[Computerized tomography and magnetic resonance in radical and reconstructive surgery using pedunculated and revascularized flaps in advanced-stage tumors of the head and neck. Analysis of recurrences. II].

January, 1992, to October, 1995, sixty-four patients with advanced head and neck carcinoma were submitted to reconstructive surgery using pedunculated myocutaneous or revascularized flaps and then to CT and MRI. Myocutaneous flaps were used in 26 cases--12 tubular and 14 linear flaps--and revascularized flaps in 38--14 latissimus dorsi flaps, 12 temporal flaps, 7 jejunal flaps and 5 radial flaps. Twenty-six of 64 cases (41%) relapsed: MRI correctly depicted the recurrence in 24 cases and CT in 19, with 2 false positives and 2 false negatives at MRI and 6 false positives and 7 false negatives at CT. The sensitivity, specificity, accuracy, positive and negative predictive value of CT were 84%, 78%, 73%, 76% and 82%, respectively, while the corresponding MR rates were 95%, 94%, 92%, 92% and 95%. MRI was more accurate than CT in demonstrating postoperative and postirradiation changes (92% for MRI versus 73% for CT) thanks to its higher sensitivity in depicting tumor tissue on T2-weighted and post-Gd-DTPA images. CT is very useful in the early postoperative period, to follow-up poorly collaborative patients, because its acquisition time is short; MRI should be performed when CT findings are questionable and the revascularized flap is used to repair a large defect at the skull base.