skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: SU-C-BRA-06: Automatic Brain Tumor Segmentation for Stereotactic Radiosurgery Applications

Abstract

Purpose: Stereotactic radiosurgery (SRS), which delivers a potent dose of highly conformal radiation to the target in a single fraction, requires accurate tumor delineation for treatment planning. We present an automatic segmentation strategy, that synergizes intensity histogram thresholding, super-voxel clustering, and level-set based contour evolving methods to efficiently and accurately delineate SRS brain tumors on contrast-enhance T1-weighted (T1c) Magnetic Resonance Images (MRI). Methods: The developed auto-segmentation strategy consists of three major steps. Firstly, tumor sites are localized through 2D slice intensity histogram scanning. Then, super voxels are obtained through clustering the corresponding voxels in 3D with reference to the similarity metrics composited from spatial distance and intensity difference. The combination of the above two could generate the initial contour surface. Finally, a localized region active contour model is utilized to evolve the surface to achieve the accurate delineation of the tumors. The developed method was evaluated on numerical phantom data, synthetic BRATS (Multimodal Brain Tumor Image Segmentation challenge) data, and clinical patients’ data. The auto-segmentation results were quantitatively evaluated by comparing to ground truths with both volume and surface similarity metrics. Results: DICE coefficient (DC) was performed as a quantitative metric to evaluate the auto-segmentation in the numerical phantom withmore » 8 tumors. DCs are 0.999±0.001 without noise, 0.969±0.065 with Rician noise and 0.976±0.038 with Gaussian noise. DC, NMI (Normalized Mutual Information), SSIM (Structural Similarity) and Hausdorff distance (HD) were calculated as the metrics for the BRATS and patients’ data. Assessment of BRATS data across 25 tumor segmentation yield DC 0.886±0.078, NMI 0.817±0.108, SSIM 0.997±0.002, and HD 6.483±4.079mm. Evaluation on 8 patients with total 14 tumor sites yield DC 0.872±0.070, NMI 0.824±0.078, SSIM 0.999±0.001, and HD 5.926±6.141mm. Conclusion: The developed automatic segmentation strategy, which yields accurate brain tumor delineation in evaluation cases, is promising for its application in SRS treatment planning.« less

Authors:
; ; ; ; ; ;  [1]
  1. UT Southwestern Medical Center, Dallas, TX (United States)
Publication Date:
OSTI Identifier:
22624328
Resource Type:
Journal Article
Resource Relation:
Journal Name: Medical Physics; Journal Volume: 43; Journal Issue: 6; Other Information: (c) 2016 American Association of Physicists in Medicine; Country of input: International Atomic Energy Agency (IAEA)
Country of Publication:
United States
Language:
English
Subject:
60 APPLIED LIFE SCIENCES; 61 RADIATION PROTECTION AND DOSIMETRY; BRAIN; IMAGES; NEOPLASMS; NMR IMAGING; NOISE; PATIENTS; PHANTOMS; PLANNING; RADIOTHERAPY; SURGERY

Citation Formats

Liu, Y, Stojadinovic, S, Jiang, S, Timmerman, R, Abdulrahman, R, Nedzi, L, and Gu, X. SU-C-BRA-06: Automatic Brain Tumor Segmentation for Stereotactic Radiosurgery Applications. United States: N. p., 2016. Web. doi:10.1118/1.4955567.
Liu, Y, Stojadinovic, S, Jiang, S, Timmerman, R, Abdulrahman, R, Nedzi, L, & Gu, X. SU-C-BRA-06: Automatic Brain Tumor Segmentation for Stereotactic Radiosurgery Applications. United States. doi:10.1118/1.4955567.
Liu, Y, Stojadinovic, S, Jiang, S, Timmerman, R, Abdulrahman, R, Nedzi, L, and Gu, X. 2016. "SU-C-BRA-06: Automatic Brain Tumor Segmentation for Stereotactic Radiosurgery Applications". United States. doi:10.1118/1.4955567.
@article{osti_22624328,
title = {SU-C-BRA-06: Automatic Brain Tumor Segmentation for Stereotactic Radiosurgery Applications},
author = {Liu, Y and Stojadinovic, S and Jiang, S and Timmerman, R and Abdulrahman, R and Nedzi, L and Gu, X},
abstractNote = {Purpose: Stereotactic radiosurgery (SRS), which delivers a potent dose of highly conformal radiation to the target in a single fraction, requires accurate tumor delineation for treatment planning. We present an automatic segmentation strategy, that synergizes intensity histogram thresholding, super-voxel clustering, and level-set based contour evolving methods to efficiently and accurately delineate SRS brain tumors on contrast-enhance T1-weighted (T1c) Magnetic Resonance Images (MRI). Methods: The developed auto-segmentation strategy consists of three major steps. Firstly, tumor sites are localized through 2D slice intensity histogram scanning. Then, super voxels are obtained through clustering the corresponding voxels in 3D with reference to the similarity metrics composited from spatial distance and intensity difference. The combination of the above two could generate the initial contour surface. Finally, a localized region active contour model is utilized to evolve the surface to achieve the accurate delineation of the tumors. The developed method was evaluated on numerical phantom data, synthetic BRATS (Multimodal Brain Tumor Image Segmentation challenge) data, and clinical patients’ data. The auto-segmentation results were quantitatively evaluated by comparing to ground truths with both volume and surface similarity metrics. Results: DICE coefficient (DC) was performed as a quantitative metric to evaluate the auto-segmentation in the numerical phantom with 8 tumors. DCs are 0.999±0.001 without noise, 0.969±0.065 with Rician noise and 0.976±0.038 with Gaussian noise. DC, NMI (Normalized Mutual Information), SSIM (Structural Similarity) and Hausdorff distance (HD) were calculated as the metrics for the BRATS and patients’ data. Assessment of BRATS data across 25 tumor segmentation yield DC 0.886±0.078, NMI 0.817±0.108, SSIM 0.997±0.002, and HD 6.483±4.079mm. Evaluation on 8 patients with total 14 tumor sites yield DC 0.872±0.070, NMI 0.824±0.078, SSIM 0.999±0.001, and HD 5.926±6.141mm. Conclusion: The developed automatic segmentation strategy, which yields accurate brain tumor delineation in evaluation cases, is promising for its application in SRS treatment planning.},
doi = {10.1118/1.4955567},
journal = {Medical Physics},
number = 6,
volume = 43,
place = {United States},
year = 2016,
month = 6
}
  • Purpose: Radiation therapy following resection of a brain metastasis increases the probability of disease control at the surgical site. We analyzed our experience with postoperative stereotactic radiosurgery (SRS) as an alternative to whole-brain radiotherapy (WBRT), with an emphasis on identifying factors that might predict intracranial disease control and overall survival (OS). Methods and Materials: We retrospectively reviewed all patients through December 2008, who, after surgical resection, underwent SRS to the tumor bed, deferring WBRT. Multiple factors were analyzed for time to intracranial recurrence (ICR), whether local recurrence (LR) at the surgical bed or “distant” recurrence (DR) in the brain, formore » time to WBRT, and for OS. Results: A total of 49 lesions in 47 patients were treated with postoperative SRS. With median follow-up of 9.3 months (range, 1.1-61.4 months), local control rates at the resection cavity were 85.5% at 1 year and 66.9% at 2 years. OS rates at 1 and 2 years were 52.5% and 31.7%, respectively. On univariate analysis (preoperative) tumors larger than 3.0 cm exhibited a significantly shorter time to LR. At a cutoff of 2.0 cm, larger tumors resulted in significantly shorter times not only for LR but also for DR, ICR, and salvage WBRT. While multivariate Cox regressions showed preoperative size to be significant for times to DR, ICR, and WBRT, in similar multivariate analysis for OS, only the graded prognostic assessment proved to be significant. However, the number of intracranial metastases at presentation was not significantly associated with OS nor with other outcome variables. Conclusions: Larger tumor size was associated with shorter time to recurrence and with shorter time to salvage WBRT; however, larger tumors were not associated with decrements in OS, suggesting successful salvage. SRS to the tumor bed without WBRT is an effective treatment for resected brain metastases, achieving local control particularly for tumors up to 3.0 cm diameter.« less
  • Purpose: Given the neurocognitive toxicity associated with whole-brain irradiation (WBRT), approaches to defer or avoid WBRT after surgical resection of brain metastases are desirable. Our initial experience with stereotactic radiosurgery (SRS) targeting the resection cavity showed promising results. We examined the outcomes of postoperative resection cavity SRS to determine the effect of adding a 2-mm margin around the resection cavity on local failure (LF) and toxicity. Patients and Methods: We retrospectively evaluated 120 cavities in 112 patients treated from 1998-2009. Factors associated with LF and distant brain failure (DF) were analyzed using competing risks analysis, with death as a competingmore » risk. The overall survival (OS) rate was calculated by the Kaplan-Meier product-limit method; variables associated with OS were evaluated using the Cox proportional hazards and log rank tests. Results: The 12-month cumulative incidence rates of LF and DF, with death as a competing risk, were 9.5% and 54%, respectively. On univariate analysis, expansion of the cavity with a 2-mm margin was associated with decreased LF; the 12-month cumulative incidence rates of LF with and without margin were 3% and 16%, respectively (P=.042). The 12-month toxicity rates with and without margin were 3% and 8%, respectively (P=.27). On multivariate analysis, melanoma histology (P=.038) and number of brain metastases (P=.0097) were associated with higher DF. The median OS time was 17 months (range, 2-114 months), with a 12-month OS rate of 62%. Overall, WBRT was avoided in 72% of the patients. Conclusion: Adjuvant SRS targeting the resection cavity of brain metastases results in excellent local control and allows WBRT to be avoided in a majority of patients. A 2-mm margin around the resection cavity improved local control without increasing toxicity compared with our prior technique with no margin.« less
  • Purpose: To evaluate local control rates and predictors of individual tumor local control for brain metastases from non-small cell lung cancer (NSCLC) treated with stereotactic radiosurgery (SRS). Methods and Materials: Between June 1998 and May 2011, 401 brain metastases in 228 patients were treated with Gamma Knife single-fraction SRS. Local failure was defined as an increase in lesion size after SRS. Local control was estimated using the Kaplan-Meier method. The Cox proportional hazards model was used for univariate and multivariate analysis. Receiver operating characteristic analysis was used to identify an optimal cutpoint for conformality index relative to local control. Amore » P value <.05 was considered statistically significant. Results: Median age was 60 years (range, 27-84 years). There were 66 cerebellar metastases (16%) and 335 supratentorial metastases (84%). The median prescription dose was 20 Gy (range, 14-24 Gy). Median overall survival from time of SRS was 12.1 months. The estimated local control at 12 months was 74%. On multivariate analysis, cerebellar location (hazard ratio [HR] 1.94, P=.009), larger tumor volume (HR 1.09, P<.001), and lower conformality (HR 0.700, P=.044) were significant independent predictors of local failure. Conformality index cutpoints of 1.4-1.9 were predictive of local control, whereas a cutpoint of 1.75 was the most predictive (P=.001). The adjusted Kaplan-Meier 1-year local control for conformality index ≥1.75 was 84% versus 69% for conformality index <1.75, controlling for tumor volume and location. The 1-year adjusted local control for cerebellar lesions was 60%, compared with 77% for supratentorial lesions, controlling for tumor volume and conformality index. Conclusions: Cerebellar tumor location, lower conformality index, and larger tumor volume were significant independent predictors of local failure after SRS for brain metastases from NSCLC. These results warrant further investigation in a prospective setting.« less
  • Background: The primary dose-limiting toxicity of stereotactic radiosurgery (SRS) is radiation necrosis (RN), which occurs after approximately 5% to 10% of treatments. This adverse event may worsen neurologic deficits, increase the frequency and cost of imaging, and necessitate prolonged treatment with steroids or antiangiogenic agents. Previous investigations have primarily identified lesion size and dosimetric constraints as risk factors for RN in small populations. We hypothesized that disease histology, receptor status, and mutational status are associated with RN. Methods and Materials: All patients presenting with brain metastasis between 1997 and 2015 who underwent SRS and subsequent radiographic follow-up at a singlemore » tertiary-care institution were eligible for inclusion. The primary outcome was the cumulative incidence of radiographic RN. Multivariate competing risks regression was used to identify biological risk factors for RN. Results: 1939 patients (5747 lesions) were eligible for inclusion; 285 patients (15%) experienced radiographic RN after the treatment of 427 (7%) lesions. After SRS, the median time to RN was 7.6 months. After multivariate analysis, graded prognostic assessment, renal pathology, lesion diameter, and the heterogeneity index remained independently predictive of RN in the pooled cohort. In subset analyses of individual pathologies, HER2-amplified status (hazard ratio [HR] 2.05, P=.02), BRAF V600+ mutational status (HR 0.33, P=.04), lung adenocarcinoma histology (HR 1.89, P=.04), and ALK rearrangement (HR 6.36, P<.01) were also associated with RN. Conclusions: In the present investigation constituting the largest series of RN, several novel risk factors were identified, including renal histology, lung adenocarcinoma histology, HER2 amplification, and ALK/BRAF mutational status. These risk factors may be used to guide clinical trial design incorporating biological risk stratification or dose escalation. Future studies determining the optimal timing of targeted therapies are warranted to further define the risk of RN.« less
  • Purpose: To determine how the omission of whole brain radiotherapy (WBRT) affects the neurocognitive function of patients with one to four brain metastases who have been treated with stereotactic radiosurgery (SRS). Methods and Materials: In a prospective randomized trial between WBRT+SRS and SRS alone for patients with one to four brain metastases, we assessed the neurocognitive function using the Mini-Mental State Examination (MMSE). Of the 132 enrolled patients, MMSE scores were available for 110. Results: In the baseline MMSE analyses, statistically significant differences were observed for total tumor volume, extent of tumor edema, age, and Karnofsky performance status. Of themore » 92 patients who underwent the follow-up MMSE, 39 had a baseline MMSE score of {<=}27 (17 in the WBRT+SRS group and 22 in the SRS-alone group). Improvements of {>=}3 points in the MMSEs of 9 WBRT+SRS patients and 11 SRS-alone patients (p = 0.85) were observed. Of the 82 patients with a baseline MMSE score of {>=}27 or whose baseline MMSE score was {<=}26 but had improved to {>=}27 after the initial brain treatment, the 12-, 24-, and 36-month actuarial free rate of the 3-point drop in the MMSE was 76.1%, 68.5%, and 14.7% in the WBRT+SRS group and 59.3%, 51.9%, and 51.9% in the SRS-alone group, respectively. The average duration until deterioration was 16.5 months in the WBRT+SRS group and 7.6 months in the SRS-alone group (p = 0.05). Conclusion: The results of the present study have revealed that, for most brain metastatic patients, control of the brain tumor is the most important factor for stabilizing neurocognitive function. However, the long-term adverse effects of WBRT on neurocognitive function might not be negligible.« less