Enhancing decision-making in surgery for a large temporocorneal meningioma through an explainable human-AI collaboration: a case report.
Authors
Affiliations (6)
Affiliations (6)
- Department of Neurosurgery, Faculty of Medicine, Université Libre de Bruxelles & Cliniques Universitaires de Bruxelles, Brussels, Belgium. [email protected].
- Faculty of Medicine, Neurosurgery Division, Catholic University of Graben, Butembo, Democratic Republic of the Congo. [email protected].
- Department of Surgery, Neurosurgery, College of Medicine, Makerere University, Kampala, Uganda. [email protected].
- Department of Neurosurgery, New Deal Medical Ngaliema Medical Center and C.I.E. Medical/Kinshasa & Clinique International de Médecine Avancée au Kivu (CIMAK), Goma, Democratic Republic of the Congo. [email protected].
- Department of Neurosurgery, Faculty of Medicine, Université Libre de Bruxelles & Cliniques Universitaires de Bruxelles, Brussels, Belgium.
- Service de Neurochirurgie, Cliniques Universitaires de Kinshasa, Kinshasa, Democratic Republic of the Congo.
Abstract
Meningiomas, particularly large temporocorneal meningiomas, pose significant surgical challenges due to their proximity to critical brain structures. Achieving optimal tumor resection while minimizing neurological deficits requires advanced decision-making strategies. This case report explores the integration of an explainable artificial intelligence (AI) system into the neurosurgical workflow to enhance preoperative planning, intraoperative decision-making, and postoperative outcome prediction. MAIN SYMPTOMS AND CLINICAL FINDINGS: We report the case of a 48-year-old Algerian Arab female with a one-year history of right-lateralized headaches that became generalized over time, along with episodes of loss of consciousness lasting 15-45 minutes, occurring 3-8 times daily. These episodes were characterized by a prodrome of palpitations and chest tightness, followed by transient unresponsiveness, urinary incontinence, and prolonged postictal periods. Neurological examination was unremarkable, with preserved motor and sensory functions. Initial evaluation in Algeria led to a diagnosis of epilepsy, for which the patient was prescribed multiple antiepileptic drugs. Further assessment in Belgium, including MRI and electroencephalogram (EEG), revealed a right temporal extra-axial mass (46 × 36 × 45 mm) consistent with meningioma. EEG findings were normal, suggesting psychogenic non-epileptic seizures (PNES) rather than epileptic seizures. A multidisciplinary approach, incorporating AI-driven imaging analysis and predictive modeling, was employed to optimize surgical strategies. The AI system provided insights into tumor segmentation, vascular involvement, and risk assessment, aiding in determining the safest resection trajectory. The patient underwent surgical resection of the tumor via a right pterional craniotomy, with total excision achieved with preserved neurological function. Intraoperative bleeding was significant (2 L), but the postoperative course was favorable. Antiepileptic medication withdrawal was initiated, and no recurrent seizures were reported postoperatively. This case demonstrates that explainable AI can enhance preoperative planning and surgical confidence by improving visualization and risk anticipation. However, its role remains supportive, as surgical outcomes continue to depend primarily on tumor characteristics and surgical expertise. The report also highlights the importance of accurate differentiation between PNES and epilepsy in patients with intracranial tumors. Overall, AI should be considered a complementary decision-support tool rather than a determinant of clinical outcomes.