Aim: To determine the diagnostic accuracy of MRI, degree of confidence in diagnosis and interobserver agreement in patients with COVID-19 associated acute invasive fungal sinusitis (AIFS).
Methods: Study was undertaken in a tertiary care hospital using retrospective case-control design. Cases were 68 confirmed AIFS patients and controls were 23 non-AIFS patients with concurrent or recent COVID-19 infection who underwent MRI scan of paranasal sinuses during a period of 8 weeks. MRI scans were independently reviewed by two consultant radiologists who were blinded to the clinical presentation, laboratory test results and nasal endoscopy findings. Detailed imaging parameters and predefined degree of confidence in diagnosis were recorded and analysed using SPSS software version 23.
Results: Sensitivity of the two radiologists (TP and SG) for MRI diagnosis of definite AIFS was 93% and 91% respectively and specificity was 100% for both. Sensitivity when using ‘loss of contrast enhancement’ to diagnose as probable or definite AIFS were 89% and 92% and when using ‘extrasinosal involvement’ were 91% and 90%, for TP and SG respectively. Specificity of both the MRI parameters was 100% for both radiologists. There was almost perfect statistical agreement between the radiologists for all imaging parameters (k = 0.84 to 0.97) and for degree of confidence in diagnosis of definite AIFS Vs possible AIFS Vs non-fungal (k = 0.88).
Conclusion: MRI is a highly sensitive and specific modality to diagnose AIFS with excellent interobserver agreement. Loss of contrast enhancement and extrasinosal involvement are two highly reliable MRI parameters in ascertaining diagnosis in the appropriate clinical scenario.