Automated hippocampal segmentation in patients with epilepsy: Available free online

Abstract

PurposeHippocampal sclerosis, a common cause of refractory focal epilepsy, requires hippocampal volumetry for accurate diagnosis and surgical planning. Manual segmentation is time-consuming and subject to interrater/intrarater variability. Automated algorithms perform poorly in patients with temporal lobe epilepsy. We validate and make freely available online a novel automated method.MethodsManual hippocampal segmentation was performed on 876, 3T MRI scans and 202, 1.5T scans. A template database of 400 high-quality manual segmentations was used to perform automated segmentation of all scans with a multi-atlas–based segmentation propagation method adapted to perform label fusion based on local similarity to ensure accurate segmentation regardless of pathology. Agreement between manual and automated segmentations was assessed by degree of overlap (Dice coefficient) and comparison of hippocampal volumes.Key FindingsThe automated segmentation algorithm provided robust delineation of the hippocampi on 3T scans with no more variability than that seen between different human raters (Dice coefficients: interrater 0.832, manual vs. automated 0.847). In addition, the algorithm provided excellent results with the 1.5T scans (Dice coefficient 0.827), and automated segmentation remained accurate even in small sclerotic hippocampi. There was a strong correlation between manual and automated hippocampal volumes (Pearson correlation coefficient 0.929 on the left and 0.941 on the right in 3T scans).SignificanceWe demonstrate reliable identification of hippocampal atrophy in patients with hippocampal sclerosis, which is crucial for clinical management of epilepsy, particularly if surgical treatment is being contemplated. We provide a free online Web-based service to enable hippocampal volumetry to be available globally, with consequent greatly improved evaluation of those with epilepsy.

Automated versus manual hippocampal segmentation in preoperative and postoperative patients with epilepsy

“…The manual method uses visual identification of hippocampal margins, whereas FreeSurfer transforms the images to standard coordinates, and then utilizes voxel intensities and probabilities to delineate the hippocampus. A study using a different method found a good correlation between automated and manual hippocampal volume measurements, with a tendency for the former to be larger …”

“…Recently, an automated segmentation algorithm was developed specifically for adult patients with epilepsy by using a template database comprising epilepsy patients with a range of pathologies rather than healthy subjects . Following alignment, segmentation is performed using a local ranking strategy .…”

Multi-atlas segmentation of biomedical images: A survey

“…Many methods have been developed to parcellate the whole brain, segmenting it into a large number of regions (Aljabar et al, 2008; Babalola et al, 2009; Fonov et al, 2012; Han et al, 2009; Heckemann et al, 2010, 2011; Keihaninejad et al, 2010; Kotrotsou et al, 2014; Svarer et al, 2005; Wang et al, 2012; Ledig et al, 2014), while other studies have focused on small sets or individual ROIs, such as the caudate nucleus (van Rikxoort et al, 2007b); the cerebellum (Park et al, 2014; Van Der Lijn et al, 2012; Weier et al, 2014); the amygdala (Hanson et al, 2012; Klein-Koerkamp et al, 2014); the corpus callosum (Ardekani et al, 2014; Gao et al, 2014; Meyer, 2014); the striatum (Janes et al, 2014); the subthalamic nucleus, red nucleus and substantia nigra (Xiao et al, 2014b,a); the ventricles (Chou et al, 2008; Raamana et al, 2014); and, most notably, the hippocampus, which has attracted much attention due to its association with dementia and Alzheimer’s disease (Akhondi-Asl et al, 2010; Bishop et al, 2010; Clerx et al, 2013; Hammers et al, 2007; Iglesias et al, 2010; Kim et al, 2012; Leung et al, 2010; Pipitone et al, 2014; Pluta et al, 2012; Raamana et al, 2014; van der Lijn et al, 2008; Van Der Lijn et al, 2012; Winston et al, 2013; Wolz et al, 2010b; Yushkevich et al, 2010; Platero et al, 2014; Ta et al, 2014). …”

Multiatlas-Based Segmentation Editing With Interaction-Guided Patch Selection and Label Fusion

“…III-B, we used all user interactions inserted on erroneous parts in the entire image to perform the segmentation editing. After the first round of editing, the DSC scores of most cases were higher than some inter-rater DSC scores reported in the literature [38–40], and also had no notable qualitative errors. Thus, we usually performed only the first round of editing in this study.…”