Department of Radiology, Turku University Hospital, Turku, Finland.
Department of Biostatistics, Turku University Hospital, Turku, Finland.
Eur Radiol Exp. 2023 Oct 24;7(1):65. doi: 10.1186/s41747-023-00379-0.
We investigated whether a short, 5-min magnetic resonance imaging (MRI) protocol consisting of only axial T2-weighted and diffusion-weighted imaging (DWI) sequences can discriminate between tonsillar infections, peritonsillar abscesses and deeply extending abscesses in a retrospective, blinded, multireader setting.
We included patients sent by emergency physicians with suspected pharyngotonsillar infections who underwent emergency neck 3-T MRI from April 1 2013 to December 31 2018. Three radiologists (with 10-16 years of experience) reviewed the images for abscesses and their extension into deep neck spaces. Data were reviewed first using only axial T2-weighted Dixon images and DWI (short protocol) and second including other sequences and contrast-enhanced T1-weighted Dixon images (full protocol). Diagnostic accuracy, interobserver agreement, and reader confidence were measured. Surgical findings and clinical course served as standard of reference.
The final sample consisted of 52 patients: 13 acute tonsillitis with no abscesses, 19 peritonsillar abscesses, and 20 deeply extending abscesses. Using the short protocol, diagnostic accuracy for abscesses across all readers was good-to-excellent: sensitivity 0.93 (95% confidence interval 0.87-0.97), specificity 0.85 (0.70-0.93), accuracy 0.91 (0.85-0.95). Using the full protocol, respective values were 0.98 (0.93-1.00), 0.85 (0.70-0.93), and 0.95 (0.90-0.97), not significantly different compared with the short protocol. Similar trends were seen with detecting deep extension. Interobserver agreement was similar between protocols. However, readers had higher confidence in diagnosing abscesses using the full protocol.
Short MRI protocol showed good-to-excellent accuracy for tonsillar abscesses. Contrast-enhanced images improved reader confidence but did not affect diagnostic accuracy or interobserver agreement.
Short protocol consisting only of T2-weighted Dixon and DWI sequences can accurately image tonsillar abscesses, which may improve feasibility of emergency neck MRI.
• The short 3-T MRI protocol (T2-weighted images and DWI) was faster (5 min) than the full protocol including T1-weighted contrast-enhanced images (24 min). • The short 3-T MRI protocol showed good diagnostic accuracy for pharyngotonsillar abscesses. • Contrast-enhanced sequences improved reader confidence but did not impact diagnostic accuracy or interobserver agreement.
我们研究了在回顾性、盲法、多读者环境中，仅包括轴位 T2 加权和弥散加权成像（DWI）序列的 5 分钟短磁共振成像（MRI）方案是否可以区分扁桃体感染、扁桃体周围脓肿和深部扩展脓肿。
我们纳入了因疑似咽扁桃体感染而由急诊医师转介的患者，这些患者于 2013 年 4 月 1 日至 2018 年 12 月 31 日在 3T 急诊颈部 MRI 检查。三位有 10-16 年经验的放射科医生对脓肿及其向深部颈部间隙扩展的情况进行了阅片。数据首先仅使用轴位 T2 加权 Dixon 图像和 DWI（短方案）进行回顾，然后包括其他序列和对比增强 T1 加权 Dixon 图像（全方案）。测量了诊断准确性、观察者间一致性和读者信心。手术结果和临床病程作为标准参考。
最终样本包括 52 例患者：13 例急性扁桃体炎无脓肿，19 例扁桃体周围脓肿，20 例深部扩展脓肿。所有读者使用短方案的脓肿诊断准确性均为良好至极好：敏感性 0.93（95%置信区间 0.87-0.97），特异性 0.85（0.70-0.93），准确性 0.91（0.85-0.95）。使用全方案时，相应的值分别为 0.98（0.93-1.00）、0.85（0.70-0.93）和 0.95（0.90-0.97），与短方案相比无显著差异。在检测深部扩展方面也有类似的趋势。两个方案的观察者间一致性相似。然而，读者使用全方案诊断脓肿的信心更高。
短 MRI 方案对扁桃体脓肿的准确性为良好至极好。增强对比图像提高了读者的信心，但不影响诊断准确性或观察者间一致性。
仅包括 T2 加权 Dixon 和 DWI 序列的短方案（5 分钟）比包括 T1 加权增强对比图像的全方案（24 分钟）更快。
短 3T MRI 方案对咽扁桃体脓肿有良好的诊断准确性。
增强序列提高了读者的信心，但不影响诊断准确性或观察者间一致性。
Zotero 插件