Dense, deep learning-based intracranial aneurysm detection on TOF MRI using two-stage regularized U-Net.

BACKGROUND AND PURPOSE: The prevalence of unruptured intracranial aneurysms in the general population is high and aneurysms are usually asymptomatic. Their diagnosis is often fortuitous on MRI and might be difficult and time consuming for the radiologist. The purpose of this study was to develop a deep learning neural network tool for automated segmentation of intracranial arteries and automated detection of intracranial aneurysms from 3D time-of-flight magnetic resonance angiography (TOF-MRA). MATERIALS AND METHODS: 3D TOF-MRA with aneurysms were retrospectively extracted. All were confirmed with angiography. The data were divided into two sets: a training set of 24 examinations and a test set of 25 examinations. Manual annotations of intracranial blood vessels and aneurysms were performed by neuroradiologists. A double convolutional neuronal network based on the U-Net architecture with regularization was used to increase performance despite a small amount of training data. The performance was evaluated for the test set. Subgroup analyses according to size and location of aneurysms were performed. RESULTS: The average processing time was 15 min. Overall, the sensitivity and the positive predictive value of the proposed algorithm were 78% (21 of 27; 95% CI: 62-94) and 62% (21 of 34; 95%CI: 46-78) respectively, with 0.5 FP/case. Despite gradual improvement in sensitivity regarding aneurysm size, there was no significant difference of sensitivity detection between subgroups of size and location. CONCLUSIONS: This developed tool based on a double CNN with regularization trained with small dataset, enables accurate intracranial arteries segmentation as well as effective aneurysm detection on 3D TOF MRA.

MRI predictive score of pial vascularization of supratentorial intracranial meningioma.

OBJECTIVE: Meningiomas are highly vascularized tumors which may recruit pial blood supply. Pial supply complicates tumor treatment in numerous ways. The objective of this study was to establish a reliable MRI-based diagnostic score to predict the existence of pial blood supply in supratentorial intracranial meningiomas and then correlate the score with clinical and surgical outcomes and histopathological findings. METHODS: We performed a retrospective analysis of supratentorial histologically proven meningiomas in our institution from 2010 to 2018. A score was built based on MRI criteria and correlated with digital subtraction angiography (DSA) pial vascularization assessment. The score was then validated on a second independent population recruited with the same modalities. RESULTS: Logistic regression identified four parameters related to pial blood supply which were used to build the score: skull base location, tumor size > 45 mm, peritumoral flow voids, and incomplete cerebrospinal fluid rim. The overall diagnostic performance in predicting pial blood supply was as follows: sensitivity 97.8%, specificity 76.9%, predictive positive value 88.2%, negative predictive value 95.2%, and accuracy 90.3%. Inter-reader agreement and Cohen's kappa were good, respectively, of 90.7% and 0.69. A high score was associated with aggressive meningioma (World Health Organization II-III) (p = 0.04) and with greater importance of pial supply relative to dural supply. CONCLUSIONS: We have identified a reliable way to use MRI to predict the existence of pial blood supply in supratentorial intracranial meningiomas. A higher score also predicted higher grade meningioma. KEY POINTS: • Accurate and reproducible MRI score composed of four items to predict the existence of pial blood supply in supratentorial meningioma. • High score is associated with high-grade meningioma (WHO II-III) but also with greater importance of pial supply relative to dural supply.