Automatic Rating of Perivascular Spaces in Brain MRI Using Bag of Visual Words
Perivascular spaces (PVS), if enlarged and visible in magnetic resonance imaging (MRI), relate to poor cognition, depression in older age, Parkinson’s disease, inflammation, hypertension and cerebral small vessel disease. In this paper we present a fully
- PDF / 803,675 Bytes
- 8 Pages / 439.37 x 666.142 pts Page_size
- 66 Downloads / 192 Views
partment of Neuroimaging Sciences, Centre for Clinical Brain Sciences, University of Edinburgh, Edinburgh, UK [email protected] Department of Cardiovascular Sciences, University of Sheffield, Sheffield, UK
Abstract. Perivascular spaces (PVS), if enlarged and visible in magnetic resonance imaging (MRI), relate to poor cognition, depression in older age, Parkinson’s disease, inflammation, hypertension and cerebral small vessel disease. In this paper we present a fully automatic method to rate the burden of PVS in the basal ganglia (BG) region using structural brain MRI. We used a Support Vector Machine classifier and described the BG following the bag of visual words (BoW) model. The latter was evaluated using a) Scale Invariant Feature Transform (SIFT) descriptors of points extracted from a dense sampling and b) textons, as local descriptors. BoW using SIFT yielded a global accuracy of 82.34 %, whereas using textons it yielded 79.61 %.
Keywords: Brain MRI SIFT · SVM
1
·
Perivascular spaces
·
Bag of visual words
·
Introduction
Perivascular spaces (PVS) are fluid-containing spaces that surround the walls of the brain small arteries, veins and capillaries. They are normally microscopic. When enlarged, they are seen in T2-weighted (T2W) brain MRI as round or linear structures with intensities close to the cerebrospinal fluid (CSF) and with less than 3 mm diameter in cross section. An increase of enlarged PVS has been associated with worse cognition, depression at older ages, Parkinson’s disease, inflammation, hypertension and cerebral small vessel disease in the form of lacunar stroke and vascular dementia [7]. They are increasingly recognised as an important component of the brain’s circulation and fluid drainage pathways. Therefore, quantifying the PVS has a huge interest [1,9,13]. Manual quantifications (i.e. segmentations) of PVS are very time consuming, and the existent automatic methods present serious limitations, due to the overlap in PVS shape, intensity, location and size with these of lacunes [13]. Recently, Wang et al. presented a method based on thresholding T2-weighted (T2W) images acquired using a 1.5T MRI scanner to quantify PVS in the basal c Springer International Publishing Switzerland 2016 A. Campilho and F. Karray (Eds.): ICIAR 2016, LNCS 9730, pp. 642–649, 2016. DOI: 10.1007/978-3-319-41501-7 72
Automatic Rating of Perivascular Spaces in Brain MRI
643
ganglia with good results [15], but it required manual intervention. Cai et al. also developed an automatic method to quantify PVS, but using high resolution 7T MRI scans [1], which still have limited applicability for clinical use. As an alternative to quantitative measurements, several similar visual rating scales have been proposed in recent years. Potter et al. reviewed their ambiguities and combined their strengths to develop a more comprehensive scale which proved to be robust [7]. However, as any visual recognition process, it is subject to observer bias and relatively insensitive to subtle changes. An automatic PVS rating method (e.g
Data Loading...
