Optimal Window Settings for Detection and Characterization of Ground-Glass Opacities on Computed Tomography in COVID-19 Patients Using a Simplex Algorithm-Based Approach.

Abstract

Background: Coronavirus disease-2019 (COVID-19) chest computed tomography (CT) involves ground-glass opacity (GGO) and denser consolidations, which are crucial for diagnosis.

Objectives: To determine optimal window settings for characterization and detection of GGO and dense consolidation on CT imaging in COVID-19 patients using a Simplex-based approach.

Methods: The study included 54 conventional CTs of COVID patients in two phases. First, CT images of 14 patients with GGO and 4 with dense consolidation were included. Seven radiologists evaluated representative images in different windows of varied width and center. They were graded for adequacy of characterization and detection. A Simplex algorithm was used to iteratively determine the optimal window settings. Surface response maps expressing the relationship between window settings and overall reader grades were constructed. Next, the reviewers compared manufacturer recommendations to the new optimal windows found on CT images of 40 patients.

Results: Overall, 12 different window settings were evaluated over a total of 1176 reads. Optimal characterization and detection of pure GGO was seen with a center of 630 HU and width 1460 HU, producing higher grades for both detection and characterization than the manufacturer window settings (P = 0.005). Optimal windowing for dense consolidation was like manufacturer measures (-585 HU and 1800 HU). In phase 2, an overwhelming preference of 78% favoring the optimal window compared to conventional settings was found.

Conclusions: GGO lung opacities characteristic for COVID-19 can be best seen using a lower CT windowing width than the manufacturer's recommendations, unlike denser consolidations, possibly due to differences in underlying pathophysiology.