Accuracy and reproducibility of automated white matter hyperintensities segmentation with lesion segmentation tool: A European multi-site 3T study

Ribaldi, Federica; Altomare, Daniele; Jovicich, Jorge; Ferrari, Clarissa; Picco, Agnese; Pizzini, Francesca Benedetta; Soricelli, Andrea; Mega, Anna; Ferretti, Andrea; Drevelegas, Antonios; Bosch, Beatríz; Müller, Bernhard; Marra, Camillo; Cavaliere, Carlo; Bartrés‐Faz, David; Nobili, Flavio; Alessandrini, Franco; Barkhof, Frederik; Gros‐Dagnac, Hélène; Ranjeva, Jean Philippe; Wiltfang, Jens; Kuijer, Joost P.A.; Sein, Julien; Hoffmann, Karl‐Titus; Roccatagliata, Luca; Parnetti, Lucilla; Tsolaki, Magda; Constantinidis, Manos; Aiello, Marco; Salvatore, Marco; Montalti, Martina; Caulo, Massimo; Didic, Mira; Bargalló, Núria; Blin, Olivier; Rossini, Paolo Maria; Schönknecht, Peter; Floridi, Piero; Payoux, Pierre; Visser, Pieter Jelle; Bordet, Régis; Lopes, Renaud; Tarducci, R.; Bombois, Stéphanie; Hensch, Tilman; Fiedler, Ute; Richardson, Jill C.; Frisoni, Giovanni B.; Marizzoni, Moira

doi:10.1016/j.mri.2020.11.008

Cited by 32 publications

(28 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…As discussed in the literature, WML segmentation of age-related, mild loads is a challenging problem with the low detection rates in the 30% range as opposed to the > 80% range seen with large WML loads (> 15 mL). In this study, the average Dice index was lower than 0.5 for all automatic methods, in line with the current state of the literature 25 , 29 , 45 , 50 , 51 and not as promising as segmentation of the typically large multiple sclerosis (MS) lesions reported in the literature 22 . The main cause of low indices is likely missed segmentations in parts of the brain that contain high density white matter fibers, with small intensity differences between the white matter lesions and background normal white matter fibers.…”

Section: Discussionsupporting

confidence: 80%

“…Roura et al 25 introduced the SALEM-LS (SLS) algorithm, which uses an adaptive outlier algorithm to threshold outliers as WML from grey matter, ranking first on this dataset with a total score of 82.34. The Lesion Prediction Algorithm (LPA) developed by Schmidt et al 26 – 28 was based on logistic regression and successfully applied to the longitudinal analysis of WML recently 29 . Vanderbecq and colleagues 30 recommended both the SLS and LPA algorithms as reasonable first choice WML segmentation tools after comparison and validation of seven different algorithms in their recent study.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Detection of subtle white matter lesions in MRI through texture feature extraction and boundary delineation using an embedded clustering strategy

Ong

Young

Sulaiman

et al. 2022

Sci Rep

View full text Add to dashboard Cite

White matter lesions (WML) underlie multiple brain disorders, and automatic WML segmentation is crucial to evaluate the natural disease course and effectiveness of clinical interventions, including drug discovery. Although recent research has achieved tremendous progress in WML segmentation, accurate detection of subtle WML present early in the disease course remains particularly challenging. Here we propose an approach to automatic WML segmentation of mild WML loads using an intensity standardisation technique, gray level co-occurrence matrix (GLCM) embedded clustering technique, and random forest (RF) classifier to extract texture features and identify morphology specific to true WML. We precisely define their boundaries through a local outlier factor (LOF) algorithm that identifies edge pixels by local density deviation relative to its neighbors. The automated approach was validated on 32 human subjects, demonstrating strong agreement and correlation (excluding one outlier) with manual delineation by a neuroradiologist through Intra-Class Correlation (ICC = 0.881, 95% CI 0.769, 0.941) and Pearson correlation (r = 0.895, p-value < 0.001), respectively, and outperforming three leading algorithms (Trimmed Mean Outlier Detection, Lesion Prediction Algorithm, and SALEM-LS) in five of the six established key metrics defined in the MICCAI Grand Challenge. By facilitating more accurate segmentation of subtle WML, this approach may enable earlier diagnosis and intervention.

show abstract

Section: Discussionsupporting

confidence: 80%

Section: Introductionmentioning

confidence: 99%

Detection of subtle white matter lesions in MRI through texture feature extraction and boundary delineation using an embedded clustering strategy

Ong

Young

Sulaiman

et al. 2022

Sci Rep

View full text Add to dashboard Cite

show abstract

“…However, the spatial relationships between brain functional connectivity and structural changes require further exploration in our future research. Finally, further studies employing more advanced imaging methodologies to map the brain connectome and segment lesions appropriately and precisely may provide more comprehensive insights into the connectome-based mechanisms underlying ILA-related cognitive impairment ( Ribaldi et al, 2021 ).…”

Section: Discussionmentioning

confidence: 99%

Disrupted Structural Brain Connectome Is Related to Cognitive Impairment in Patients With Ischemic Leukoaraiosis

Lü

Wang

Cui

et al. 2021

Front. Hum. Neurosci.

View full text Add to dashboard Cite

Ischemic leukoaraiosis (ILA) is related to cognitive impairment and vascular dementia in the elderly. One possible mechanism could be the disruption of white matter (WM) tracts and network function that connect distributed brain regions involved in cognition. The purpose of this study was to investigate the relationship between structural connectome and cognitive functions in ILA patients. A total of 89 patients with ILA (Fazekas score ≥ 3) and 90 healthy controls (HCs) underwent comprehensive neuropsychological examinations and diffusion tensor imaging scans. The tract-based spatial statistics approach was employed to investigate the WM integrity. Graph theoretical analysis was further applied to construct the topological architecture of the structural connectome in ILA patients. Partial correlation analysis was used to investigate the relationships between network measures and cognitive performances in the ILA group. Compared with HCs, the ILA patients showed widespread WM integrity disruptions. The ILA group displayed increased characteristic path length (Lp) and decreased global network efficiency at the level of the whole brain relative to HCs, and reduced nodal efficiencies, predominantly in the frontal–subcortical and limbic system regions. Furthermore, these structural connectomic alterations were associated with cognitive impairment in ILA patients. The association between WM changes (i.e., fractional anisotropy and mean diffusivity measures) and cognitive function was mediated by the structural connectivity measures (i.e., local network efficiency and Lp). In conclusion, cognitive impairment in ILA patients is related to microstructural disruption of multiple WM fibers and topological disorganization of structural networks, which have implications in understanding the relationship between ILA and the possible attendant cognitive impairment.

show abstract

“…As a first step, the LST (Schmidt et al, 2012 ) FLAIR segmentation software was applied to all 2D FLAIR images. Despite LST being considered accurate and a strong performer in comparative studies (Heinen et al, 2019 ; Ribaldi et al, 2021 ; Vanderbecq et al, 2020 ), a small but significant fraction of the results inevitably contained some inaccurate or unsatisfactory segmentation (Figure S2 ). While we could simply review all images and discard these failed cases, this would risk losing some interesting corner‐cases, which would run contrary to our goal of robustness.…”

Section: Methodsmentioning

confidence: 99%

Higher‐resolution quantification of white matter hypointensities by large‐scale transfer learning from 2D images on the JPSC‐AD cohort

Thyreau

Tatewaki

Chen

et al. 2022

Human Brain Mapping

View full text Add to dashboard Cite

White matter lesions (WML) commonly occur in older brains and are quantifiable on MRI, often used as a biomarker in Aging research. Although algorithms are regularly proposed that identify these lesions from T2‐fluid‐attenuated inversion recovery (FLAIR) sequences, none so far can estimate lesions directly from T1‐weighted images with acceptable accuracy. Since 3D T1 is a polyvalent and higher‐resolution sequence, it could be beneficial to obtain the distribution of WML directly from it. However a serious difficulty, both for algorithms and human, can be found in the ambiguities of brain signal intensity in T1 images. This manuscript shows that a cross‐domain ConvNet (Convolutional Neural Network) approach can help solve this problem. Still, this is non‐trivial, as it would appear to require a large and varied dataset (for robustness) labelled at the same high resolution (for spatial accuracy). Instead, our model was taught from two‐dimensional FLAIR images with a loss function designed to handle the super‐resolution need. And crucially, we leveraged a very large training set for this task, the recently assembled, multi‐sites Japan Prospective Studies Collaboration for Aging and Dementia (JPSC‐AD) cohort. We describe the two‐step procedure that we followed to handle such a large number of imperfectly labeled samples. A large‐scale accuracy evaluation conducted against FreeSurfer 7, and a further visual expert rating revealed that WML segmentation from our ConvNet was consistently better. Finally, we made a directly usable software program based on that trained ConvNet model, available at https://github.com/bthyreau/deep-T1-WMH .

show abstract

Accuracy and reproducibility of automated white matter hyperintensities segmentation with lesion segmentation tool: A European multi-site 3T study

Cited by 32 publications

References 48 publications

Detection of subtle white matter lesions in MRI through texture feature extraction and boundary delineation using an embedded clustering strategy

Detection of subtle white matter lesions in MRI through texture feature extraction and boundary delineation using an embedded clustering strategy

Disrupted Structural Brain Connectome Is Related to Cognitive Impairment in Patients With Ischemic Leukoaraiosis

Higher‐resolution quantification of white matter hypointensities by large‐scale transfer learning from 2D images on the JPSC‐AD cohort

Contact Info

Product

Resources

About