A Fully Automatic Method to Segment Choroid Plexuses in Multiple Sclerosis Using Conventional MRI Sequences

Abstract

BackgroundChoroid plexus (CP) volume has been recently proposed as a proxy for brain neuroinflammation in multiple sclerosis (MS).PurposeTo develop and validate a fast automatic method to segment CP using routinely acquired brain T1‐weighted and FLAIR MRI.Study TypeRetrospective.PopulationFifty‐five MS patients (33 relapsing–remitting, 22 progressive; mean age = 46.8 ± 10.2 years; 31 women) and 60 healthy controls (HC; mean age = 36.1 ± 12.6 years, 33 women).Field Strength/Sequence3D T2‐weighted FLAIR and 3D T1‐weighted gradient echo sequences at 3.0 T.AssessmentBrain tissues were segmented on T1‐weighted sequences and a Gaussian Mixture Model (GMM) was fitted to FLAIR image intensities obtained from the ventricle masks of the SIENAX. A second GMM was then applied on the thresholded and filtered ventricle mask. CP volumes were automatically determined and compared with those from manual segmentation by two raters (with 3 and 10 years' experience; reference standard). CP volumes from previously published automatic segmentation methods (freely available Freesurfer [FS] and FS‐GMM) were also compared with reference standard. Expanded Disability Status Scale (EDSS) score was assessed within 3 days of MRI. Computational time was assessed for each automatic technique and manual segmentation.Statistical TestsComparisons of CP volumes with reference standard were evaluated with Bland Altman analysis. Dice similarity coefficients (DSC) were computed to assess automatic CP segmentations. Volume differences between MS and HC for each method were assessed with t‐tests and correlations of CP volumes with EDSS were assessed with Pearson's correlation coefficients (R). A P value <0.05 was considered statistically significant.ResultsCompared to manual segmentation, the proposed method had the highest segmentation accuracy (mean DSC = 0.65 ± 0.06) compared to FS (mean DSC = 0.37 ± 0.08) and FS‐GMM (0.58 ± 0.06). The percentage CP volume differences relative to manual segmentation were −0.1% ± 0.23, 4.6% ± 2.5, and −0.48% ± 2 for the proposed method, FS, and FS‐GMM, respectively. The Pearson's correlations between automatically obtained CP volumes and the manually obtained volumes were 0.70, 0.54, and 0.56 for the proposed method, FS, and FS‐GMM, respectively. A significant correlation between CP volume and EDSS was found for the proposed automatic pipeline (R = 0.2), for FS‐GMM (R = 0.3) and for manual segmentation (R = 0.4). Computational time for the proposed method (32 ± 2 minutes) was similar to the manual segmentation (20 ± 5 minutes) but <25% of the FS (120 ± 15 minutes) and FS‐GMM (125 ± 15 minutes) methods.Data ConclusionThis study developed an accurate and easily implementable method for automatic CP segmentation in MS using T1‐weighted and FLAIR MRI.Evidence Level1Technical EfficacyStage 4

Longitudinal enlargement of choroid plexus is associated with chronic lesion expansion and neurodegeneration in RRMS patients

“…The development of automated or semi-automated segmentation methods for CP could potentially overcome this limitation, enabling more widespread and efficient analysis of CP changes in patients with RRMS. 19,38,39 In conclusion, this study provides evidence of progressive CP enlargement in patients with RRMS and its association with chronic lesion expansion and brain atrophy, supporting a potential role for the CP in the chronic inflammatory processes and neurodegeneration in RRMS.…”

Associations between the Choroid Plexus and Tau in Alzheimer’s Disease using an Active Learning Segmentation Pipeline

“…For example, it may be one of the possible reasons that no correlation was detected between the ChP and Aβ42 using the multivariate regression model [5]. Although enhanced ChP segmentation methods were proposed through the Gaussian Mixture Model, these methods relied on image intensity, making them susceptible to performance degradation when artifacts are present [11,12]. Second, the overall clearance e ciency of the glymphatic system is associated with the brain volume directly.…”

“…Medical image segmentation requires pixel-wise annotation from experienced radiologists, which can be tedious and costly. For example, the ChP annotation by a radiologist requires 20 ± 5 minutes [12]. Although deep learning ChP segmentation models were proposed, it is challenging to achieve su cient accuracy with a small number of manually labeled images [16,17].…”

Editorial for “A Fully‐Automatic Method to Segment Choroid Plexuses in Multiple Sclerosis Using Conventional MRI Sequences”

“…In the present study, the authors adopted an alternative automatic approach based on a GMM. 8 Accurately segmenting small structures with significant appearance variations, like CPs, from MRI images presents a formidable challenge. In this issue of JMRI, Rocca et al 8 present a readily implementable fully automatic method for CP segmentation leveraging both three-dimensional (3D) T1-weighted and 3D FLAIR brain MRI sequences.…”

“…8 Accurately segmenting small structures with significant appearance variations, like CPs, from MRI images presents a formidable challenge. In this issue of JMRI, Rocca et al 8 present a readily implementable fully automatic method for CP segmentation leveraging both three-dimensional (3D) T1-weighted and 3D FLAIR brain MRI sequences. Notably, this approach employs GMM without relying on FS segmentation as a starting point, setting it apart from other existing methods.…”