February 14, 2026European Spine Journal0 citations

Feasibility and accuracy of AI-assisted 3D MRI/CT fusion imaging for preoperative simulation of full-endoscopic transforaminal discectomy

DUDaisuke UkebaHokkaido University KNKen NagahamaHokkaido University YHYoshinori HyugajiSapporo Science Center

Key Points

Evaluate the accuracy and safety of AI-assisted 3D MRI/CT fusion imaging for preoperative simulation of full-endoscopic transforaminal discectomy.
Retrospective case series of 40 patients with L5/S1 intervertebral disc herniation.
Preoperative simulations guided virtual endoscopic axis to avoid nerve injury.
Measured and compared preoperative simulation angles with actual insertion angles postoperatively.
Significant improvement in VAS score for leg pain from 7.0 preoperatively to 1.5 at day 90.
Significant improvement in VAS scores for low back pain from 5.5 to 2.1 at day 90.
JOA scores improved from 13.9 preoperatively to 25.7 at day 90 (all p < 0.05).

Abstract

Full-endoscopic discectomy via the transforaminal approach (FED-TF) carries the risk of exiting nerve root (ENR) injury, and its feasibility at the lumbosacral level may be limited by anatomical constraints. Precise preoperative assessment is therefore essential. This study aimed to evaluate the accuracy, safety, and clinical efficacy of artificial intelligence (AI)-assisted three-dimensional (3D) MRI/CT fusion imaging for preoperative simulation of FED-TF in a retrospective case series. This retrospective case series included 40 patients with L5/S1 intervertebral disc herniation who were judged feasible for FED-TF without superior articular process (SAP) resection based on preoperative simulations. A virtual endoscopic axis was established to avoid interference from the iliac crest, SAP, and ENR. The axial and sagittal angles of the virtual axis were measured and compared with the actual endoscopic insertion angles postoperatively. Clinical outcomes were assessed using the visual analogue scale (VAS) for leg and low back pain, and the Japanese Orthopaedic Association (JOA) scores perioperatively. All the procedures were completed without any complications. The mean axial angle was 59.9° (simulation) versus 58.4° (actual), with no significant difference ( p = 0.236). The sagittal angle was 8.6° (simulation) versus 8.9° (actual), with no significant difference ( p = 0.739). The VAS score for leg pain improved significantly from 7.0 preoperatively to 2.1 on day 1 and 1.5 at day 90. The VAS scores for low back pain significantly improved from 5.5 to 3.5 and 2.1, respectively. The JOA scores improved significantly from 13.9 preoperatively to 21.5 on day 1 and 25.7 on day 90 (all p < 0.05). AI-assisted 3D MRI/CT fusion imaging demonstrated high accuracy in preoperative simulation and enabled safe and effective FED-TF in this case series. However, because CT and MRI were acquired separately and only cases not requiring SAP resection were included, further studies with larger cohorts and control groups are needed to validate and expand the applicability of this technique.

AIに質問

Bookmark

View Full Paper