Evolution of Radiographic Changes of a Vascularized Pedicled Nasoseptal Flap after Endonasal Endoscopic Skull Base Surgery.

BACKGROUND AND PURPOSE: There is active research involving the radiographic appearance of the skull base following reconstruction. The purpose of this study was to describe the radiographic appearance of the vascularized pedicle nasoseptal flap after endoscopic skull base surgery across time. MATERIALS AND METHODS: We performed chart and imaging review of all patients with intraoperative nasoseptal flap placement during endoscopic skull base surgery at a tertiary academic skull base surgery program between July 2018 and March 2021. All patients underwent immediate and delayed (>3 months) postoperative MR imaging. Primary outcome variables included flap and pedicle enhancement, flap thickness, and flap adherence to the skull base. RESULTS: Sixty-eight patients were included. Flap (P = .003) enhancement significantly increased with time. Mean nasoseptal flap thickness on immediate and delayed postoperative scans was 3.8 and 3.9 mm, respectively (P = .181). The nasoseptal flap adhered entirely to the skull base in 37 (54.4%) and 67 (98.5%) patients on immediate and delayed imaging, respectively (P < .001). CONCLUSIONS: Our findings demonstrate heterogeneity of the nasoseptal flap appearance after skull base reconstruction. While it is important for surgeons and radiologists to evaluate variations in flap appearance, the absence of enhancement and lack of adherence to the skull base on immediate postoperative imaging do not appear to predict reconstructive success and healing, with many flaps "self-adjusting" with time.

Hybrid 3D/2D Convolutional Neural Network for Hemorrhage Evaluation on Head CT.

BACKGROUND AND PURPOSE: Convolutional neural networks are a powerful technology for image recognition. This study evaluates a convolutional neural network optimized for the detection and quantification of intraparenchymal, epidural/subdural, and subarachnoid hemorrhages on noncontrast CT. MATERIALS AND METHODS: This study was performed in 2 phases. First, a training cohort of all NCCTs acquired at a single institution between January 1, 2017, and July 31, 2017, was used to develop and cross-validate a custom hybrid 3D/2D mask ROI-based convolutional neural network architecture for hemorrhage evaluation. Second, the trained network was applied prospectively to all NCCTs ordered from the emergency department between February 1, 2018, and February 28, 2018, in an automated inference pipeline. Hemorrhage-detection accuracy, area under the curve, sensitivity, specificity, positive predictive value, and negative predictive value were assessed for full and balanced datasets and were further stratified by hemorrhage type and size. Quantification was assessed by the Dice score coefficient and the Pearson correlation. RESULTS: A 10,159-examination training cohort (512,598 images; 901/8.1% hemorrhages) and an 862-examination test cohort (23,668 images; 82/12% hemorrhages) were used in this study. Accuracy, area under the curve, sensitivity, specificity, positive predictive value, and negative-predictive value for hemorrhage detection were 0.975, 0.983, 0.971, 0.975, 0.793, and 0.997 on training cohort cross-validation and 0.970, 0.981, 0.951, 0.973, 0.829, and 0.993 for the prospective test set. Dice scores for intraparenchymal hemorrhage, epidural/subdural hemorrhage, and SAH were 0.931, 0.863, and 0.772, respectively. CONCLUSIONS: A customized deep learning tool is accurate in the detection and quantification of hemorrhage on NCCT. Demonstrated high performance on prospective NCCTs ordered from the emergency department suggests the clinical viability of the proposed deep learning tool.