Federated Learning: A Cross-Institutional Feasibility Study of Deep Learning Based Intracranial Tumor Delineation Framework for Stereotactic Radiosurgery.

BACKGROUND: Deep learning-based segmentation algorithms usually required large or multi-institute data sets to improve the performance and ability of generalization. However, protecting patient privacy is a key concern in the multi-institutional studies when conventional centralized learning (CL) is used. PURPOSE: To explores the feasibility of a proposed lesion delineation for stereotactic radiosurgery (SRS) scheme for federated learning (FL), which can solve decentralization and privacy protection concerns. STUDY TYPE: Retrospective. SUBJECTS: 506 and 118 vestibular schwannoma patients aged 15-88 and 22-85 from two institutes, respectively; 1069 and 256 meningioma patients aged 12-91 and 23-85, respectively; 574 and 705 brain metastasis patients aged 26-92 and 28-89, respectively. FIELD STRENGTH/SEQUENCE: 1.5T, spin-echo, and gradient-echo [Correction added after first online publication on 21 August 2023. Field Strength has been changed to "1.5T" from "5T" in this sentence.]. ASSESSMENT: The proposed lesion delineation method was integrated into an FL framework, and CL models were established as the baseline. The effect of image standardization strategies was also explored. The dice coefficient was used to evaluate the segmentation between the predicted delineation and the ground truth, which was manual delineated by neurosurgeons and a neuroradiologist. STATISTICAL TESTS: The paired t-test was applied to compare the mean for the evaluated dice scores (p < 0.05). RESULTS: FL performed the comparable mean dice coefficient to CL for the testing set of Taipei Veterans General Hospital regardless of standardization and parameter; for the Taichung Veterans General Hospital data, CL significantly (p < 0.05) outperformed FL while using bi-parameter, but comparable results while using single-parameter. For the non-SRS data, FL achieved the comparable applicability to CL with mean dice 0.78 versus 0.78 (without standardization), and outperformed to the baseline models of two institutes. DATA CONCLUSION: The proposed lesion delineation successfully implemented into an FL framework. The FL models were applicable on SRS data of each participating institute, and the FL exhibited comparable mean dice coefficient to CL on non-SRS dataset. Standardization strategies would be recommended when FL is used. LEVEL OF EVIDENCE: 4 TECHNICAL EFFICACY: Stage 1.

Using the deformity index of vital structures to predict outcome of patients with large vestibular schwannomas after Gamma Knife radiosurgery.

PURPOSE: Microsurgery is the mainstay of treatment for large vestibular schwannomas (VS), but the benefits of radiosurgery remain incompletely defined. Here, we aim to use automated volumetric analysis software to quantify the degree of brain stem deformity to predict long-term outcomes of patients with large VS following GKRS. METHODS: Between 2003 and 2020, 39 patients with large VS (volume > 8 cc) undergoing GKRS with a margin dose of 10-12 Gy were analyzed. The reconstruction 3D MRI was used to evaluate the extent of deformity for predicting the long-term outcome of patients. RESULTS: Their mean tumor volume was 13.7 ± 6.3 cc, and their mean follow-up after GKRS was 86.7 ± 65.3 months. Favorable clinical outcome was observed in 26 (66.7%) patients, while 13 (33.3%) patients had treatment failure. Patients with small tumor volumes, low vital structure deformity indice [(TV/(BSV + CerV) and (TV + EV)/(BSV + CerV)], and long distance of tumor to the central line were more likely to have favorable clinical outcome after GKRS. Significant prognostic value was with tumor shrinkage ratio (< 50%) were CV, CV/TV, TV/CerV, (TV + EV)/(BSV + CerV), and the distance of tumor to the central line. In cox regression, favorable clinical outcome was correlated with the Charlson comorbidity index and cochlear dosage (both p < 0.05). In multivariant analysis, tumor regression was highly correlated with the CV/TV ratio (p < 0.001). CONCLUSIONS: The brainstem deformity ratio is likely a useful index to assess the clinical and tumor regression outcomes. Clinical outcomes are multifactorial and the tumor regression was highly correlated with the ratio of cystic components.

Gamma Knife Radiosurgery for Indirect Dural Carotid-Cavernous Fistula: Long-Term Ophthalmological Outcome.

Objective: The leading treatment option for dural carotid−cavernous sinus fistula is an endovascular approach with immediate improvement. Alternatively, radiosurgery is a slow response for obliterating the fistula and poses a radiation risk to the optic apparatus and the associated cranial nerves and blood vessels. In this study, we retrieved cases from a prospective database to assess the ophthalmological outcomes and complications in treating dural carotid cavernous sinus fistula with gamma knife radiosurgery (GKRS). Material and Methods: We retrieved a total of 65 cases of carotid cavernous sinus fistula treated with GKRS with margin dose of 18−20 Gy from 2003 to 2018 and reviewed the ophthalmological records required for our assessment. Results: The mean target volume was 2 ± 1.43 cc. The onset of symptom alleviated after GKRS was 3.71 ± 7.68 months. There were two cases with residual chemosis, two with cataract, two with infarction, one with transient optic neuropathy, and four with residual cranial nerve palsy, but none with glaucoma or dry eyes. In MRA analysis, total obliteration of the fistula was noted in 64 cases with no detectable ICA stenosis nor cavernous sinus thrombosis. In the Cox regression analysis, post-GKRS residual cranial nerve palsy was highly correlated to targeted volume (p < 0.05) and age (p < 0.05). The occurrence of post-GKRS cataract was related to the initial symptom of chemosis (p < 0.05). Conclusion: GKRS for carotid cavernous sinus fistula offers a high obliteration rate and preserves the cavernous sinus vascular structure while conferring a low risk of treatment complications such as adverse radiation risk to the optic apparatus and adjacent cranial nerves.

Magnetic Resonance Angiography in the Diagnosis of Cerebral Arteriovenous Malformation and Dural Arteriovenous Fistulas: Comparison of Time-Resolved Magnetic Resonance Angiography and Three Dimensional Time-of-Flight Magnetic Resonance Angiography.

BACKGROUND: Traditional digital subtraction angiography (DSA) is currently the gold standard diagnostic method for the diagnosis and evaluation of cerebral arteriovenous malformation (AVM) and dural arteriovenous fistulas (dAVF). OBJECTIVES: The aim of this study was to analyze different less invasive magnetic resonance angiography (MRA) images, time-resolved MRA (TR-MRA) and three-dimensional time-of-flight MRA (3D TOF MRA) to identify their diagnostic accuracy and to determine which approach is most similar to DSA. PATIENTS AND METHODS: A total of 41 patients with AVM and dAVF at their initial evaluation or follow-up after treatment were recruited in this study. We applied time-resolved angiography using keyhole (4D-TRAK) MRA to perform TR-MRA and 3D TOF MRA examinations simultaneously followed by DSA, which was considered as a standard reference. Two experienced neuroradiologists reviewed the images to compare the diagnostic accuracy, arterial feeder and venous drainage between these two MRA images. Inter-observer agreement for different MRA images was assessed by Kappa coefficient and the differences of diagnostic accuracy between MRA images were evaluated by the Wilcoxon rank sum test. RESULTS: Almost all vascular lesions (92.68%) were correctly diagnosed using 4D-TRAK MRA. However, 3D TOF MRA only diagnosed 26 patients (63.41%) accurately. There were statistically significant differences regarding lesion diagnostic accuracy (P = 0.008) and venous drainage identification (P < 0.0001) between 4D-TRAK MRA and 3D TOF MRA. The results indicate that 4D-TRAK MRA is superior to 3D TOF MRA in the assessment of lesions. CONCLUSION: Compared with 3D TOF MRA, 4D-TRAK MRA proved to be a more reliable screening modality and follow-up method for the diagnosis of cerebral AVM and dAVF.

Civilian gunshot wounds to the brain.

BACKGROUND: Civilian gunshot wounds to the brain are relatively rare, and a much-neglected subject in Taiwan. We present our experience with 16 patients who sustained gunshot wounds to the brain, and then identify factors determining the respective outcomes. METHODS: From 1988-2002, data from 16 patients with civilian gunshot wounds to the brain, who were treated at Taichung Veterans General Hospital, were retrospectively reviewed. Historical information, clinical manifestations, and imaging findings were described. Based on Glasgow Outcome Score (GOS), patients were divided into a poor-outcome group (GOS, 1-2) and satisfactory-outcome group (GOS, 3-5) for further analysis. RESULTS: The overall mortality rate was 31.3% (5 of 16 patients). Thirteen patients underwent surgery, and the surgical mortality rate was 15.4% (2 of 13 patients). Of the 7 patients with a Glasgow Coma Scale (GCS) score of more than 8, all survived with satisfactory outcomes; of the 9 patients with a GCS score of less than 8, 8 had poor outcomes (5 died, and 3 were in a persistent vegetative state); p < 0.005 between the 2 GCS groups. In addition, the rate of satisfactory outcome was significantly higher in 7 patients with limited brain injury, as determined by computed tomography (CT) scan, than in 8 patients with extensive brain injury (86% vs 25%; p < 0.05). CONCLUSION: GCS score on admission, and the extent of brain injury as visualized by CT scan, seem to be the 2 most significant predictors of outcome in cranio-cerebral gunshot wounds. Patients with a GCS score of more than 8, or brain lesions limited to a single lobe of the brain, may benefit from aggressive management.