A Practical Roadmap to Implementing Deep Learning Segmentation in the Clinical Neuroimaging Research Workflow.

BACKGROUND: Thanks to the proliferation of open-source tools, we are seeing an exponential growth of machine-learning applications, and its integration has become more accessible, particularly for segmentation tools in neuroimaging. METHODS: This article explores a generalized methodology that harnesses these tools and aims/enables to expedite and enhance the reproducibility of clinical research. Herein, critical considerations include hardware, software, neural network training strategies, and data labeling guidelines. More specifically, we advocate an iterative approach to model training and transfer learning, focusing on internal validation and outlier handling early in the labeling process and fine-tuning later. RESULTS: The iterative refinement process allows experts to intervene and improve model reliability while cutting down on their time spent in manual work. A seamless integration of the final model's predictions into clinical research is proposed to ensure standardized and reproducible results. CONCLUSIONS: In short, this article provides a comprehensive framework for accelerating research using machine-learning techniques for image segmentation.

Clinical and radiologic outcomes after stereotactic radiosurgery for meningiomas in direct contact with the optic apparatus: an international multicenter study.

OBJECTIVE: Resection of meningiomas in direct contact with the anterior optic apparatus carries risk of injury to the visual pathway. Stereotactic radiosurgery (SRS) offers a minimally invasive alternative. However, its use is limited owing to the risk of radiation-induced optic neuropathy. Few SRS studies have specifically assessed the risks and benefits of treating meningiomas in direct contact with the optic nerve, chiasm, or optic tract. The authors hypothesized that SRS is safe for select patients with meningiomas in direct contact with the anterior optic apparatus. METHODS: The authors performed an international multicenter retrospective analysis of 328 patients across 11 institutions. All patients had meningiomas in direct contract with the optic apparatus. Patients were followed for a median duration of 56 months after SRS. Neurological examinations, including visual function evaluations, were performed at follow-up visits. Clinical and treatment variables were collected at each site according to protocol. Tumor volumes were assessed with serial MR imaging. Variables predictive of visual deficit were identified using univariable and multivariable logistic regression. RESULTS: SRS was the initial treatment modality for 64.6% of patients, and 93% of patients received SRS as a single fraction. Visual information was available for 302 patients. Of these patients, visual decline occurred in 29 patients (9.6%), of whom 12 (41.4%) had evidence of tumor progression. Visual decline in the remaining 17 patients (5.6%) was not associated with tumor progression. Pre-SRS Karnofsky Performance Status predicted visual decline in adjusted analysis (adjusted OR 0.9, 95% CI 0.9-1.0, p < 0.01). Follow-up imaging data were available for 322 patients. Of these patients, 294 patients (91.3%) had radiographic evidence of stability or tumor regression at last follow up. Symptom duration was associated with tumor progression in adjusted analysis (adjusted OR 1.01, adjusted 95% CI 1.0-1.02, adjusted p = 0.02). CONCLUSIONS: In this international multicenter study, the vast majority of patients exhibited tumor control and preservation of visual function when SRS was used to treat meningioma in direct contact with the anterior optic pathways. SRS is a relatively safe treatment modality for select patients with perioptic meningiomas in direct contact with the optic apparatus.

Neurofibromatosis type 2-associated meningiomas: an international multicenter study of outcomes after Gamma Knife stereotactic radiosurgery.

OBJECTIVE: The management of neurofibromatosis type 2 (NF2)-associated meningiomas is challenging. The role of Gamma Knife radiosurgery (GKRS) in the treatment of these tumors remains to be fully defined. In this study, the authors aimed to examine the role of GKRS in the treatment of NF2-associated meningiomas and to evaluate the outcomes and complications after treatment. METHODS: Seven international medical centers contributed data for this retrospective cohort. Tumor progression was defined as a ≥ 20% increase from the baseline value. The clinical features, treatment details, outcomes, and complications were studied. The median follow-up was 8.5 years (range 0.6-25.5 years) from the time of initial GKRS. Shared frailty Cox regression was used for analysis. RESULTS: A total of 204 meningiomas in 39 patients treated with GKRS were analyzed. Cox regression analysis showed that increasing the maximum dose (p = 0.02; HR 12.2, 95% CI 1.287-116.7) and a lower number of meningiomas at presentation (p = 0.03; HR 0.9, 95% CI 0.821-0.990) were predictive of better tumor control in both univariable and multivariable settings. Age at onset, sex, margin dose, location, and presence of neurological deficit were not predictive of tumor progression. The cumulative 10-year progression-free survival was 94.8%. Radiation-induced adverse effects were noted in 4 patients (10%); these were transient and managed medically. No post-GKRS malignant transformation was noted in 287 person-years of follow-up. CONCLUSIONS: GKRS achieved effective tumor control with a low and generally acceptable rate of complications in NF2-associated meningiomas. There did not appear to be an appreciable risk of post-GKRS-induced malignancy in patients with NF2-treated meningiomas.