Deep learning–based detection and segmentation-assisted management of brain metastases

Xue, Jie; Bao, Wang; Yang, Ming; Liu, Xuejun; Jiang, Zekun; Wang, Chengwei; Liu, Xiyu; Chen, Ligang; Qu, Jianhua; Xu, Shangchen; Tang, Xuqun; Mao, Ying; Liu, Yingchao; Li, Dengwang

doi:10.1093/neuonc/noz234

Cited by 90 publications

(67 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A very recent study by Xue et al [25] is not listed in the comparison in Table 3. They also used data gained from SRS and claimed an accuracy of 100% on their data.…”

Section: Discussionmentioning

confidence: 99%

“…The smallest lesion in their training set had a size of 0.07 ml, while 26% of lesions in our test set were in fact smaller. Additionally Xue et al [25] calculated sensitivity and specificity of their model per pixel and nor per lesion, which heavily favors larger lesions. The used dataset also has a typical distribution of primary cancer types, which we used to show the robust performance of the DCNN for different primary tumors.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Deep convolutional neural networks for automated segmentation of brain metastases trained on clinical data

et al. 2020

View full text Add to dashboard Cite

Introduction: Deep learning-based algorithms have demonstrated enormous performance in segmentation of medical images. We collected a dataset of multiparametric MRI and contour data acquired for use in radiosurgery, to evaluate the performance of deep convolutional neural networks (DCNN) in automatic segmentation of brain metastases (BM). Methods: A conventional U-Net (cU-Net), a modified U-Net (moU-Net) and a U-Net trained only on BM smaller than 0.4 ml (sU-Net) were implemented. Performance was assessed on a separate test set employing sensitivity, specificity, average false positive rate (AFPR), the dice similarity coefficient (DSC), Bland-Altman analysis and the concordance correlation coefficient (CCC). Results: A dataset of 509 patients (1223 BM) was split into a training set (469 pts) and a test set (40 pts). A combination of all trained networks was the most sensitive (0.82) while maintaining a specificity 0.83. The same model achieved a sensitivity of 0.97 and a specificity of 0.94 when considering only lesions larger than 0.06 ml (75% of all lesions). Type of primary cancer had no significant influence on the mean DSC per lesion (p = 0.60). Agreement between manually and automatically assessed tumor volumes as quantified by a CCC of 0.87 (95% CI, 0.77-0.93), was excellent. Conclusion: Using a dataset which properly captured the variation in imaging appearance observed in clinical practice, we were able to conclude that DCNNs reach clinically relevant performance for most lesions. Clinical applicability is currently limited by the size of the target lesion. Further studies should address if small targets are accurately represented in the test data.

show abstract

“…A very recent study by Xue et al [25] is not listed in the comparison in Table 3. They also used data gained from SRS and claimed an accuracy of 100% on their data.…”

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Deep convolutional neural networks for automated segmentation of brain metastases trained on clinical data

et al. 2020

View full text Add to dashboard Cite

show abstract

“…still needed to scan physical films for subsequent slice-by-slice segmentation ( 6 ), semiautomatic techniques have emerged that are much more time-efficient and reduce inter- and intra-observer variability ( 18 – 20 ). Recently, the advent of artificial neural networks has even enabled accurate fully automatic segmentation of brain tumors ( 32 , 33 ). Moreover, radiomic analyses also necessarily require tumor segmentations and are increasingly incorporated into clinical trials ( 34 ).…”

Section: Discussionmentioning

confidence: 99%

Volumetric Regression in Brain Metastases After Stereotactic Radiotherapy: Time Course, Predictors, and Significance

et al. 2021

View full text Add to dashboard Cite

BackgroundThere is insufficient understanding of the natural course of volumetric regression in brain metastases after stereotactic radiotherapy (SRT) and optimal volumetric criteria for the assessment of response and progression in radiotherapy clinical trials for brain metastases are currently unknown.MethodsVolumetric analysis via whole-tumor segmentation in contrast-enhanced 1 mm³-isotropic T1-Mprage sequences before SRT and during follow-up. A total of 3,145 MRI studies of 419 brain metastases from 189 patients were segmented. Progression was defined using a volumetric extension of the RANO-BM criteria. A subset of 205 metastases without progression/radionecrosis during their entire follow-up of at least 3 months was used to study the natural course of volumetric regression after SRT. Predictors for volumetric regression were investigated. A second subset of 179 metastases was used to investigate the prognostic significance of volumetric response at 3 months (defined as ≥20% and ≥65% volume reduction, respectively) for subsequent local control.ResultsMedian relative metastasis volume post-SRT was 66.9% at 6 weeks, 38.6% at 3 months, 17.7% at 6 months, 2.7% at 12 months and 0.0% at 24 months. Radioresistant histology and FSRT vs. SRS were associated with reduced tumor regression for all time points. In multivariate linear regression, radiosensitive histology (p=0.006) was the only significant predictor for metastasis regression at 3 months. Volumetric regression ≥20% at 3 months post-SRT was the only significant prognostic factor for subsequent control in multivariate analysis (HR 0.63, p=0.023), whereas regression ≥65% was no significant predictor.ConclusionsVolumetric regression post-SRT does not occur at a constant rate but is most pronounced in the first 6 weeks to 3 months. Despite decreasing over time, volumetric regression continues beyond 6 months post-radiotherapy and may lead to complete resolution of controlled lesions by 24 months. Radioresistant histology is associated with slower regression. We found that a cutoff of ≥20% regression for the volumetric definition of response at 3 months post-SRT was predictive for subsequent control whereas the currently proposed definition of ≥65% was not. These results have implications for standardized volumetric criteria in future radiotherapy trials for brain metastases.

show abstract

“…(2019) [45] Patients with metastatic castration-resistant prostate cancer NR 193/NR 193(NR)/0(NR) 69.6(7.9;NR) NR Jie Xue et al. (2019) [46] Definitely histopathological results of the primary tumor lesion; patients with only metastatic lesions in brain; with an age over 18 years old; 3D T1 MPRAGE sequence was acquired. Unqualified imaging quality of 3D T1 MPRAGE; data missing; skull metastases and meningeal metastases Dataset 1:1201/1201 Dataset 2:231/231 Dataset 3:220/220 Dataset 1:1201(1201)/0(0) Dataset 2:231(231)/0(0) Dataset 3:220(220)/0(0) Dataset 1:58(18;NR) Dataset 2:60(18;NR) Dataset 3:59(15;NR) Dataset 1:57% Dataset 2:53% Dataset 3:52% Bettina Baessler et al.…”

Section: Resultsmentioning

confidence: 99%

Artificial intelligence performance in detecting tumor metastasis from medical radiology imaging: A systematic review and meta-analysis

et al. 2021

View full text Add to dashboard Cite

Background Early diagnosis of tumor metastasis is crucial for clinical treatment. Artificial intelligence (AI) has shown great promise in the field of medicine. We therefore aimed to evaluate the diagnostic accuracy of AI algorithms in detecting tumor metastasis using medical radiology imaging. Methods We searched PubMed and Web of Science for studies published from January 1, 1997, to January 30, 2020. Studies evaluating an AI model for the diagnosis of tumor metastasis from medical images were included. We excluded studies that used histopathology images or medical wave-form data and those focused on the region segmentation of interest. Studies providing enough information to construct contingency tables were included in a meta-analysis. Findings We identified 2620 studies, of which 69 were included. Among them, 34 studies were included in a meta-analysis with a pooled sensitivity of 82% (95% CI 79–84%), specificity of 84% (82–87%) and AUC of 0·90 (0·87–0·92). Analysis for different AI algorithms showed a pooled sensitivity of 87% (83–90%) for machine learning and 86% (82–89%) for deep learning, and a pooled specificity of 89% (82–93%) for machine learning, and 87% (82–91%) for deep learning. Interpretation AI algorithms may be used for the diagnosis of tumor metastasis using medical radiology imaging with equivalent or even better performance to health-care professionals, in terms of sensitivity and specificity. At the same time, rigorous reporting standards with external validation and comparison to health-care professionals are urgently needed for AI application in the medical field. Funding College students' innovative entrepreneurial training plan program .

show abstract

Deep learning–based detection and segmentation-assisted management of brain metastases

Cited by 90 publications

References 32 publications

Deep convolutional neural networks for automated segmentation of brain metastases trained on clinical data

Deep convolutional neural networks for automated segmentation of brain metastases trained on clinical data

Volumetric Regression in Brain Metastases After Stereotactic Radiotherapy: Time Course, Predictors, and Significance

Artificial intelligence performance in detecting tumor metastasis from medical radiology imaging: A systematic review and meta-analysis

Contact Info

Product

Resources

About