Search Results (16)

Background/Objectives: The integration of Artificial Intelligence (AI) into clinical workflows raises critical questions regarding decision-making responsibility, as fully autonomous systems inevitably carry a margin of error that can be fatal in high-stakes fields like surgery. This study addresses this challenge by evaluating a “Human-in-the-Loop” (HITL) workflow, using intraoperative Optical Coherence Tomography (OCT) for glioma detection. We aimed to determine if integrating Machine Learning (ML)-generated segmentation maps with human contextual analysis resolves the tension between automation and clinical responsibility, yielding superior diagnostic reliability compared to structural or quantitative imaging alone. Methods: We retrospectively analyzed 86 intraoperative OCT scans from 27 patients. Five neurosurgeons blindly assessed the data across three progressive levels of processing: (1) structural scans, (2) physics-based parametric maps, and (3) SVM-based generated segmentation maps. Crucially, the HITL inference performance on segmentation maps was benchmarked against “models-only” inference pipeline: a SVM and a state-of-the-art multimodal reasoning model, Gemini 3.1 Pro. To evaluate interpretability and the operator’s ability to confidently exercise their authority, we measured inter-rater consistency alongside diagnostic performance. Results: The results demonstrate that, while quantitative parametric maps improved Global Accuracy (87% [95% CI: 82–92%]) compared to structural scans (80% [95% CI: 73–86%]), they suffered from an “interpretability gap,” resulting in a moderate inter-rater consistency of 0.68 [95% CI: 0.59–0.78]. In contrast, the HITL approach using segmentation maps maximized consensus to 0.98 [95% CI: 0.95–1.00] and achieved the highest performance (Accuracy 94% [95% CI: 88–98%] and Sensitivity 98% [95% CI: 92–100%]). Compared to the standalone models, the HITL approach significantly outperformed the SVM baseline (Accuracy 84% [95% CI: 81–87%]; Sensitivity 83% [95% CI: 78–88%]). Furthermore, it surpassed the SOTA Gemini 3.1 Pro model (Accuracy 90% [95% CI: 83–95%]; Sensitivity 86% [95% CI: 74–95%]). While the HITL sensitivity demonstrated a definitive and statistically significant edge over the Gemini model, the accuracy improvement fell just slightly short of undisputed statistical significance due to overlapping confidence intervals. Conclusions: By utilizing their clinical domain knowledge of tumor invasion patterns and topological priors, surgeons effectively filtered algorithmic noise—overriding ML errors in 69% (9 out of 13) false positive cases that models alone could not resolve. This demonstrates exactly how and where HITL optimally utilizes human contextual intelligence to outperform autonomous “models-only” pipelines, confirming a human-ML synergy that augments the objectivity of machine learning with human domain knowledge. This paradigm ensures that the ultimate responsibility for diagnostic inference remains safely and practically in human hands. Open Data Initiative: To ensure essential reproducibility, enable independent multi-center validation and support open science, all examples of intraoperative in vivo OCT brain scans used in this study are made publicly available. To the best of our knowledge, this represents the first open-access data of its kind globally. Full article

Background: Stereotactic brain biopsy is the gold standard for the pathological diagnosis of malignant brain tumors. However, conventional procedures rely heavily on manual path planning and unintuitive navigation, which significantly increase the risk of severe complications and impose an additional cognitive burden on surgeons. Methods: We propose an augmented reality-based navigation system that synergizes multi-objective path planning with hybrid registration. Preoperatively, the system utilizes a constrained multi-objective optimization (MOO) model derived from clinical criteria to automatically calculate and visualize optimal biopsy paths within a three-dimensional anatomical environment. Intraoperatively, the system performs rapid initial alignment using quick response (QR) codes, followed by precise refinement through anatomical landmarks. This process ultimately enables the highly accurate, real-time overlay of the surgical path and anatomical models onto the patient’s operative field. Results: An expert study across four common brain tumor locations demonstrated that the MOO model significantly outperformed manual methods in satisfying safety criteria. The hybrid registration reduced the mean fiducial registration error (FRE) from 4.19 ± 1.11 mm to 2.37 ± 0.91 mm (p < 0.001), with a mean target registration error (TRE) of 2.34 ± 0.71 mm and a mean clinical setup time of 2.63 ± 0.36 min. Conclusions: This system assists stereotactic brain biopsy through automated path planning and immersive augmented reality-based guidance, highlighting its potential to support surgical workflow consistency and procedural safety. Full article

Acquired tracheoesophageal fistulas (TEF) are a rare but severe complication in post-coma neurorehabilitation patients, particularly those requiring long-term tracheostomy and enteral nutrition. Early recognition and proper surgical management are critical to prevent life-threatening outcomes and functional deterioration. However, variability in clinical presentation and the lack of standardized multidisciplinary pathways often delay referral to thoracic surgeons. We present the case of a young patient with severe traumatic brain injury, prolonged tracheostomy, and percutaneous endoscopic gastrostomy (PEG), who developed a TEF due to tracheal ischemic injury. Clinical suspicion arose from indirect signs—such as recurrent aspiration and air in the PEG system—the diagnosis was confirmed by bronchoscopy and sagittal CT imaging. Surgical planning was carried out in close collaboration between rehabilitation physicians and thoracic surgeons, based on shared criteria involving ventilator weaning, nutritional status, and clinical stability. This case highlights the importance of a multidisciplinary, protocol-driven approach in managing TEF. Current literature supports timely but carefully selected surgical intervention, particularly in patients who are no longer ventilator-dependent, significantly reducing perioperative mortality (reported up to 60% in ventilated patients). Recent reviews advocate for standardized surgical techniques—such as single-stage repair with muscle flap interposition—and emphasize the value of early diagnosis using a combination of bronchoscopy, videofluoroscopy, and sagittal CT. We propose a structured clinical pathway integrating neurorehabilitation and thoracic surgery, aimed at optimizing timing and surgical outcomes in patients with acquired TEF. This model may serve as a foundation for future guidelines, improving both safety and efficiency in the multidisciplinary management of this complex complication. Full article

Background: Different lateral and postero-lateral cranial approaches to the petroclival region and to the mid-upper brain stem have been described so far, some of which require extensive osseous demolition and possible damage of neurovascular structures. Neuronavigational systems are now extensively available for preoperative planning and intraoperative navigation to assist the surgeons in choosing the optimally invasive approach for each pathology. Herein, we describe a detailed navigation-augmented morphometric analysis to bring insight into the usefulness of an anterior petrosectomy (AP) to handle lesions in the petroclival region. Methods: Eight cadaveric, silicone injected heads were used. A total of 14 approaches (AP, n = 8; retrosigmoid, RS, n = 6) using a standard microsurgical dissection technique were performed. All specimens had preoperative CT and MRI scans, as well as a post-dissection CT. The neuronavigational system was used for distance measurements, craniotomy sizes and surgical corridor volumes, for each approach. Results: The distance from the skull surface to the petrous apex was significantly shorter in the AP approach when compared with the RS (46.0 ± 1.9 mm versus 71.3 ± 1.8 mm, respectively, p < 0.001). Although the craniotomy size was not different, the volume of the surgical corridor was significantly larger with the AP approach (21.31 ± 1.91 cm³ vs. 13.39 ± 1.8 cm³). The AP approach increased the length of the basilar artery exposure from 6.9 ± 1.5 mm (obtained with a standard subtemporal approach) to 22.1 ± 1.7 mm (p < 0.0001). Conclusions: The surgical corridor to the petroclival region achieved by virtue of an AP was significantly larger and featured shorter working distances, resulting in a higher degree of surgical freedom. Although significant individual anatomical variations of fundamental neurovascular and bony structures were found, these difficulties were overcome by careful pre- and intraoperative use of neuronavigation. Full article

Introduction: A supraorbital keyhole craniotomy (SOKC) is a novel alternative to frontal craniotomies for accessing the anterior fossa for resecting tumors and clipping aneurysms; however, its implementation is limited in patients at a high risk of complications. We present two cases involving the use of augmented reality (AR) and virtual reality (VR) for patient selection and preoperative planning for a supraorbital tumor resection of anterior fossa meningiomas. Methods: This is a prospective, single-center case series at a research institute. We identified patients with an anterior or middle fossa meningioma regardless of age, gender, and tumor characteristics who could undergo an SOKC and MRI. The preoperative planning was performed with the BrainLab Magic Leap AR/VR platform. The meningiomas were resected through the SOKC under neuronavigation. Results: We identified two cases: a 37-year-old male with a meningioma in the sellar region and an 84-year-old male with a right anterior fossa meningioma, both confirmed by MRI. Both patients had a complete tumor resection by a minimally invasive SOKC after preoperative planning with the AR/VR platform. Postoperatively, hyponatremia complicated the first case, while the second case developed an intracranial hemorrhage. They both recovered after the appropriate interventions. Conclusions: The use of an SOKC for anterior skull base meningiomas should be individualized after considering the lesion characteristics, vascular control needs, and the surgeon’s expertise. VR/AR-assisted preoperative evaluation and planning will optimize the patient selection and surgical outcomes. We can utilize VR/AR technologies to identify patients that will benefit from an SOKC and expand the implementation of the approach beyond its current limitations. Full article

Background: Neurosurgery demands exceptional precision due to the brain’s complex and delicate structures, necessitating precise targeting of pathological targets. Achieving optimal outcomes depends on the surgeon’s ability to accurately differentiate between healthy and pathological tissues during operations. Raman spectroscopy (RS) has emerged as a promising innovation, offering real-time, in vivo non-invasive biochemical tissue characterization. This literature review evaluates the current research on RS applications in intraoperative neurosurgery, emphasizing its potential to enhance surgical precision and patient outcomes. Methods: Following PRISMA guidelines, a comprehensive systematic review was conducted using PubMed to extract relevant peer-reviewed articles. The inclusion criteria focused on original research discussing real-time RS applications with human tissue samples in or near the operating room, excluding retrospective studies, reviews, non-human research, and other non-relevant publications. Results: Our findings demonstrate that RS significantly improves tumor margin delineation, with handheld devices achieving high sensitivity and specificity. Stimulated Raman Histology (SRH) provides rapid, high-resolution tissue images comparable to traditional histopathology but with reduced time to diagnosis. Additionally, RS shows promise in identifying tumor types and grades, aiding precise surgical decision-making. RS techniques have been particularly beneficial in enhancing the accuracy of glioma surgeries, where distinguishing between tumor and healthy tissue is critical. By providing real-time molecular data, RS aids neurosurgeons in maximizing the extent of resection (EOR) while minimizing damage to normal brain tissue, potentially improving patient outcomes and reducing recurrence rates. Conclusions: This review underscores the transformative potential of RS in neurosurgery, advocating for continued innovation and research to fully realize its benefits. Despite its substantial potential, further research is needed to validate RS’s clinical utility and cost-effectiveness. Full article

Background and Objectives: As brain lesions present complex diagnostic challenges, accurate tissue sampling via biopsy is critical for effective treatment planning. Traditional frame-based stereotactic biopsy has been complemented by navigated biopsy techniques, leveraging advancements in imaging and navigation technology. This study aims to compare the navigated and frame-based stereotactic biopsy methods in a clinical setting, evaluating their efficacy, safety, and diagnostic outcomes to determine the optimal approach for precise brain lesion targeting. Materials and Methods: retrospective analysis was conducted on patients who underwent brain biopsies between January 2017 and August 2023 at an academic medical center. Data on patient demographics, clinical characteristics, biopsy technique (navigated vs. frame-based), and outcomes including accuracy, complications, and hospital stay duration were analyzed. Results: The cohort comprised 112 patients, with no significant age or gender differences between groups. Symptoms leading to biopsy were predominantly diminished muscle strength (42.0%), cognitive issues (28.6%), and aphasia (24.1%). Tumors were most common in the deep hemisphere (24.1%). The median hospital stay was 5 days, with a rehospitalization rate of 27.7%. Complications occurred in 4.47% of patients, showing no significant difference between biopsy methods. However, navigated biopsies resulted in fewer samples (p < 0.001) but with comparable diagnostic accuracy as frame-based biopsies. Conclusions: Navigated and frame-based stereotactic biopsies are both effective and safe, with comparable accuracy and complication rates. The choice of technique should consider lesion specifics, surgeon preference, and technological availability. The findings highlight the importance of advanced neurosurgical techniques in enhancing patient care and outcomes. Full article

Background: Meckel’s cave is a challenging surgical target due to its deep location and proximity to vital neurovascular structures. Surgeons have developed various microsurgical transcranial approaches (MTAs) to access it, but there is no consensus on the best method. Newer endoscopic approaches have also emerged. This study seeks to quantitatively compare these surgical approaches to Meckel’s cave, offering insights into surgical volumes and exposure areas. Methods: Fifteen surgical approaches were performed bilaterally in six specimens, including the pterional approach (PTA), fronto-temporal-orbito-zygomatic approach (FTOZA), subtemporal approach (STA), Kawase approach (KWA), retrosigmoid approach (RSA), retrosigmoid approach with suprameatal extension (RSAS), endoscopic endonasal transpterygoid approach (EETPA), inferolateral transorbital approach (ILTEA) and superior eyelid approach (SEYA). All the MTAs were performed both with 10 mm and 15 mm of brain retraction, to consider different percentages of surface exposure. A dedicated navigation system was used to quantify the surgical working volumes and exposure of different areas of Meckel’s cave (ApproachViewer, part of GTx-Eyes II, University Health Network, Toronto, Canada). Microsurgical transcranial approaches were quantified with two different degrees of brain retraction (10 mm and 15 mm). Statistical analysis was performed using a mixed linear model with bootstrap resampling. Results: The RSAS with 15 mm of retraction offered the maximum exposure of the trigeminal stem (TS). If compared to the KWA, the RSA exposed more of the TS (69% vs. 46%; p = 0.01). The EETPA and ILTEA exposed the Gasserian ganglion (GG) mainly in the anteromedial portion, but with a significant 20% gain in exposure provided by the EETPA compared to ILTEA (42% vs. 22%; p = 0.06). The STA with 15 mm of retraction offered the maximum exposure of the GG, with a significant gain in exposure compared to the STA with 10 mm of retraction (50% vs. 35%; p = 0.03). The medial part of the three trigeminal branches was mainly exposed by the EETPA, particularly for the ophthalmic (66%) and maxillary (83%) nerves. The EETPA offered the maximum exposure of the medial part of the mandibular nerve, with a significant gain in exposure compared to the ILTEA (42% vs. 11%; p = 0.01) and the SEY (42% vs. 2%; p = 0.01). The FTOZA offered the maximum exposure of the lateral part of the ophthalmic nerve, with a significant gain of 67% (p = 0.03) and 48% (p = 0.04) in exposure compared to the PTA and STA, respectively. The STA with 15 mm of retraction offered the maximum exposure of the lateral part of the maxillary nerve, with a significant gain in exposure compared to the STA with 10 mm of retraction (58% vs. 45%; p = 0.04). The STA with 15 mm of retraction provided a significant exposure gain of 23% for the lateral part of the mandibular nerve compared to FTOZA with 15 mm of retraction (p = 0.03). Conclusions: The endoscopic approaches, through the endonasal and transorbital routes, can provide adequate exposure of Meckel’s cave, especially for its more medial portions, bypassing the impediment of major neurovascular structures and significant brain retraction. As far as the most lateral portion of Meckel’s cave, MTA approaches still seem to be the gold standard in obtaining optimal exposure and adequate surgical volumes. Full article

Brain tumors are nonlinear and present with variations in their size, form, and textural variation; this might make it difficult to diagnose them and perform surgical excision using magnetic resonance imaging (MRI) scans. The procedures that are currently available are conducted by radiologists, brain surgeons, and clinical specialists. Studying brain MRIs is laborious, error-prone, and time-consuming, but they nonetheless show high positional accuracy in the case of brain cells. The proposed convolutional neural network model, an existing blockchain-based method, is used to secure the network for the precise prediction of brain tumors, such as pituitary tumors, meningioma tumors, and glioma tumors. MRI scans of the brain are first put into pre-trained deep models after being normalized in a fixed dimension. These structures are altered at each layer, increasing their security and safety. To guard against potential layer deletions, modification attacks, and tempering, each layer has an additional block that stores specific information. Multiple blocks are used to store information, including blocks related to each layer, cloud ledger blocks kept in cloud storage, and ledger blocks connected to the network. Later, the features are retrieved, merged, and optimized utilizing a Genetic Algorithm and have attained a competitive performance compared with the state-of-the-art (SOTA) methods using different ML classifiers. Full article

Volterra models allow modeling nonlinear dynamical systems, even though they require the estimation of a large number of parameters and have, consequently, potentially large computational costs. The pruning of Volterra models is thus of fundamental importance to reduce the computational costs of nonlinear calibration, and improve stability and speed, while preserving accuracy. Several techniques (LASSO, DOMP and OBS) and their variants (WLASSO and OBD) are compared in this paper for the experimental calibration of an IF amplifier. The results show that Volterra models can be simplified, yielding models that are 4–5 times sparser, with a limited impact on accuracy. About 6 dB of improved Error Vector Magnitude (EVM) is obtained, improving the dynamic range of the amplifiers. The Symbol Error Rate (SER) is greatly reduced by calibration at a large input power, and pruning reduces the model complexity without hindering SER. Hence, pruning allows improving the dynamic range of the amplifier, with almost an order of magnitude reduction in model complexity. We propose the OBS technique, used in the neural network field, in conjunction with the better known DOMP technique, to prune the model with the best accuracy. The simulations show, in fact, that the OBS and DOMP techniques outperform the others, and OBD, LASSO and WLASSO are, in turn, less efficient. A methodology for pruning in the complex domain is described, based on the Frisch–Waugh–Lovell (FWL) theorem, to separate the linear and nonlinear sections of the model. This is essential because linear models are used for equalization and cannot be pruned to preserve model generality vis-a-vis channel variations, whereas nonlinear models must be pruned as much as possible to minimize the computational overhead. This methodology can be extended to models other than the Volterra one, as the only conditions we impose on the nonlinear model are that it is feedforward and linear in the parameters. Full article

Precise intraoperative brain tumor visualization supports surgeons in achieving maximal safe resection. In this sense, improved prognosis in patients with high-grade gliomas undergoing protoporphyrin IX fluorescence-guided surgery has been demonstrated. Phase fluorescence lifetime imaging in the frequency-domain has shown promise to distinguish weak protoporphyrin IX fluorescence from competing endogenous tissue fluorophores, thus allowing for brain tumor detection with high sensitivity. In this work, we show that this technique can be further improved by minimizing the crosstalk of autofluorescence signal contributions when only detecting the fluorescence emission above 615 nm. Combining fluorescence lifetime and spectroscopic measurements on a set of 130 ex vivo brain tumor specimens (14 low- and 56 high-grade gliomas, 39 meningiomas and 21 metastases) coherently substantiated the resulting increase of the fluorescence lifetime with respect to the detection band employed in previous work. This is of major interest for obtaining a clear-cut distinction from the autofluorescence background of the physiological brain. In particular, the median fluorescence lifetime of low- and high-grade glioma specimens lacking visual fluorescence during surgical resection was increased from 4.7 ns to 5.4 ns and 2.9 ns to 3.3 ns, respectively. While more data are needed to create statistical evidence, the coherence of what was observed throughout all tumor groups emphasized that this optimization should be taken into account for future studies. Full article

The intracranial measurement of local cerebral tissue oxygen levels—PbtO₂—has become a useful tool for the critical care unit to investigate severe trauma and ischemia injury in patients. Our preliminary work in animal models supports the hypothesis that multi-site depth electrode recording of PbtO₂ may give surgeons and critical care providers needed information about brain viability and the capacity for better recovery. Here, we present a surface morphology characterization and an electrochemical evaluation of the analytical properties toward oxygen detection of an FDA-approved, commercially available, clinical grade depth recording electrode comprising 12 Pt recording contacts. We found that the surface of the recording sites is composed of a thin film of smooth Pt and that the electrochemical behavior evaluated by cyclic voltammetry in acidic and neutral electrolyte is typical of polycrystalline Pt surface. The smoothness of the Pt surface was further corroborated by determination of the electrochemical active surface, confirming a roughness factor of 0.9. At an optimal working potential of −0.6 V vs. Ag/AgCl, the sensor displayed suitable values of sensitivity and limit of detection for in vivo PbtO₂ measurements. Based on the reported catalytical properties of Pt toward the electroreduction reaction of O₂, we propose that these probes could be repurposed for multisite monitoring of PbtO₂ in vivo in the human brain. Full article

In brain tumor surgery, an appropriate and careful surgical planning process is crucial for surgeons and can determine the success or failure of the surgery. A deep comprehension of spatial relationships between tumor borders and surrounding healthy tissues enables accurate surgical planning that leads to the identification of the optimal and patient-specific surgical strategy. A physical replica of the region of interest is a valuable aid for preoperative planning and simulation, allowing the physician to directly handle the patient’s anatomy and easily study the volumes involved in the surgery. In the literature, different anatomical models, produced with 3D technologies, are reported and several methodologies were proposed. Many of them share the idea that the employment of 3D printing technologies to produce anatomical models can be introduced into standard clinical practice since 3D printing is now considered to be a mature technology. Therefore, the main aim of the paper is to take into account the literature best practices and to describe the current workflow and methodology used to standardize the pre-operative virtual and physical simulation in neurosurgery. The main aim is also to introduce these practices and standards to neurosurgeons and clinical engineers interested in learning and implementing cost-effective in-house preoperative surgical planning processes. To assess the validity of the proposed scheme, four clinical cases of preoperative planning of brain cancer surgery are reported and discussed. Our preliminary results showed that the proposed methodology can be applied effectively in the neurosurgical clinical practice both in terms of affordability and in terms of simulation realism and efficacy. Full article

The main goal of brain cancer surgery is to perform an accurate resection of the tumor, preserving as much normal brain tissue as possible for the patient. The development of a non-contact and label-free method to provide reliable support for tumor resection in real-time during neurosurgical procedures is a current clinical need. Hyperspectral imaging is a non-contact, non-ionizing, and label-free imaging modality that can assist surgeons during this challenging task without using any contrast agent. In this work, we present a deep learning-based framework for processing hyperspectral images of in vivo human brain tissue. The proposed framework was evaluated by our human image database, which includes 26 in vivo hyperspectral cubes from 16 different patients, among which 258,810 pixels were labeled. The proposed framework is able to generate a thematic map where the parenchymal area of the brain is delineated and the location of the tumor is identified, providing guidance to the operating surgeon for a successful and precise tumor resection. The deep learning pipeline achieves an overall accuracy of 80% for multiclass classification, improving the results obtained with traditional support vector machine (SVM)-based approaches. In addition, an aid visualization system is presented, where the final thematic map can be adjusted by the operating surgeon to find the optimal classification threshold for the current situation during the surgical procedure. Full article

Glioblastoma (GBM) is the most common primary intracranial neoplasia, and is characterized by its extremely poor prognosis. Despite maximum surgery, chemotherapy, and radiation, the histological heterogeneity of GBM makes total eradication impossible, due to residual cancer cells invading the parenchyma, which is not otherwise seen in radiographic images. Even with gross total resection, the heterogeneity and the dormant nature of brain tumor initiating cells allow for therapeutic evasion, contributing to its recurrence and malignant progression, and severely impacting survival. Visual delimitation of the tumor’s margins with common surgical techniques is a challenge faced by many surgeons. In an attempt to achieve optimal safe resection, advances in approaches allowing intraoperative analysis of cancer and non-cancer tissue have been developed and applied in humans resulting in improved outcomes. In addition, functional paradigms based on stimulation techniques to map the brain’s electrical activity have optimized glioma resection in eloquent areas such as the Broca’s, Wernike’s and perirolandic areas. In this review, we will elaborate on the current standard therapy for newly diagnosed and recurrent glioblastoma with a focus on surgical approaches. We will describe current technologies used for glioma resection, such as awake craniotomy, fluorescence guided surgery, laser interstitial thermal therapy and intraoperative mass spectrometry. Additionally, we will describe a newly developed tool that has shown promising results in preclinical experiments for brain cancer: optical coherence tomography. Full article