evaluate target volume coverage and organ-at-risk dose constraints. Conventional clinical target volumes and Cho/NAA abnormalities identified significantly different regions of microscopic infiltration with substantial Cho/NAA abnormalities falling outside of the conventional 60 Gy isodose line (41.1, 22.2, and 12.7 cm³, respectively). Clinical target volumes using Cho/NAA thresholds exhibited significantly higher coverage of contrast enhancement at recurrence on average (92.4%, 90.5%, and 88.6%, respectively) than conventional plans (82.5%). sMRI-based plans targeting tumor infiltration met planning objectives in all cases with no significant change in target coverage. In 2 cases, the sMRI-modified plan exhibited better coverage of contrast-enhancing tumor at recurrence than the original plan. Integration of the high-resolution, whole-brain sMRI into RT planning is feasible, resulting in RT target volumes that can effectively target tumor infiltration while adhering to conventional constraints.

Automated population-based planning for whole brain radiation therapy

Treatment planning for whole‐brain radiation treatment is technically a simple process, but in practice it takes valuable clinical time of repetitive and tedious tasks. This report presents a method that automatically segments the relevant target and normal tissues, and creates a treatment plan in only a few minutes after patient simulation. Segmentation of target and critical structures is performed automatically through morphological operations on the soft tissue and was validated by comparing with manual clinical segmentation using the Dice coefficient and Hausdorff distance. The treatment plan is generated by searching a database of previous cases for patients with similar anatomy. In this search, each database case is ranked in terms of similarity using a customized metric designed for sensitivity by including only geometrical changes that affect the dose distribution. The database case with the best match is automatically modified to replace relevant patient info and isocenter position while maintaining original beam and MLC settings. Fifteen patients with marginally acceptable treatment plans were used to validate the method. In each of these cases the anatomy was accurately segmented, but the beams and MLC settings led to a suboptimal treatment plan by either underdosing the brain or excessively irradiating critical normal tissues. For each case, the anatomy was automatically segmented with the proposed method, and the automated and manual segmentations were then compared. The mean Dice coefficient was 0.97, with a standard deviation of 0.008 for the brain, 0.85±0.009 for the eyes, and 0.67±0.11 for the lens. The mean Euclidian distance was 0.13±0.13 mm for the brain, 0.27±0.31 for the eye, and 2.34±7.23 for the lens. Each case was then subsequently matched against a database of 70 validated treatment plans and the best matching plan (termed autoplanned), was compared retrospectively with the clinical plans in terms of brain coverage and maximum doses to critical structures. Maximum doses were reduced by a maximum of 8.37 Gy for the left eye (mean 2.08), 11.67 for the right eye (1.90) and, respectively, 25.44 (5.59) for the left lens and 24.40 (4.85) for the right lens. Time to generate the autoplan, including the segmentation, was 3−4 min. Automated database‐ based matching is an alternative to classical treatment planning that improves quality while providing a cost‐effective solution to planning through modifying previous validated plans to match a current patient's anatomy.

PACS number: 87.55.D, 87.55.tg, 87.57.nm