Abstract
Enlarged perivascular spaces (PVS) are structural brain changes visible in MRI, and are a marker of cerebral small vessel disease. Most studies use time-consuming and subjective visual scoring to assess these structures. Recently, automated methods to quantify enlarged perivascular spaces have been proposed. Most of these methods have been evaluated only in high resolution scans acquired in controlled research settings. We evaluate and compare two recently published automated methods for the quantification of enlarged perivascular spaces in 76 clinical scans acquired from 9 different scanners. Both methods are neural networks trained on high resolution research scans and are applied without fine-tuning the networks’ parameters. By adapting the preprocessing of clinical scans, regions of interest similar to those computed from research scans can be processed. The first method estimates only the number of PVS, while the second method estimates simultaneously also a high resolution attention map that can be used to detect and segment PVS. The Pearson correlations between visual and automated scores of enlarged perivascular spaces were higher with the second method. With this method, in the centrum semiovale, the correlation was similar to the inter-rater agreement, and also similar to the performance in high resolution research scans. Results were slightly lower than the inter-rater agreement for the hippocampi, and noticeably lower in the basal ganglia. By computing attention maps, we show that the neural networks focus on the enlarged perivascular spaces. Assessing the burden of said structures in the centrum semiovale with the automated scores reached a satisfying performance, could be implemented in the clinic and, e.g., help predict the bleeding risk related to cerebral amyloid angiopathy.
Highlights
Enlarged perivascular spaces (PVS) are structural brain changes visible on MRI
The preprocessed images were given as input to two different types of neural networks proposed for automated PVS quantification: (1) a neural network with four convolutional layers and a max-pooling layer which outputs the number of PVS in a region [4] and that we call CNN, and (2) GP-Unet, a similar neural network proposed by the same authors [5], in which the downsampling path is followed by an upsampling path to enable weakly supervised detection of PVS
The visual PVS scores were highly correlated to the automated PVS scores of GP-Unet in the centrum semiovale (0.78 Pearson correlation), were moderately correlated in the hippocampi (0.52), and a lower correlation in the basal ganglia (0.28)
Summary
Enlarged perivascular spaces (PVS) are structural brain changes visible on MRI. They can be identified as thin hyperintense tubular structures on T2-weighted MRI scans. In most studies, enlarged perivascular spaces are quantified using visual scores that either classify the burden of PVS in several categories [8], or count PVS [1] These quantification methods are tedious and observer-dependent. Boespflug et al [2] proposed an automated quantification method combining image intensities and morphologic features from several MRI sequences They evaluated their method in the centrum semiovale in research scans. Dubost et al [4] proposed to quantify PVS burden in four brain regions – midbrain, hippocampi, basal ganglia, and centrum semiovale – with neural network regressors trained with image level labels: the count of PVS in the target brain region. The same authors [5] proposed to use a more advanced model (GP-Unet) for weakly supervised detection of enlarged perivascular spaces This method estimates simultaneously the number of PVS and a high resolution attention map that can be used to detect and segment PVS.
Sign-in/Register to view highlights & summary 
Submitted Version (
Free)
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call