Artificial intelligence-based automatic nidus segmentation of cerebral arteriovenous malformation on time-of-flight magnetic resonance angiography.

OBJECTIVE: Accurate nidus segmentation and quantification have long been challenging but important tasks in the clinical management of Cerebral Arteriovenous Malformation (CAVM). However, there are still dilemmas in nidus segmentation, such as difficulty defining the demarcation of the nidus, observer-dependent variation and time consumption. The aim of this study isto develop an artificial intelligence model to automatically segment the nidus on Time-Of-Flight Magnetic Resonance Angiography (TOF-MRA) images. METHODS: A total of 92patients with CAVM who underwent both TOF-MRA and DSA examinations were enrolled. Two neurosurgeonsmanually segmented the nidusonTOF-MRA images,which were regarded as theground-truth reference. AU-Net-basedAImodelwascreatedfor automatic nidus detectionand segmentationonTOF-MRA images. RESULTS: The meannidus volumes of the AI segmentationmodeland the ground truthwere 5.427 ± 4.996 and 4.824 ± 4.567 mL,respectively. The meandifference in the nidus volume between the two groups was0.603 ± 1.514 mL,which wasnot statisticallysignificant (P = 0.693). The DSC,precision and recallofthe testset were 0.754 ± 0.074, 0.713 ± 0.102 and 0.816 ± 0.098, respectively. The linear correlation coefficient of the nidus volume betweenthesetwo groupswas 0.988, p < 0.001. CONCLUSION: The performance of the AI segmentationmodel is moderate consistent with that of manual segmentation. This AI model has great potential in clinical settings, such as preoperative planning, treatment efficacy evaluation, riskstratification and follow-up.

Identification of predictive factors for better outcomes in LINAC-based radiation treatment for cerebral cavernous malformation.

The effectiveness of radiation therapy in the treatment of cerebral cavernous malformations (CCM) remains debatable. However, numerous studies have shown a reduction in hemorrhage risk following radiotherapy for CCM. Therefore, herein, we share our experiences utilizing linear accelerator (LINAC)-based radiation for treating CCMs, with the aim of identifying the key factors influencing the therapeutic outcomes. We conducted a retrospective review of all patients with non-brainstem CCMs who underwent radiation treatment, as recorded in the NOVALIS registry at our institution. T2-weighted MR images were used for volumetric assessments using the iPlan radiotherapy planning software. To determine the independent predictors of nidus volume reduction and perilesional brain edema (PBE), we performed multivariate Cox regression analysis to calculate hazard ratios. Twenty patients with 31 non-brainstem CCMs were enrolled in this study. Analysis revealed age as an independent predictive factor for both nidus volume reduction and PBE after radiation treatment for CCM. Furthermore, a single fraction dose of 17 Gy or more was identified as an independent predictor of nidus volume decrease, while a single fraction dose of 18 Gy or more was found to be an independent risk factor for PBE in patients with CCM following LINAC-based radiation therapy. LINAC-based radiation therapy for non-brainstem CCMs with a single fraction radiation dose between 16.5 and 17.5 Gy, or a biologically equivalent dose of approximately 120 Gy, may be the most effective at reducing nidus volume and limiting side effects, particularly in patients under the age of 55 years. We further observed that the risk of PBE increased as the maximum radiation dose delivered to a 1 cc volume of the surrounding normal brain exceeded approximately 17.3 Gy. Therefore, we believe that calculating the D1cc of the normal brain may help to predict the occurrence of PBE when radiotherapy is administered to non-brainstem CCMs.