Radiographic detection of retained neuropatties in the paranasal sinuses: a cadaveric diagnostic accuracy study.

backgroundRetained surgical materials are a preventable source of postoperative complications. This study evaluates the radiographic detectability of retained neuropatties in the paranasal sinuses to support safer postoperative imaging and surgical practice.

methodsSixteen previously dissected cadaveric heads were labeled and assigned to predefined sinonasal or nasopharyngeal site using an allocation table (e.g., frontal sinus anterior table, maxillary sinus, sphenoid sinus floor, nasal cavity, nasopharynx). For each head–site combination, a large neuropatty (1 × 3 in) was first inserted and imaged with anteroposterior and lateral digital radiographs, then removed and replaced by a small neuropatty (1/2 × 1/2 in) with repeat imaging, followed by anteroposterior and lateral radiographs without a neuropatty in place. This protocol yielded 32 images with large neuropatties, 32 with small neuropatties, and 32 without neuropatties (96 radiographs in total) for blinded interpretation by three neuroradiologists and subsequent diagnostic accuracy analysis.

resultsThere was substantial inter- and intra-rater reliability for neuropatty detection, with Fleiss’ κ values ranging from 0.67 to 0.78, and agreement with the previously established distribution of neuropatty locations was similarly high (κ = 0.61–0.82). Sensitivity and specificity for small neuropatties were 0.43 and 0.80 in AP views, respectively, and 0.68 and 0.93 in lateral views, respectively. Larger neuropatties yielded higher sensitivity (0.62 in AP and 0.78 in lateral views) with comparable specificities. Detection remained poor in anatomically complex regions, particularly the sphenoid sinus and nasopharynx, where sensitivities were as low as 0.00–0.17 in some views.

conclusionRadiologic accuracy is greater for larger neuropatties and lateral views but remains limited in anatomically complex regions, such as the sphenoid sinus and nasopharynx.