Adverse intraoperative events during surgical repair of ruptured cerebral aneurysms: a systematic review

A Comparative Analysis of AI Models in Complex Medical Decision-Making Scenarios: Evaluating ChatGPT, Claude AI, Bard, and Perplexity

This study rigorously evaluates the performance of four artificial intelligence (AI) language models - ChatGPT, Claude AI, Google Bard, and Perplexity AI - across four key metrics: accuracy, relevance, clarity, and completeness. We used a strong mix of research methods, getting opinions from 14 scenarios. This helped us make sure our findings were accurate and dependable. The study showed that Claude AI performs better than others because it gives complete responses. Its average score was 3.64 for relevance and 3.43 for completeness compared to other AI tools. ChatGPT always did well, and Google Bard had unclear responses, which varied greatly, making it difficult to understand it, so there was no consistency in Google Bard. These results give important information about what AI language models are doing well or not for medical suggestions. They help us use them better, telling us how to improve future tech changes that use AI. The study shows that AI abilities match complex medical scenarios.

A synthetic model simulator for intracranial aneurysm clipping: validation of the UpSurgeOn AneurysmBox

Background and objectives

In recent decades, the rise of endovascular management of aneurysms has led to a significant decline in operative training for surgical aneurysm clipping. Simulation has the potential to bridge this gap and benchtop synthetic simulators aim to combine the best of both anatomical realism and haptic feedback. The aim of this study was to validate a synthetic benchtop simulator for aneurysm clipping (AneurysmBox, UpSurgeOn).

Methods

Expert and novice surgeons from multiple neurosurgical centres were asked to clip a terminal internal carotid artery aneurysm using the AneurysmBox. Face and content validity were evaluated using Likert scales by asking experts to complete a post-task questionnaire. Construct validity was evaluated by comparing expert and novice performance using the modified Objective Structured Assessment of Technical Skills (mOSATS), developing a curriculum-derived assessment of Specific Technical Skills (STS), and measuring the forces exerted using a force-sensitive glove.

Results

Ten experts and eighteen novices completed the task. Most experts agreed that the brain looked realistic (8/10), but far fewer agreed that the brain felt realistic (2/10). Half the expert participants (5/10) agreed that the aneurysm clip application task was realistic. When compared to novices, experts had a significantly higher median mOSATS (27 vs. 14.5; p < 0.01) and STS score (18 vs. 9; p < 0.01); the STS score was strongly correlated with the previously validated mOSATS score (p < 0.01). Overall, there was a trend towards experts exerting a lower median force than novices, however, these differences were not statistically significant (3.8 N vs. 4.0 N; p = 0.77). Suggested improvements for the model included reduced stiffness and the addition of cerebrospinal fluid (CSF) and arachnoid mater.

Conclusion

At present, the AneurysmBox has equivocal face and content validity, and future versions may benefit from materials that allow for improved haptic feedback. Nonetheless, it has good construct validity, suggesting it is a promising adjunct to training.

Shifted-windows transformers for the detection of cerebral aneurysms in microsurgery

Purpose

Microsurgical Aneurysm Clipping Surgery (MACS) carries a high risk for intraoperative aneurysm rupture. Automated recognition of instances when the aneurysm is exposed in the surgical video would be a valuable reference point for neuronavigation, indicating phase transitioning and more importantly designating moments of high risk for rupture. This article introduces the MACS dataset containing 16 surgical videos with frame-level expert annotations and proposes a learning methodology for surgical scene understanding identifying video frames with the aneurysm present in the operating microscope’s field-of-view.

Methods

Despite the dataset imbalance (80% no presence, 20% presence) and developed without explicit annotations, we demonstrate the applicability of Transformer-based deep learning architectures (MACSSwin-T, vidMACSSwin-T) to detect the aneurysm and classify MACS frames accordingly. We evaluate the proposed models in multiple-fold cross-validation experiments with independent sets and in an unseen set of 15 images against 10 human experts (neurosurgeons).

Results

Average (across folds) accuracy of 80.8% (range 78.5–82.4%) and 87.1% (range 85.1–91.3%) is obtained for the image- and video-level approach, respectively, demonstrating that the models effectively learn the classification task. Qualitative evaluation of the models’ class activation maps shows these to be localized on the aneurysm’s actual location. Depending on the decision threshold, MACSWin-T achieves 66.7–86.7% accuracy in the unseen images, compared to 82% of human raters, with moderate to strong correlation.

Conclusions

Proposed architectures show robust performance and with an adjusted threshold promoting detection of the underrepresented (aneurysm presence) class, comparable to human expert accuracy. Our work represents the first step towards landmark detection in MACS with the aim to inform surgical teams to attend to high-risk moments, taking precautionary measures to avoid rupturing.

Adverse events associated with microsurgial treatment for ruptured intracerebral aneurysms: a prospective nationwide study on subarachnoid haemorrhage in Sweden

BACKGROUND

Adverse events (AEs) or complications may arise secondary to the treatment of aneurysmal subarachnoid haemorrhage (SAH). The aim of this study was to identify AEs associated with microsurgical occlusion of ruptured aneurysms, as well as to analyse their risk factors and impact on functional outcome.

METHODS

Patients with aneurysmal SAH admitted to the neurosurgical centres in Sweden were prospectively registered during a 3.5-year period (2014-2018). AEs were categorised as intraoperative or postoperative. A range of variables from patient history and SAH characteristics were explored as potential risk factors for an AE. Functional outcome was assessed approximately 1 year after the bleeding using the extended Glasgow Outcome Scale.

RESULTS

In total, 1037 patients were treated for ruptured aneurysms, of which, 322 patients were treated with microsurgery. There were 105 surgical AEs in 97 patients (30%); 94 were intraoperative AEs in 79 patients (25%). Aneurysm rerupture occurred in 43 patients (13%), temporary occlusion of the parent artery >5 min in 26 patients (8%) and adjacent vessel injury in 25 patients (8%). High Fisher grade and brain oedema on CT were related to increased risk of AEs. At follow-up, 38% of patients had unfavourable outcome. Patients suffering AEs were more likely to have unfavourable outcome (OR 2.3, 95% CI 1.10 to 4.69).

CONCLUSION

Intraoperative AEs occurred in 25% of patients treated with microsurgery for ruptured intracerebral aneurysm in this nationwide survey. Although most operated patients had favourable outcome, AEs were associated with increased risk of unfavourable outcome.

Robotic Arm-Protected Microsurgical Pericallosal and Middle Cerebral Artery Aneurysm Clipping: A Technical Note and Case Series

BACKGROUND

Managing intraoperative aneurysm rupture (IAR) during intracranial aneurysm clipping can be challenging given the excessive hemorrhage and limited field of view under the microscope for visualizing the proximal artery and safe temporary clipping.

OBJECTIVE

To describe the first known use of robotic arm for safeguarding IAR in microsurgical aneurysm clipping.

METHODS

A robotic arm was used to safeguard 3 microsurgical clipping cases (1 pericallosal and 2 middle cerebral artery) performed by a single surgeon. The device was installed onto the side rail of the operating table along with the clip applier attachment. After dissecting the cerebral artery segment proximal to the aneurysm, a temporary aneurysm clip was loaded and established at the appropriate segment before dissecting distally toward the aneurysm.

RESULTS

Setup for the robotic arm and temporary clip was simple, quick, precise, and without any unforeseen accommodations needed in all 3 instances. The temporary clip acted as an emergency gate and could be deployed either manually or remotely through a controller. IAR occurred in case 1, and the robotic-assisted temporary clip deployment achieved immediate hemostasis without complications. This method bypassed the need for significant suctioning, packing, and further exploration for safe temporary clipping. Case 2 and 3 demonstrated the feasibility for middle cerebral artery protection and ease of intraoperative readjustment.

CONCLUSION

This technical note highlights the feasibility and relative ease of using a robotic arm as a safeguard device, and it enables on-demand control of proximal blood flow and may enhance the safety of microsurgical aneurysm procedures.

Microsurgery for intracranial aneurysms: A qualitative survey on technical challenges and technological solutions

Introduction

Microsurgery for the clipping of intracranial aneurysms remains a technically challenging and high-risk area of neurosurgery. We aimed to describe the technical challenges of aneurysm surgery, and the scope for technological innovations to overcome these barriers from the perspective of practising neurovascular surgeons.

Materials and Methods

Consultant neurovascular surgeons and members of the British Neurovascular Group (BNVG) were electronically invited to participate in an online survey regarding surgery for both ruptured and unruptured aneurysms. The free text survey asked three questions: what do they consider to be the principal technical barriers to aneurysm clipping? What technological advances have previously contributed to improving the safety and efficacy of aneurysm clipping? What technological advances do they anticipate improving the safety and efficacy of aneurysm clipping in the future? A qualitative synthesis of responses was performed using multi-rater emergent thematic analysis.

Results

The most significant reported historical advances in aneurysm surgery fell into five themes: (1) optimising clip placement, (2) minimising brain retraction, (3) tissue handling, (4) visualisation and orientation, and (5) management of intraoperative rupture. The most frequently reported innovation by far was indocyanine green angiography (84% of respondents). The three most commonly cited future advances were hybrid surgical and endovascular techniques, advances in intraoperative imaging, and patient-specific simulation and planning.

Conclusions

While some surgeons perceive that the rate of innovation in aneurysm clipping has been dwarfed in recent years by endovascular techniques, surgeons surveyed highlighted a broad range of future technologies that have the potential to continue to improve the safety of aneurysm surgery in the future.

Plasma Amino Acids May Improve Prediction Accuracy of Cerebral Vasospasm after Aneurysmal Subarachnoid Haemorrhage

Aneurysmal subarachnoid haemorrhages (aSAH) account for 5% of strokes and continues to place a great burden on patients and their families. Cerebral vasospasm (CVS) is one of the main causes of death after aSAH, and is usually diagnosed between day 3 and 14 after bleeding. Its pathogenesis remains poorly understood. To verify whether plasma concentration of amino acids have prognostic value in predicting CVS, we analysed data from 35 patients after aSAH (median age 55 years, IQR 39-62; 20 females, 57.1%), and 37 healthy volunteers (median age 50 years, IQR 38-56; 19 females, 51.4%). Fasting peripheral blood samples were collected on postoperative day one and seven. High performance liquid chromatography-mass spectrometry (HPLC-MS) analysis was performed. The results showed that plasma from patients after aSAH featured a distinctive amino acids concentration which was presented in both principal component analysis and direct comparison. No significant differences were noted between postoperative day one and seven. A total of 18 patients from the study group (51.4%) developed CVS. Hydroxyproline (AUC = 0.7042, 95%CI 0.5259-0.8826, p = 0.0248) and phenylalanine (AUC = 0.6944, 95%CI 0.5119-0.877, p = 0.0368) presented significant CVS prediction potential. Combining the Hunt-Hess Scale and plasma levels of hydroxyproline and phenylalanine provided the model with the best predictive performance and the lowest leave-one-out cross-validation of performance error. Our results suggest that plasma amino acids may improve sensitivity and specificity of Hunt-Hess scale in predicting CVS.