Lesion detection in epilepsy surgery: Lessons from a prospective evaluation of a machine learning algorithm

AIM

To evaluate a lesion detection algorithm designed to detect focal cortical dysplasia (FCD) in children undergoing stereoelectroencephalography (SEEG) as part of their presurgical evaluation for drug-resistant epilepsy.

METHOD

This was a prospective, single-arm, interventional study (Idea, Development, Exploration, Assessment, and Long-Term Follow-Up phase 1/2a). After routine SEEG planning, structural magnetic resonance imaging sequences were run through an FCD lesion detection algorithm to identify putative clusters. If the top three clusters were not already sampled, up to three additional SEEG electrodes were added. The primary outcome measure was the proportion of patients who had additional electrode contacts in the SEEG-defined seizure-onset zone (SOZ).

RESULTS

Twenty patients (median age 12 years, range 4-18 years) were enrolled, one of whom did not undergo SEEG. Additional electrode contacts were part of the SOZ in 1 out of 19 patients while 3 out of 19 patients had clusters that were part of the SOZ but they were already implanted. A total of 16 additional electrodes were implanted in nine patients and there were no adverse events from the additional electrodes.

INTERPRETATION

We demonstrate early-stage prospective clinical validation of a machine learning lesion detection algorithm used to aid the identification of the SOZ in children undergoing SEEG. We share key lessons learnt from this evaluation and emphasize the importance of robust prospective evaluation before routine clinical adoption of such algorithms.

WHAT THIS PAPER ADDS

The focal cortical dysplasia detection algorithm collocated with the seizure-onset zone (SOZ) in 4 out of 19 patients. The algorithm changed the resection boundaries in 1 of 19 patients undergoing stereoelectroencephalography for drug-resistant epilepsy. The patient with an altered resection due to the algorithm was seizure-free 1 year after resective surgery. Overall, the algorithm did not increase the proportion of patients in whom SOZ was identified.

Paediatric subarachnoid haemorrhage and severe vasospasm secondary to traumatic pseudoaneurysm of a fenestrated vertebral artery: a case report and review of the literature

Paediatric intracranial aneurysms are rare entities accounting for less than 5% of all age intracranial aneurysms. Traumatic aneurysms are more common in children and have an association with anatomical variations such as arterial fenestrations. Here, we present a case of a child initially presenting with traumatic subarachnoid haemorrhage who returned to baseline and was discharged home only to return within 2 weeks with diffuse subarachnoid and intraventricular re-haemorrhage. A dissecting aneurysm of a duplicated (fenestrated) V4 vertebral artery segment was identified as a rare cause of rebleeding. We describe a course complicated by severe vasospasm delaying aneurysm detection and treatment. Dissecting aneurysms in children should be considered in all cases of delayed post-traumatic cranial rebleeding, particularly where there is anomalous arterial anatomy.

Epileptogenic Tubers Are Associated with Increased Kurtosis of Susceptibility Values: A Combined Quantitative Susceptibility Mapping and Stereoelectroencephalography Pilot Study

BACKGROUND AND PURPOSE

Prior studies have found an association between calcification and the epileptogenicity of tubers in tuberous sclerosis complex. Quantitative susceptibility mapping is a novel tool sensitive to magnetic susceptibility alterations due to tissue calcification. We assessed the utility of quantitative susceptibility mapping in identifying putative epileptogenic tubers in tuberous sclerosis complex using stereoelectroencephalography data as ground truth.

MATERIALS AND METHODS

We studied patients with tuberous sclerosis complex undergoing stereoelectroencephalography at a single center who had multiecho gradient-echo sequences available. Quantitative susceptibility mapping and R2* values were extracted for all tubers on the basis of manually drawn 3D ROIs using T1- and T2-FLAIR sequences. Characteristics of quantitative susceptibility mapping and R2* distributions from implanted tubers were compared using binary logistic generalized estimating equation models designed to identify ictal (involved in seizure onset) and interictal (persistent interictal epileptiform activity) tubers. These models were then applied to the unimplanted tubers to identify potential ictal and interictal tubers that were not sampled by stereoelectroencephalography.

RESULTS

A total of 146 tubers were identified in 10 patients, 76 of which were sampled using stereoelectroencephalography. Increased kurtosis of the tuber quantitative susceptibility mapping values was associated with epileptogenicity (P = .04 for the ictal group and P = .005 for the interictal group) by the generalized estimating equation model. Both groups had poor sensitivity (35.0% and 44.1%, respectively) but high specificity (94.6% and 78.6%, respectively).

CONCLUSIONS

Our finding of increased kurtosis of quantitative susceptibility mapping values (heavy-tailed distribution) was highly specific, suggesting that it may be a useful biomarker to identify putative epileptogenic tubers in tuberous sclerosis complex. This finding motivates the investigation of underlying tuber mineralization and other properties driving kurtosis changes in quantitative susceptibility mapping values.

IDEAL approach to the evaluation of machine learning technology in epilepsy surgery: protocol for the MAST trial

Epilepsy and epilepsy surgery lend themselves well to the application of machine learning (ML) and artificial intelligence (AI) technologies. This is evidenced by the plethora of tools developed for applications such as seizure detection and analysis of imaging and electrophysiological data. However, few of these tools have been directly used to guide patient management. In recent years, the Idea, Development, Exploration, Assessment, Long-Term Follow-Up (IDEAL) collaboration has formalised stages for the evaluation of surgical innovation and medical devices, and, in many ways, this pragmatic framework is also applicable to ML/AI technology, balancing innovation and safety.

In this protocol paper, we outline the preclinical (IDEAL stage 0) evaluation and the protocol for a prospective (IDEAL stage 1/2a) study to evaluate the utility of an ML lesion detection algorithm designed to detect focal cortical dysplasia from structural MRI, as an adjunct in the planning of stereoelectroencephalography trajectories in children undergoing intracranial evaluation for drug-resistant epilepsy.

Planning stereoelectroencephalography using automated lesion detection: Retrospective feasibility study

Objective

This retrospective, cross‐sectional study evaluated the feasibility and potential benefits of incorporating deep‐learning on structural magnetic resonance imaging (MRI) into planning stereoelectroencephalography (sEEG) implantation in pediatric patients with diagnostically complex drug‐resistant epilepsy. This study aimed to assess the degree of colocalization between automated lesion detection and the seizure onset zone (SOZ) as assessed by sEEG.

Methods

A neural network classifier was applied to cortical features from MRI data from three cohorts. (1) The network was trained and cross‐validated using 34 patients with visible focal cortical dysplasias (FCDs). (2) Specificity was assessed in 20 pediatric healthy controls. (3) Feasibility of incorporation into sEEG implantation plans was evaluated in 34 sEEG patients. Coordinates of sEEG contacts were coregistered with classifier‐predicted lesions. sEEG contacts in seizure onset and irritative tissue were identified by clinical neurophysiologists. A distance of <10 mm between SOZ contacts and classifier‐predicted lesions was considered colocalization.

Results

In patients with radiologically defined lesions, classifier sensitivity was 74% (25/34 lesions detected). No clusters were detected in the controls (specificity = 100%). Of the total 34 sEEG patients, 21 patients had a focal cortical SOZ, of whom eight were histopathologically confirmed as having an FCD. The algorithm correctly detected seven of eight of these FCDs (86%). In patients with histopathologically heterogeneous focal cortical lesions, there was colocalization between classifier output and SOZ contacts in 62%. In three patients, the electroclinical profile was indicative of focal epilepsy, but no SOZ was localized on sEEG. In these patients, the classifier identified additional abnormalities that had not been implanted.

Significance

There was a high degree of colocalization between automated lesion detection and sEEG. We have created a framework for incorporation of deep‐learning–based MRI lesion detection into sEEG implantation planning. Our findings support the prospective evaluation of automated MRI analysis to plan optimal electrode trajectories.

Altered White-Matter Microstructure in Conduct Disorder Is Specifically Associated with Elevated Callous-Unemotional Traits

Adolescents with conduct disorder (CD) and elevated callous-unemotional (CU) traits have been reported to present with a more severe and persistent pattern of antisocial behaviour than those with low levels of CU traits. However, relatively few studies have investigated whether there are differences in brain structure between these subgroups.We acquired diffusion tensor imaging data and used tract-based spatial statistics (TBSS) to compare adolescents with CD and high levels of CU traits (CD/CU+; n = 18, CD and low levels of CU traits (CD/CU-; n = 17) and healthy controls (HC; n = 32) on measures of fractional anisotropy (FA), axial (AD), radial (RD) and mean (MD) diffusivity. Compared to CD/CU- adolescents, those with CD/CU+ presented increased FA and reduced RD and MD (lower diffusivity) in several tracts including: body and splenium of the corpus callosum, right inferior longitudinal fasciculus, ILF; right inferior fronto-occipital fasciculus, IFOF; left superior longitudinal fasciculus, SLF; left cerebral peduncle, bilateral internal capsule, left superior and posterior corona radiata, bilateral thalamic radiation and left external capsule. In addition, relative to CD/CU- individuals, adolescents with CD/CU+ showed lower diffusivity (indexed by reduced RD and MD) in left uncinate fasciculus and bilateral fornix. Finally, relative to healthy controls, CD/CU+ individuals showed lower diffusivity (reduced RD) in the genu and body of the corpus callosum and left anterior corona radiata. These results suggest that CD/CU+ individuals present with white-matter microstructural abnormalities compared to both CD/CU- individuals and age-matched healthy controls. This finding is consistent with emerging evidence suggesting that CD/CU+ represents a distinct subtype of CD, and illustrates the importance of accounting for heterogeneity within CD populations.

Natural biological variation of white matter microstructure is accentuated in Huntington's disease

Huntington's disease (HD) is a monogenic neurodegenerative disorder caused by a CAG‐repeat expansion in the Huntingtin gene. Presence of this expansion signifies certainty of disease onset, but only partly explains age at which onset occurs. Genome‐wide association studies have shown that naturally occurring genetic variability influences HD pathogenesis and disease onset. Investigating the influence of biological traits in the normal population, such as variability in white matter properties, on HD pathogenesis could provide a complementary approach to understanding disease modification. We have previously shown that while white matter diffusivity patterns in the left sensorimotor network were similar in controls and HD gene‐carriers, they were more extreme in the HD group. We hypothesized that the influence of natural variation in diffusivity on effects of HD pathogenesis on white matter is not limited to the sensorimotor network but extends to cognitive, limbic, and visual networks. Using tractography, we investigated 32 bilateral pathways within HD‐related networks, including motor, cognitive, and limbic, and examined diffusivity metrics using principal components analysis. We identified three independent patterns of diffusivity common to controls and HD gene‐carriers that predicted HD status. The first pattern involved almost all tracts, the second was limited to sensorimotor tracts, and the third encompassed cognitive network tracts. Each diffusivity pattern was associated with network specific performance. The consistency in diffusivity patterns across both groups coupled with their association with disease status and task performance indicates that naturally‐occurring patterns of diffusivity can become accentuated in the presence of the HD gene mutation to influence clinical brain function.

Impact of induction chemotherapy, hyperfractionated accelerated radiotherapy and high-dose thiotepa on brain volume loss and functional status of children with primitive neuroectodermal tumour

BACKGROUND

The introduction of aggressive chemo-radiotherapy regimens has improved overall survival in children with primitive neuroectodermal tumours (PNET). However, these combinations may result in neurotoxicity. Previously reported magnetic resonance imaging abnormalities in children receiving intensive sequential chemotherapy, hyperfractionated accelerated radiotherapy (HART) and high-dose thiotepa prompted us to investigate the degree of brain volume loss and patients' functional status after therapy.

METHODS

We retrospectively reviewed clinico-radiological data of children with PNET treated in this way at our centre.

RESULTS

We studied 14 children treated between December 2009 and April 2013. Data were not complete for one child. Performance status was severely restricted in four children, and mildly to moderately impaired in 7 of the 13 children. Eleven of 13 children showed mild-to-severe generalised neuroparenchymal atrophy, in 7 of whom neuroparenchymal volume loss was moderate to severe. Of these seven, six had received high-dose thiotepa. There was no correlation between brain volume loss and Lansky performance status. However, unexpected neurotoxicities, such as symptoms of transverse myelitis, were observed.

CONCLUSION

Measurement of brain volume loss in patients treated with HART and high-dose thiotepa may not be sufficient to predict function. However, correlation of brain volume loss due to late neurotoxicity with performance decline may be more obvious over longer period of follow-up. The combination of HART and myeloablative courses of thiotepa is associated with severe neurotoxicity and subsequent decline in performance status in a significant proportion of patients.

Prolonged febrile seizures cause reversible reductions in white matter integrity

Prolonged febrile seizures (PFS) are the commonest cause of childhood status epilepticus and are believed to carry a risk of neuronal damage, in particular to the mesial temporal lobe. This study was designed to determine: i) the effect of prolonged febrile seizures on white matter and ii) the temporal evolution of any changes seen.

33 children were recruited 1 month following PFS and underwent diffusion tensor imaging (DTI) with repeat imaging at 6 and 12 months after the original episode of PFS. 18 age-matched healthy control subjects underwent similar investigations at a single time point. Tract-based spatial statistics (TBSS) was used to compare fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD) and radial diffusivity (RD) between patients and controls on a voxel-wise basis within the white matter skeleton.

Widespread reductions in FA along multiple white matter tracts were found at 1 and 6 months post-PFS, but these had resolved at 12 months. At one month post-PFS the main changes seen were reductions in AD but at 6 months these had predominantly changed to increases in RD.

These widespread white matter changes have not previously been noted following PFS. There are many possible explanations, but one plausible hypothesis is that this represents a temporary halting of normal white matter development caused by the seizure, that then resumes and normalises in the majority of children.

•

Widespread reductions in FA occur in children after prolonged febrile seizures.

•

These reductions persist up to 6 months post-PFS but resolve by 1 year.

•

This may represent a seizure-related disruption of white matter development.

The CONNECT project: Combining macro- and micro-structure

In recent years, diffusion MRI has become an extremely important tool for studying the morphology of living brain tissue, as it provides unique insights into both its macrostructure and microstructure. Recent applications of diffusion MRI aimed to characterize the structural connectome using tractography to infer connectivity between brain regions. In parallel to the development of tractography, additional diffusion MRI based frameworks (CHARMED, AxCaliber, ActiveAx) were developed enabling the extraction of a multitude of micro-structural parameters (axon diameter distribution, mean axonal diameter and axonal density). This unique insight into both tissue microstructure and connectivity has enormous potential value in understanding the structure and organization of the brain as well as providing unique insights to abnormalities that underpin disease states. The CONNECT (Consortium Of Neuroimagers for the Non-invasive Exploration of brain Connectivity and Tracts) project aimed to combine tractography and micro-structural measures of the living human brain in order to obtain a better estimate of the connectome, while also striving to extend validation of these measurements. This paper summarizes the project and describes the perspective of using micro-structural measures to study the connectome.

A hierarchical response of cells to perpendicular micro- and nanometric textural cues

In this paper, we report on the influence of shallow micro- and nanopatterned substrata on the attachment and behavior of a human fibroblast [human telomerase transfected immortalized (hTERT)] cells. We identify a hierarchy of textural guidance cues with respect to cell alignment on these substrates. Cells were seeded and cultured for 48 h on silicon substrates patterned with two linear textures overlaid at 90 degrees, both with 24 microm pitch: a micrograting and a nanopattern of rows of 140- nm-diameter pits arranged in a rectangular array with 300 nm centre-to-centre spacing. The cell response to these textures was shown to be highly dependent on textural feature dimensions. We show that cells align to the stripes of nanopits. Stripes of 30-nm deep nanopits were also shown to elicit a stronger response from cells than 160-nm deep nanopits.

3D polymer scaffolds for tissue engineering

This review discusses some of the most common polymer scaffold fabrication techniques used for tissue engineering applications. Although the field of scaffold fabrication is now well established and advancing at a fast rate, more progress remains to be made, especially in engineering small diameter blood vessels and providing scaffolds that can support deep tissue structures. With this in mind, we introduce two new lithographic methods that we expect to go some way to addressing this problem.