Predurotomy intraoperative ultrasound in intradural extramedullary spinal tumor surgery: agreement with preoperative magnetic resonance imaging and correlation with microsurgical findings.

backgroundIntraoperative ultrasound (IOUS) has been widely used in spinal tumor surgery. However, this imaging modality is highly operator-dependent and the available evidence regarding its applications in this field is derived predominantly from descriptive studies, while studies systematically assessing its agreement with magnetic resonance imaging (MRI) or its correlation with microsurgical findings remain scarce. The present study provides an assessment of the agreement between predurotomy IOUS and preoperative MRI in intradural extramedullary (IDEM) spinal tumor surgery based on an interobserver comparative analysis of several anatomical structures that are consistently identifiable and reproducible in this setting. In addition, corresponding microsurgical images are presented, establishing an integrated MRI-IOUS-Microsurgery correlation framework that may be valuable for both neuroradiologists and neurosurgeons.

methodsA retrospective case series of patients with IDEM spinal tumors was compiled. Each case was characterised using a set of representative images depicting the same anatomical area of interest (tumor location), obtained by predurotomy IOUS, preoperative MRI and an operating microscope. Experienced neurosurgeons independently evaluated the IOUS images and MRI scans using a structured questionnaire and rated the perceived agreement on a 5-point scale by comparing several structures. Agreement between predurotomy IOUS and preoperative MRI was subsequently analysed. These structures were identified in an integrated MRI-IOUS-Microsurgery correlation framework.

resultsNine patients with IDEM spinal tumors were included. A total of 768 agreement assessments were obtained, of which 579 (75.39%) were comparable. Tumor and spinal cord showed the highest levels of agreement between IOUS and MRI, followed by the subarachnoid space and dura mater, which also showed high, although somewhat lower, agreement. In contrast, pia mater and nerve roots showed the lowest levels of agreement. Compared with MRIT1c, MRIT2 demonstrated significantly higher agreement with IOUS for nerve roots, dura mater, pia mater and subarachnoid space, while there were no significant differences for tumor or spinal cord. Sagittal images achieved significantly higher agreement scores than axial images. Despite systematic differences in scoring among observers, interobserver reliability was good when ratings from four observers were averaged, with comparable reproducibility for IOUS-MRIT1c and IOUS-MRIT2.

conclusionsTaking interobserver variability into account, IOUS-MRI agreement varied according to the anatomical structure evaluated, being highest for tumor and spinal cord, followed by subarachnoid space and dura mater, and lowest for pia mater and nerve roots. Sagittal images provided superior anatomical orientation compared with axial images. Interobserver reproducibility remained good despite systematic differences in individual scoring behavior.