Artificial intelligence tools in clinical neuroradiology: essential medico-legal aspects.

Commercial software based on artificial intelligence (AI) is entering clinical practice in neuroradiology. Consequently, medico-legal aspects of using Software as a Medical Device (SaMD) become increasingly important. These medico-legal issues warrant an interdisciplinary approach and may affect the way we work in daily practice. In this article, we seek to address three major topics: medical malpractice liability, regulation of AI-based medical devices, and privacy protection in shared medical imaging data, thereby focusing on the legal frameworks of the European Union and the USA. As many of the presented concepts are very complex and, in part, remain yet unsolved, this article is not meant to be comprehensive but rather thought-provoking. The goal is to engage clinical neuroradiologists in the debate and equip them to actively shape these topics in the future.

Distended Seminal Vesicles Are Involved in Specific Cerebral Sexual Arousal: A Pilot Study Using Functional Brain Imaging in Young Healthy Men.

Background: Whether seminal vesicles play a role in sexual activity in men is unknown. No study so far has compared the neural processing of visual sexual stimuli in men depending on the filling state of the seminal vesicles. Objective: To evaluate potential specific cortical activation by visual sexual stimuli with distended and empty seminal vesicles. Design setting and participants: A prospective case-control trial was conducted. Six male individuals underwent two visits on 2 consecutive days for hormone analyses; Derogatis Interview for Sexual Functioning (DISF) questionnaire; functional magnetic resonance imaging (fMRI) with passively viewing sexual, neutral, positive, and negative emotional pictures; and structural pelvic MRI. After the first visit, the participants had to empty seminal vesicles by masturbation. During fMRI, every participant viewed alternating blocks of sexual, neutral, positive, and negative emotional pictures. Outcome measurements and statistical analysis: Comparisons between days 1 and 2 were evaluated using paired t tests. Results and limitations: No significant differences were observed regarding hormone analyses, DISF questionnaire score, and arousal scoring between days 1 and 2. Seminal vesicle volume was significantly lower on day 2 (p = 0.003). Significantly higher activation was observed in the right precentral gyrus, middle frontal gyrus, and right superior temporal sulcus when contrasted for sexual over neutral (p < 0.05). Conclusions: In response to pictures with sexual emotional content, significantly higher activation was detected in brain areas involved in motor preparation (arousal) and coding of desirability of visual sexual stimuli in men with distended seminal vesicles than in the same men with emptied seminal vesicles. This suggests that the filling state of the seminal vesicles may influence sexual desire in men. Patient summary: We compared brain activity of men with filled and emptied seminal vesicles by functional magnetic resonance imaging. We found that men with filled seminal vesicles had higher activation of brain areas involved in arousal and sexual desire.

Simultaneous lesion and brain segmentation in multiple sclerosis using deep neural networks.

Segmentation of white matter lesions and deep grey matter structures is an important task in the quantification of magnetic resonance imaging in multiple sclerosis. In this paper we explore segmentation solutions based on convolutional neural networks (CNNs) for providing fast, reliable segmentations of lesions and grey-matter structures in multi-modal MR imaging, and the performance of these methods when applied to out-of-centre data. We trained two state-of-the-art fully convolutional CNN architectures on the 2016 MSSEG training dataset, which was annotated by seven independent human raters: a reference implementation of a 3D Unet, and a more recently proposed 3D-to-2D architecture (DeepSCAN). We then retrained those methods on a larger dataset from a single centre, with and without labels for other brain structures. We quantified changes in performance owing to dataset shift, and changes in performance by adding the additional brain-structure labels. We also compared performance with freely available reference methods. Both fully-convolutional CNN methods substantially outperform other approaches in the literature when trained and evaluated in cross-validation on the MSSEG dataset, showing agreement with human raters in the range of human inter-rater variability. Both architectures showed drops in performance when trained on single-centre data and tested on the MSSEG dataset. When trained with the addition of weak anatomical labels derived from Freesurfer, the performance of the 3D Unet degraded, while the performance of the DeepSCAN net improved. Overall, the DeepSCAN network predicting both lesion and anatomical labels was the best-performing network examined.

Correction to: Constant severe imbalance following traumatic otoconial loss: a new explanation of residual dizziness.

The original article was updated.

The Influence of Various Cerebral and Extracerebral Pathologies on Apparent Diffusion Coefficient Values in the Fetal Brain.

BACKGROUND AND PURPOSE: The changing MRI signal accompanying brain maturation in fetal brains can be quantified on apparent diffusion coefficient (ADC) maps. Deviations from the natural course of ADC values may reflect structural pathology. The purpose of this study was to determine the influence of fetal pathologies on the ADC values in different regions of the fetal brain and their evolution with increasing gestational age. METHODS: This was a retrospective study of 291 fetuses evaluated between the 14th and the 40th week of gestation using diffusion-weighted imaging (DWI). Fetuses with normal MRI findings but sonographically suspected pathology or fetuses with abnormalities not affecting the brain were analyzed in the control group and compared to fetuses suffering from different pathologies like hydrocephalus/ventriculomegaly, brain malformations, infections, ischemia/hemorrhage, diaphragmatic hernias, and congenital heart disease. Pairwise ADC measurements in each side of the white matter (WM) of the frontal, parietal, and occipital lobes, in the basal ganglia and the cerebellum, as well as a single measurement in the pons were performed and were plotted against gestational age. RESULTS: In the control group, brain maturation followed a defined gradient, resulting in lower ADC values in the most mature regions. Each disorder group experienced abnormal patterns of evolution of the ADC values over time deviating from the expected course. CONCLUSIONS: The ADC values in different regions of the fetal brain and their evolution with increasing gestational age are influenced by pathologies compromising the cerebral maturation.

Constant severe imbalance following traumatic otoconial loss: a new explanation of residual dizziness.

Benign paroxysmal positional vertigo (BPPV) is the most common type of vertigo, caused by otoconia falling from the utricle into a semicircular canal (SCC). After successful repositioning maneuvers residual dizziness (RD) has been described and several reasons are used to explain RD. It can last for only a few days or weeks, but also much longer. We present a patient with a severe traumatic loss of otoconia from both maculae utriculi and a persistent imbalance more than 9 years. We think that the loss of otoconia from the utricular and probably also saccular macula induced a sudden reduction of her ability to sense gravity thus logically explaining her symptoms. We show the vestibular test results also supporting our hypothesis and we extrapolate this support to other forms of so far unexplained dizziness especially increasing imbalance with aging. We also discuss the normal c- and oVEMP indicating intact haircell function and supporting our hypothesis of isolated otoconial loss as the major cause for imbalance.

Automatic detection of lesion load change in Multiple Sclerosis using convolutional neural networks with segmentation confidence.

The detection of new or enlarged white-matter lesions is a vital task in the monitoring of patients undergoing disease-modifying treatment for multiple sclerosis. However, the definition of 'new or enlarged' is not fixed, and it is known that lesion-counting is highly subjective, with high degree of inter- and intra-rater variability. Automated methods for lesion quantification, if accurate enough, hold the potential to make the detection of new and enlarged lesions consistent and repeatable. However, the majority of lesion segmentation algorithms are not evaluated for their ability to separate radiologically progressive from radiologically stable patients, despite this being a pressing clinical use-case. In this paper, we explore the ability of a deep learning segmentation classifier to separate stable from progressive patients by lesion volume and lesion count, and find that neither measure provides a good separation. Instead, we propose a method for identifying lesion changes of high certainty, and establish on an internal dataset of longitudinal multiple sclerosis cases that this method is able to separate progressive from stable time-points with a very high level of discrimination (AUC = 0.999), while changes in lesion volume are much less able to perform this separation (AUC = 0.71). Validation of the method on two external datasets confirms that the method is able to generalize beyond the setting in which it was trained, achieving an accuracies of 75 % and 85 % in separating stable and progressive time-points.

Impact of stroke volume on motor outcome in neonatal arterial ischemic stroke.

BACKGROUND AND OBJECTIVES: Neonatal arterial ischemic stroke (NAIS) can lead to long-term neurological consequences such as cerebral palsy (CP). The aim of this study was to evaluate the predictive value of acute diffusion-weighted imaging (DWI) for CP by analyzing stroke volume next to brain structure involvement. METHODS: We included 37 term-born infants with NAIS prospectively registered in a nationwide pediatric stroke registry. DWI was performed between 0 and 8 days (mean 3 days) after stroke manifestation. Participants were neurologically assessed at the age of 2 years. We calculated the stroke volume (in mm3) and the ratio of the stroke volume to the volume of the entire brain (relative stroke volume). The predictive value of the relative stroke volume was analyzed and an optimal threshold for classification of children with high- and low-rates of CP was calculated. Predictive value of brain structure involvements and the prevalence of CP in combinations of different brain structures was also assessed. RESULTS: Sixteen children (43.2%) developed CP. Relative stroke volume significantly predicted CP (p < .001). Its optimal threshold for division into high- and low-rate of CP was 3.3%. The basal ganglia (OR 8.3, 95% CI 1.2-60.0) and basis pontis (OR 18.5, 95% CI 1.8-194.8) were independently associated with CP. CONCLUSION: In addition to determining the involvement of affected brain areas, the volumetric quantification of stroke volume allows accurate prediction of cerebral palsy in newborns with NAIS.

Mild Traumatic Brain Injury: Striking Postconcussion Symptoms Due to Inadequate Management.

Pediatric mild traumatic brain injury is a frequent cause for emergency consultations. Very often, management decisions focus only on acute neurological problems, without considering possible long-term impairment. Our case describes a 14-year-old girl who developed a pronounced and prolonged postconcussive syndrome and subsequent posttraumatic stress symptoms after (mild) traumatic brain injury. Early discharge without adequate instructions about the appropriate time to return to school and daily life activities promoted these striking problems. Only the delayed interventions including reduction of school workload and initiation of physiotherapy led to an improvement of symptoms. CONCLUSIONS: Traumatologists, pediatricians, and general practitioners should call families' and teachers' attention to the risk of potential postconcussive syndrome and advise them on appropriate coping strategies. Thorough clinical examination should rule out potentially treatable physical impairments. Prescription of physical and cognitive rest at an early stage is mandatory and should be part of concussion management already at emergency department. Pediatricians or general practitioners should follow up patients and support their gradually working back into full activity.

The effect of a single dose of escitalopram on sensorimotor networks.

INTRODUCTION: Serving as a pilot study of poststroke pharmacotherapy, the present investigation was intended to establish the effect of a single dose of escitalopram on motor task performance in normal volunteers. METHODS: Ten healthy volunteers of median age 63 years including four females performed a well-studied tactile manipulation task in two fMRI sessions using a double-blind cross-over design. The sessions began approximately three hours after ingestion of 20 mg escitalopram or placebo presented in pseudorandom order. The fMRI image sequences were submitted to principal component analysis (PCA). RESULTS: Based on volume correlations of task-related principal components with the mean component images derived in our previous study, we established the reproducibility of two networks of sensorimotor activity proposed there. The network reflecting motor control (cerebral pattern I) appeared invariably in placebo and verum conditions. In contrast, the other network, attributed to diminished motor control due to distracting mental processing (cerebral pattern II), emerged less regularly and exhibited more variability. Second-level PCAs of both conditions confirmed the findings of the initial analysis. Specifically, it validated the dominant and invariable expression of cerebral pattern I after application of a single dose of escitalopram. Dynamic causal modeling confirmed enhanced motor output as a result of a significantly increased connectivity between primary motor cortex and dorsal premotor cortex. CONCLUSION: This pilot study suggests the promise of stimulation by a specific serotonin reuptake inhibitor in regard to recovery and preservation of motor control after stroke.

Publisher Correction: Personalized structural image analysis in patients with temporal lobe epilepsy.

A correction to this article has been published and is linked from the HTML version of this paper. The error has been fixed in the paper.

Resting-state connectivity and executive functions after pediatric arterial ischemic stroke.

BACKGROUND: The aim of this study was to compare the relationship between core executive functions and frontoparietal network connections at rest between children who had suffered an arterial ischemic stroke and typically developing peers. METHODS: Children diagnosed with arterial ischemic stroke more than two years previously and typically developing controls were included. Executive function (EF) measures comprised inhibition (Go-NoGo task), fluency (category fluency task), processing speed (processing speed tasks), divided attention, working memory (letter-number sequencing), conceptual reasoning (matrices) and EF in everyday life (questionnaire). High-resolution T1-weighted magnetic resonance (MR) structural images and resting-state functional MR imaging were acquired. Independent component analysis was used to identify the frontoparietal network. Functional connections were obtained through correlation matrices; associations between cognitive measures and functional connections through Pearson's correlations. RESULTS: Twenty participants after stroke (7 females; mean age 16.0 years) and 22 controls (13 females; mean age 14.8 years) were examined. Patients and controls performed within the normal range in all executive tasks. Patients who had had a stroke performed significantly less well in tests of fluency, processing speed and conceptual reasoning than controls. Resting-state functional connectivity between the left and right inferior parietal lobe was significantly reduced in patients after pediatric stroke. Fluency, processing speed and perceptual reasoning correlated positively with the interhemispheric inferior parietal lobe connection in patients and controls. CONCLUSION: Decreased interhemispheric connections after stroke in childhood may indicate a disruption of typical interhemispheric interactions relating to executive functions. The present results emphasize the relationship between functional organization of the brain at rest and cognitive processes.

Decreased oxygen saturation levels in neonates with transposition of great arteries: Impact on appearance of cerebral veins in susceptibility-weighted imaging.

Purpose of this study was to investigate a potential correlation between the pattern of cerebral veins (CV) on susceptibility-weighted imaging (SWI) and blood oxygen saturation, as well as preoperative brain injury, in neonates with transposition of the great arteries (TGA). Eleven neonates with TGA underwent MRI preoperatively, including SWI, T1- and T2-weighted scans. Images were retrospectively evaluated and appearance of CV was graded from 0 (normal appearance) to 3 (severe prominent appearance). White matter injuries (WMI) and strokes were analysed. Results were correlated with preductal arterial oxygen saturation. As findings one subject showed a normal CV appearance (grade 0) whereas 10 showed pathological prominent CV (grades 1-3); median 2. Mean oxygen saturation ranged between 67.5% and 89.0% (median 81.0%). CV grade and mean oxygen saturation correlated significantly (p = 0.011). WMI were absent in 5 cases, mild in 4, and moderate in 2 cases. We conclude, that SWI has the potential to be used to estimate the current hypoxic burden on brain tissue in TGA newborns by assessing the prominence of the CV.

Personalized structural image analysis in patients with temporal lobe epilepsy.

Volumetric and morphometric studies have demonstrated structural abnormalities related to chronic epilepsies on a cohort- and population-based level. On a single-patient level, specific patterns of atrophy or cortical reorganization may be widespread and heterogeneous but represent potential targets for further personalized image analysis and surgical therapy. The goal of this study was to compare morphometric data analysis in 37 patients with temporal lobe epilepsies with expert-based image analysis, pre-informed by seizure semiology and ictal scalp EEG. Automated image analysis identified abnormalities exceeding expert-determined structural epileptogenic lesions in 86% of datasets. If EEG lateralization and expert MRI readings were congruent, automated analysis detected abnormalities consistent on a lobar and hemispheric level in 82% of datasets. However, in 25% of patients EEG lateralization and expert readings were inconsistent. Automated analysis localized to the site of resection in 60% of datasets in patients who underwent successful epilepsy surgery. Morphometric abnormalities beyond the mesiotemporal structures contributed to subtype characterisation. We conclude that subject-specific morphometric information is in agreement with expert image analysis and scalp EEG in the majority of cases. However, automated image analysis may provide non-invasive additional information in cases with equivocal radiological and neurophysiological findings.

Therapy-related longitudinal brain perfusion changes in patients with chronic pelvic pain syndrome.

BACKGROUND: The imaging method most frequently employed to identify brain areas involved in neuronal processing of nociception and brain pain perception is blood oxygenation level-dependent (BOLD) magnetic resonance imaging (MRI). Arterial spin labelling (ASL), in contrast, offers advantages when slow varying changes in brain function are investigated. Chronic pelvic pain syndrome (CPPS) is a disorder of, mostly, young males that leads to altered pain perceptions in structures related to the pelvis. We aimed to investigate the potential of ASL to monitor longitudinal cranial blood flow (CBF) changes in patients with CPPS. METHODS: In a randomised, placebo-controlled, double-blind single centre trial, we investigated treatment effects in CPPS after 12 weeks in patients that underwent sono-electro-magnetic therapy vs placebo. We investigated changes of CBF related to treatment outcome using pseudo-continuous arterial spin labelling (pCASL)-MRI. RESULTS: We observed CBF downregulation in the prefrontal cortex and anterior cingulate cortex and upregulation in the dorsolateral prefrontal cortex in responders. Nonresponders presented with CBF upregulation in the hippocampus. In patients with a history of CPPS of less than 12 months, there were significant correlations between longitudinal CBF changes and the Chronic Prostatitis Symptom Index pain subscore within the joint clusters anterior cingulate cortex and left anterior prefrontal cortex in responders, and the right hippocampus in nonresponders. CONCLUSIONS: We demonstrated therapy-related and stimulus-free longitudinal CBF changes in core areas of the pain matrix using ASL. ASL may act as a complementary noninvasive method to functional MRI and single-photon emission computed tomography / positron emission tomography, especially in the longitudinal assessment of pain response in clinical trials.

Seven-Tesla MRI of Hippocampal Sclerosis: An In Vivo Feasibility Study With Histological Correlations.

INTRODUCTION: Temporal lobe epilepsy (TLE) is the most frequent form of focal epilepsy in adults. Because approximately half of these patients develop drug resistance, epilepsy surgery designed to remove the epileptogenic zone is an excellent option in selected patients. Histopathological analyses of hippocampal specimens in TLE patients revealed 4 types of Ammon's horn sclerosis, which are correlated with long-term epileptological outcome. The aim of this study was the correlation of noninvasive, high-resolution, morphological magnetic resonance imaging (MRI) at an ultra-high-field (7 T) of the hippocampus in TLE patients with histopathological findings. METHODS: High-resolution, T2-weighted FSE MRI in 14 patients with drug-resistant temporal lobe epilepsy was performed on a 7 T Magnetom using a 32-channel coil. Four independent investigators assessed the delineation and semiquantitative evaluation of volume, signal intensity, internal architecture, and overall grading of the hippocampal subfields CA1-4, as well as the presence of the dentate granule cell layer (DGCL), on MRI scans. Results were compared with semiquantitative evaluation of neuronal loss and astrogliosis in the histological sections of the surgical specimens. RESULTS: Seven-tesla MR examinations were evaluable in 13 cases. Volume loss and signal intensity, as well as overall grading, showed a strong correlation between MRI and histology in individual CA regions. Furthermore, sensitivity and specificity values up to 100% were found for the detection of pathology in the CA subfields. The prediction of Ammon's horn sclerosis type was correct in up to 12 of 13 cases, whereas the dentate gyrus could not be delineated on MRI. DISCUSSION: High-resolution, ultra-high-field MRI is a promising tool for the detection of subtle changes in the hippocampus in patients with temporal lobe epilepsy. Large cohorts will be necessary to confirm the predictive value of 7 T MRI in the preoperative evaluation of TLE patients.

Enlarged posterior fossa on prenatal imaging: differential diagnosis, associated anomalies and postnatal outcome.

INTRODUCTION: The primary aim of this study was to ascertain the prevalence of the individual conditions and of associated anomalies in fetuses with the prenatal diagnosis of enlarged posterior fossa (PF) and to explore the diagnostic accuracy of ultrasound in these anomalies. The secondary aim was to evaluate the postnatal outcome of children affected by PF anomalies. MATERIAL AND METHODS: All fetuses with enlarged PF detected by prenatal sonography at a referral center from 2001 to 2015 were analyzed retrospectively. Some were also studied by fetal magnetic resonance imaging (MRI) or volume ultrasound examinations. Fetal sonographic and MRI were compared using following classification: Dandy-Walker malformation (DWM); megacisterna magna (MCM); Blake's pouch cyst; isolated vermian hypoplasia; vermian agenesis; PF arachnoid cyst (AC); and cerebellar hypoplasia (CH). RESULTS: The ultrasound diagnoses of the 69 fetuses were as follows: MCM (n = 29; of these isolated n = 15), DWM (n = 28, isolated n = 4), vermian hypoplasia (n = 5, isolated n = 4), Blake's pouch cyst (n = 4, isolated n = 1), CH (n = 2; none isolated) and AC in the PF (n = 1, isolated). Thirteen of the 41 karyotyped fetuses were aneuploid, including seven with DWM. Associated malformations were found in 37/69 cases. There were 39 live births, including 11 with confirmed DWM, six of whom show a normal development. Twelve infants with isolated MCM show normal development. There were eight false-positive prenatal diagnoses (or resolution until birth) of "enlarged PF": three with Blake's pouch cyst, two with MCM and one with vermian hypoplasia. CONCLUSIONS: An enlarged PF requires specific diagnoses for the best possible counseling. The term "Dandy-Walker variant" should not be used anymore. Isolated MCM and Blake's pouch cyst can either resolve or be normal variants, but may also indicate the presence of a more severe anomaly or associated malformations.

The neuronal correlates of mirror illusion in children with spastic hemiparesis: a study with functional magnetic resonance imaging.

AIM: To investigate the neuronal activation pattern underlying the effects of mirror illusion in children/adolescents with normal motor development and in children/adolescents with hemiparesis and preserved contralateral corticospinal organisation. METHOD: The type of cortical reorganisation was classified according to results of transcranial magnetic stimulation. Only subjects with congenital lesions and physiological contralateral cortical reorganisation were included. Functional magnetic resonance imaging was performed to investigate neuronal activation patterns with and without a mirror box. Each test consisted of a unimanual and a bimanual motor task. RESULTS: Seven children/adolescents with congenital hemiparesis (10-20 years old, three boys and four girls) and seven healthy subjects (8-17 years old, four boys and three girls) participated in this study. In the bimanual experiment, children with hemiparesis showed a significant effect of the mirror illusion (p<0.001 at voxel level, family-wise error corrected at cluster level) in the dorsolateral prefrontal cortex and anterior cingulate cortex of the affected and unaffected hemispheres, respectively. No significant effects of the mirror illusion were observed in unimanual experiments and in healthy participants. INTERPRETATION: Mirror illusion in children/adolescents with hemiparesis leads to activation of brain areas involved in visual conflict detection and cognitive control to resolve this conflict. This effect is observed only in bimanual training. We consider that for mirror therapy in children and adolescents with hemiparesis a bimanual approach is more suitable than a unimanual approach.

Population Statistics Approach for Safety Assessment in Robotic Cochlear Implantation.

HYPOTHESIS: Descriptive statistics with respect to patient anatomy and image guidance accuracy can be used to assess the effectiveness of any system for minimally invasive cochlear implantation, on both an individual patient and wider population level. BACKGROUND: Minimally invasive cochlear implantation involves the drilling of a tunnel from the surface of the mastoid to cochlea, with the trajectory passing through the facial recess. The facial recess anatomy constrains the drilling path and places prohibitive accuracy requirements on the used system. Existing single thresholds are insufficient for assessing the effectiveness of these systems. METHODS: A statistical model of the anatomical situation encountered during minimally invasive drilling of the mastoid for cochlear implantation was developed. A literature review was performed to determine the statistical distribution of facial recess width; these values were confirmed through facial recess measurements on computed tomography (CT) data. Based on the accuracy of a robotic system developed by the authors, the effect of variation of system accuracy, precision, and tunnel diameter examined with respect to the potential treatable portion of the population. RESULTS: A facial recess diameter of 2.54 ±â€Š0.51 mm (n = 74) was determined from a review of existing literature; subsequent measurements on CT data revealed a facial recess diameter of 2.54 ±â€Š0.5 mm (n = 23). The developed model demonstrated the effects of varying accuracy on the treatable portion of the population. CONCLUSIONS: The presented model allows the assessment of the applicability of a system on a wider population scale beyond examining only the system's ability to reach an arbitrary threshold accuracy.