Probing artificial intelligence in neurosurgical training: ChatGPT takes a neurosurgical residents written exam

Introduction

Artificial Intelligence tools are being introduced in almost every field of human life, including medical sciences and medical education, among scepticism and enthusiasm.

Research question

to assess how a generative language tool (Generative Pretrained Transformer 3.5, ChatGPT) performs at both generating questions and answering a neurosurgical residents' written exam. Namely, to assess how ChatGPT generates questions, how it answers human-generated questions, how residents answer AI-generated questions and how AI answers its self-generated question.

Materials and methods

50 questions were included in the written exam, 46 questions were generated by humans (senior staff members) and 4 were generated by ChatGPT. 11 participants took the exam (ChatGPT and 10 residents). Questions were both open-ended and multiple-choice.8 questions were not submitted to ChatGPT since they contained images or schematic drawings to interpret.

Results

formulating requests to ChatGPT required an iterative process to precise both questions and answers. Chat GPT scored among the lowest ranks (9/11) among all the participants). There was no difference in response rate for residents' between human-generated vs AI-generated questions that could have been attributed to less clarity of the question. ChatGPT answered correctly to all its self-generated questions.

Discussion and conclusions

AI is a promising and powerful tool for medical education and for specific medical purposes, which need to be further determined. To request AI to generate logical and sound questions, that request must be formulated as precise as possible, framing the content, the type of question and its correct answers.

Modulation of epileptic networks by transient interictal epileptic activity: A dynamic approach to simultaneous EEG-fMRI

•

EEG-fMRI has been instrumental in characterizing brain networks in epilepsy.

•

Its value is documented in the pre-surgical assessment of drug-resistant epilepsy.

•

The delineation of brain areas to resect is fundamental for the post-surgical outcome.

•

Standard EEG-fMRI in epilepsy assesses static functional connectivity of the network.

•

EEG-fMRI dynamic connectivity identifies transitory features of specific connections.

•

We integrate dynamic fMRI connectivity and dynamic patterns of simultaneous scalp EEG.

•

This allows to better characterize the spatiotemporal aspects of epileptic networks.

•

This may help in more efficiently target the surgical intervention.

Epileptic networks, defined as brain regions involved in epileptic brain activity, have been mapped by functional connectivity in simultaneous electroencephalography and functional magnetic resonance imaging (EEG-fMRI) recordings. This technique allows to define brain hemodynamic changes, measured by the Blood Oxygen Level Dependent (BOLD) signal, associated to the interictal epileptic discharges (IED), which together with ictal events constitute a signature of epileptic disease. Given the highly time-varying nature of epileptic activity, a dynamic functional connectivity (dFC) analysis of EEG-fMRI data appears particularly suitable, having the potential to identify transitory features of specific connections in epileptic networks. In the present study, we propose a novel method, defined dFC-EEG, that integrates dFC assessed by fMRI with the information recorded by simultaneous scalp EEG, in order to identify the connections characterised by a dynamic profile correlated with the occurrence of IED, forming the dynamic epileptic subnetwork. Ten patients with drug-resistant focal epilepsy were included, with different aetiology and showing a widespread (or multilobar) BOLD activation, defined as involving at least two distinct clusters, located in two different lobes and/or extended to the hemisphere contralateral to the epileptic focus. The epileptic focus was defined from the IED-related BOLD map. Regions involved in the occurrence of interictal epileptic activity; i.e., forming the epileptic network, were identified by a general linear model considering the timecourse of the fMRI-defined focus as main regressor. dFC between these regions was assessed with a sliding-window approach. dFC timecourses were then correlated with the sliding-window variance of the IED signal (VarIED), to identify connections whose dynamics related to the epileptic activity; i.e., the dynamic epileptic subnetwork. As expected, given the very different clinical picture of each individual, the extent of this subnetwork was highly variable across patients, but was but was reduced of at least 30% with respect to the initially identified epileptic network in 9/10 patients. The connections of the dynamic subnetwork were most commonly close to the epileptic focus, as reflected by the laterality index of the subnetwork connections, reported higher than the one within the original epileptic network. Moreover, the correlation between dFC timecourses and VarIED was predominantly positive, suggesting a strengthening of the dynamic subnetwork associated to the occurrence of IED. The integration of dFC and scalp IED offers a more specific description of the epileptic network, identifying connections strongly influenced by IED. These findings could be relevant in the pre-surgical evaluation for the resection or disconnection of the epileptogenic zone and help in reaching a better post-surgical outcome. This would be particularly important for patients characterised by a widespread pathological brain activity which challenges the surgical intervention.

A snapshot of European neurosurgery December 2019 vs. March 2020: just before and during the Covid-19 pandemic

Background

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2 or Covid-19), which began as an epidemic in China and spread globally as a pandemic, has necessitated resource management to meet emergency needs of Covid-19 patients and other emergent cases. We have conducted a survey to analyze caseload and measures to adapt indications for a perception of crisis.

Methods

We constructed a questionnaire to survey a snapshot of neurosurgical activity, resources, and indications during 1 week with usual activity in December 2019 and 1 week during SARS-CoV-2 pandemic in March 2020. The questionnaire was sent to 34 neurosurgical departments in Europe; 25 departments returned responses within 5 days.

Results

We found unexpectedly large differences in resources and indications already before the pandemic. Differences were also large in how much practice and resources changed during the pandemic. Neurosurgical beds and neuro-intensive care beds were significantly decreased from December 2019 to March 2020. The utilization of resources decreased via less demand for care of brain injuries and subarachnoid hemorrhage, postponing surgery and changed surgical indications as a method of rationing resources. Twenty departments (80%) reduced activity extensively, and the same proportion stated that they were no longer able to provide care according to legitimate medical needs.

Conclusion

Neurosurgical centers responded swiftly and effectively to a sudden decrease of neurosurgical capacity due to relocation of resources to pandemic care. The pandemic led to rationing of neurosurgical care in 80% of responding centers. We saw a relation between resources before the pandemic and ability to uphold neurosurgical services. The observation of extensive differences of available beds provided an opportunity to show how resources that had been restricted already under normal conditions translated to rationing of care that may not be acceptable to the public of seemingly affluent European countries.

Electronic supplementary material

The online version of this article (10.1007/s00701-020-04482-8) contains supplementary material, which is available to authorized users.

The network integration of epileptic activity in relation to surgical outcome

OBJECTIVE

Epilepsy is a network disease with epileptic activity and cognitive impairment involving large-scale brain networks. A complex network is involved in the seizure and in the interictal epileptiform discharges (IEDs). Directed connectivity analysis, describing the information transfer between brain regions, and graph analysis are applied to high-density EEG to characterise networks.

METHODS

We analysed 19 patients with focal epilepsy who had high-density EEG containing IED and underwent surgery. We estimated cortical activity during IED using electric source analysis in 72 atlas-based cortical regions of the individual brain MRI. We applied directed connectivity analysis (information Partial Directed Coherence) and graph analysis on these sources and compared patients with good vs poor post-operative outcome at global, hemispheric and lobar level.

RESULTS

We found lower network integration reflected by global, hemispheric, lobar efficiency during the IED (p < 0.05) in patients with good post-surgical outcome, compared to patients with poor outcome. Prediction was better than using the IED field or the localisation obtained by electric source imaging.

CONCLUSIONS

Abnormal network patterns in epilepsy are related to seizure outcome after surgery.

SIGNIFICANCE

Our finding may help understand networks related to a more "isolated" epileptic activity, limiting the extent of the epileptic network in patients with subsequent good post-operative outcome.

Spontaneous appearance of de novo intracranial arteriovenous malformation in hepatic cirrhosis

BACKGROUND

Intracranial arteriovenous malformations (AVMs) are rare lesions that can be congenital or acquired in early childhood, with fatal outcome in approximately 30% of cases. De novo formation during adulthood without established predisposing vascular pathology or previous brain insult is even less frequent.

CASE DESCRIPTION

We present a case of de novo brain AVM in an alcoholic Child-B cirrhosis setting. Thirty previously reported cases presented de novo AVM in patients of all ages that had another previous brain pathology or insult, such as AVM resection. Seventeen of those cases occurred in adult patients, with only 2 showing no significant predisposing factor. The present pathophysiological review covers and completes Mullan's hemodynamic "two-hit" model, associating probable thrombotic predisposition to AVM with brain insult triggering a later stage based on angiogenic stimuli.

CONCLUSIONS

This case report and literature review renews previously discussed hemodynamic theories and contributes to a fuller understanding of the pathogenesis and progression of AVM. We postulate a causal link between hepatopathy and de novo AVM, which should be strengthened and interpreted based on recent genetic data and future prospective studies.

[E-Cigarettes: Assessment of Health Effects and Potential Benefits for Smokers]

E-cigarettes are electronic devices used to vaporize a flavored liquid resulting in aerosols to be inhaled by the user. The high variety of e-cigarettes and liquids makes it difficult to assess the health effects of e-cigarette use. The ingredients of the liquids - apart from nicotine - are generally recognized as safe for oral and dermal use, however, toxicological data on health effects upon chronic inhalation are missing. Studies have detected several harmful substances in e-cigarette aerosol, including carcinogens, but in lower concentrations than tobacco smoke. Evidence suggests that e-cigarette use entails acute health effects such as inflammatory responses in the lung. In animal and in vitro studies e-cigarette aerosol showed adverse effects like cytotoxicity, increased infiltration of inflammatory cells and pro-inflammatory cytokines, increased oxidative stress and moderate inflammation. Compared to tobacco smoking e-cigarettes are very likely to be less harmful, however, due to scarce data the extent of harm reduction cannot be quantified. Although smokers often use e-cigarettes to quit smoking, currently they are not approved as smoking cessation aids. As e-cigarette aerosol is a new source of air pollution, it is reasonable to assume that it poses an increased health risk to bystanders, in particular to vulnerable persons such as children and people with health conditions. E-cigarettes can have a positive effect on public health, if many smokers completely switch from smoking tobacco to e-cigarettes and, finally, also stop e-cigarette use, and if at the same time adverse consequences (such as increased initiation among youths) remain limited.

Correlation of pain, functional impairment, and health-related quality of life with radiological grading scales of lumbar degenerative disc disease

BACKGROUND

It is generally believed that radiological signs of lumbar degenerative disc disease (DDD) are associated with increased pain and functional impairment as well as lower health-related quality of life (HRQoL). Our aim was to assess the association of the Modic and Pfirrmann grading scales with established outcome questionnaires and the timed-up-and-go (TUG) test.

METHODS

In a prospective two-center study with patients scheduled for lumbar spine surgery, visual analogue scale (VAS) for back and leg pain, Roland-Morris Disability Index, Oswestry Disability Index and HRQoL, as determined by the Short-Form (SF)-12 and the Euro-Qol, were recorded. Functional mobility was measured with the TUG test. Modic type (MOD) and Pfirrmann grade (PFI) of the affected lumbar segment were assessed with preoperative imaging. Uni- and multivariate logistic regression analysis was performed to estimate the effect size of the relationship between clinical and radiological findings.

RESULTS

Two hundred eighty-four patients (mean age 58.5, 119 (42 %) females) were enrolled. None of the radiological grading scales were significantly associated with any of the subjective or objective clinical tests. There was a tendency for higher VAS back pain (3.48 vs. 4.14, p = 0.096) and lower SF-12 physical component scale (31.2 vs. 29.4, p = 0.065) in patients with high PFI (4-5) as compared to patients with low PFI (0-3). In the multivariate analysis, patients with MOD changes of the vertebral endplates were 100 % as likely as patients without changes to show an impaired TUG test performance (odds ratio (OR) 1.00, 95 % confidence interval (CI) 0.56-1.80, p = 0.982). Patients with high PFI were 145 % as likely as those with low PFI to show an impaired TUG test performance (OR 1.45, 95 % CI 0.79-2.66, p = 0.230).

CONCLUSIONS

There was no association between established outcome questionnaires of symptom severity and two widely used radiological classifications in patients undergoing surgery for lumbar DDD.

[Minimally invasive spine surgery: past and present]

In the early twentieth century, the understanding of spine biomechanics and the advent of surgical techniques of the lumbar spine, led to the currently emerging concept of minimal invasive spine surgery, By reducing surgical access, blood loss, infection rate and general morbidity, functional prognosis of patients is improved. This is a real challenge for the spine surgeon, who has to maintain a good operative result by significantly reducing surgical collateral damages due to the relatively traumatic conventional access.

[Cervical disc herniation--diagnosis and treatment]

A cervical disc herniation (CDH) is a frequently encountered pathology in primary care medicine. It may give rise to a compression of a nerve root (a radiculopathy, with or without sensory-motor deficit) or of the spinal cord (myelopathy). The majority of CDHs can be supported by means of a conservative treatment. When a radiculopathy is found and a clinico-radiological correlation is present, a moderate neurological deficit appears suddenly, or if it is progressive under conservative treatment or if pain is poorly controlled by well-conducted conservative treatment performed during 6 to 8 months, surgery is then recommended. A symptomatic cervical myelopathy is, by itself, an indication for a surgical treatment.

Improving workflows of neuro-interventional procedures with autostereoscopic 3D visualization of multi-modality imaging in hybrid interventional suites

PURPOSE

Recent developments in interventional neuroradiology techniques, medical imaging modalities, endovascular stenting and embolization materials lead to an increasing number of patients with cerebral aneurysms and arteriovenous malformations that are eligible for endovascular treatment and have opened new perspectives for novel ways for patient treatment in general. In this paper, we describe a software tool for 3D image fusion of multi-modal acquisitions to assist endovascular treatment of cerebral malformations. The software and an autostereoscopic 3D display were implemented and tested in clinical applications in a hybrid interventional suite that is used for radio-interventional as well as neurosurgical procedures. Our hypothesis is that fusion of image data acquired prior to intervention procedures with images acquired during those procedures should allow better visualizing and navigating through complex cerebral vasculature. This should also improve workflows of neuro-interventional procedures.

METHODS

Preoperative and intra-operative acquisitions of vascular images of the brain were performed and transferred to a dedicated imaging workstation to be processed with our image fusion and visualization software tool. The tool was developed as a plugin extension to the open-source DICOM viewer OsiriX and is based on a modular and scalable architecture. Several processing modules were implemented to allow spatial co-registration and fusion of preoperative and intra-operative modalities. A special extension was also implemented for interactive autostereosopic, glass-free 3D visualization of fused results.

RESULTS

The software platform was validated and evaluated in nine in vivo procedures by expert users. All patient cases were related to interventional treatment of neuro-vascular diseases. The emphasis was laid on the added value of spatial co-registration and fusion of preoperative and intra-operative modalities, as well as the overall impact on workflow during the intervention. The co-registered and fused images were visualized on an autostereoscopic 3D monitor installed in hybrid interventional suite. All experiments were evaluated and scored by interventional physicians and technicians.

CONCLUSIONS

Displaying 3D-4D representations of brain vascular anomalies based on multi-modal acquisitions on a 3D autostereoscopic display is beneficial for the workflow and efficiency of interventional radiologists. The implemented software tool fulfills the premise of applicability of an open-source platform for more advanced, multi-modal visualization and processing of brain vascular structures for image-guided therapeutic interventions.

[Lumbar disc herniation--diagnosis and treatment]

A lumbar disc herniation (LDH) is a condition frequently encountered in primary care medicine. It may give rise to a compression of one or more nerve roots, which can lead to a nerve root irritation, a so-called radiculopathy, with or without a sensorimotor deficit. The majority of LDHs can be supported by means of a conservative treatment consisting of physical therapy, ergotherapy, analgetics, anti-inflammatory therapy or corticosteroids, which may be eventually administered by infiltrations. If a clinico-radiological correlation is present and moderate neurological deficit appears suddenly, if it is progressive under conservative treatment or if pain is poorly controlled by well-conducted conservative treatment performed during four to six months, surgery is then recommended.

Intra-aneurysmal flow patterns: illustrative comparison among digital subtraction angiography, optical flow, and computational fluid dynamics

BACKGROUND AND PURPOSE

Digital subtraction angiography is the gold standard vascular imaging and it is used for all endovascular treatment of intracranial anerysms. Optical flow imaging has been described as a potential method to evaluate cerebral hemodynamics through DSA. In this study, we aimed to compare the flow patterns measured during angiography, by using an optical flow method, with those measured by using computational fluid dynamics in intracranial aneurysms.

MATERIALS AND METHODS

A consecutive series of 21 patients harboring unruptured saccular intracranial aneurysms who underwent diagnostic angiography before treatment was considered. High-frame-rate digital subtraction angiography was performed to obtain an intra-aneurysmal velocity field by following the cardiac-modulated contrast wave through the vascular structures by using optical flow principles. Additionally, computational fluid dynamics modeling was performed for every case by using patient-specific inlet-boundary conditions measured with the optical flow method from both DSA and 3D rotational angiography datasets. Three independent observers compared qualitatively both the inflow direction and the apparent recirculation in regular DSA, optical flow images, and computational fluid dynamics flow patterns for each patient; κ statistics were estimated.

RESULTS

We included 21 patients. In 14 of these 21, the flow patterns were conclusive and matching between the optical flow images and computational fluid dynamics within the same projection view (κ = .91). However, in only 8 of these 14 patients the optical flow images were conclusive and matching regular DSA images (observer κ = 0.87). In 7 of the 21 patients, the flow patterns in the optical flow images were inconclusive, possibly due to improper projection angles.

CONCLUSIONS

The DSA-based optical flow technique was considered qualitatively consistent with computational fluid dynamics outcomes in evaluating intra-aneurysmal inflow direction and apparent recirculation. Moreover, the optical flow technique may provide the premises for new solutions for improving the visibility of flow patterns when contrast motion in DSA is not apparent. This technique is a diagnostic method to evaluate intra-aneurysmal flow patterns and could be used in the future for validation and patient evaluation.

[Cell transplantation: current treatments and future prospects]

Regenerative medicine aims to replace a body function or specific cell loss. It includes therapies at the forefront of modem medicine, issuing from translational biomedical research. Transplantation of organs and cells has revolutionized the management of patients for whom medical treatment is a failure. Unfortunately, organ shortage is limiting treatment possibility. As an example, among the 15,000 patients with type I diabetes in Switzerland, only approximately 30 can receive a pancreas or an islet transplant per year. Second example, 500 patients die each year in Switzerland from alcoholic cirrhosis because no treatment is available. Transplantation of islet cells, hepatocytes, mesenchymal stem cells or dopaminergic neurons represents hope fora therapy available for large populations of patients.

Clinically relevant complications related to posterior atlanto-axial fixation in atlanto-axial instability and their management

BACKGROUND

The Magerl transarticular technique and the Harms-Goel C1 lateral mass-C2 isthmic screw technique are the two most commonly used surgical procedures to achieve fusion at C1-C2 level for atlanto-axial instability. Despite recent technological advances with an increased safety, several complications may still occur, including vascular lesions, neurological injuries, pain at the harvested bone graft site, infections, and metallic device failure.

METHODS

We retrospectively analyzed all patients (n=42 cases) undergoing a Harms-Goel C1-C2 fixation surgery with polyaxial C1 lateral mass screws and C2 isthmic screws at two different institutions between 2003 and 2012 and report clinical and radiological complications. One patient was lost to follow-up. The mean follow-up of the remaining 41 patients was 18.7 months (range 12-90). A clinically relevant complication was defined as a complication determining the onset of a new neurological deficit or requiring the need for a revision surgery.

RESULTS

A total of 14 complications occurred in 10 patients (24.4% of 41 patients). Greater occipital nerve neuralgia was evident in 4 patients (9.8%). All but one completely resolved at the end of the follow-up. Persistent neck pain was reported by 3 patients (7.3%), hypoesthesia by 1 patient (2.4%), and anesthesia in the C2 area on both sides in 1 patient (2.4%). Furthermore, a superficial, a deep, and a combined superficial and deep wound infection occurred in 1 patient each (2.4%). One patient (2.4%) had pain at the iliac bone graft donor site for several weeks with spontaneous resolution. A posterior progressive intestinal herniation through the iliac scar was seen in 1 case (2.4%), which required surgical repair. No vascular damages occurred. Altogether, 5/41 patients (12.2%) had a clinically relevant complication including 4 patients necessitating a revision surgery at the C1-C2 level (9.8%).

CONCLUSIONS

Atlanto-axial fixation surgery remains a challenging procedure because of the proximity of important neurovascular structures. Nevertheless, on the basis of our current experience, the C1 lateral mass-C2 isthmic screw technique appears to be safe with a low incidence of clinically relevant complications. Postoperative C2 neuralgia, as the most frequent problem, is due to surgical manipulation during preparation of the C1 screw entry point.

Analysis of the expression of nine secreted matrix metalloproteinases and their endogenous inhibitors in the brain of mice subjected to ischaemic stroke

Matrix metalloproteinases (MMPs) are a family of more than twenty secreted and cell-surface endopeptidases. Among them, MMP2, MMP3 and MMP9 are involved in blood-brain barrier injury and neuronal death after cerebral ischaemia. On the other hand, very little is known about the expression of the other secreted MMPs. Herein, we compared the global changes in MMP1, MMP2, MMP3, MMP7, MMP8, MMP9, MMP10, MMP12 and MMP13, and their endogenous inhibitors TIMP1 and TIMP2, both at the mRNA and protein levels, during the hyperacute (6 h), acute (24 h) and subacute (72 h) stages following transient focal cerebral ischaemia and treatment with recombinant tissue plasminogen activator (rtPA). We observed a significant increase in MMP1, MMP2, MMP9, MMP10, MMP13 and TIMP1 levels during the acute stage of reperfusion, which was further amplified during the subacute stage for MMP1, MMP2, MMP10 and TIMP1. In general, no change of MMP3, MMP7, MMP8, MMP12 and TIMP2 was observed. However, rtPA treatment induced a rapid increase in MMP1/TIMP2, MMP2/TIMP2, MMP8/TIMP2 and MMP9/TIMP2 ratios during the hyperacute stage of reperfusion compared to saline treatment, which may have potential implications in the early disruption of the blood-brain barrier after rtPA treatment.