Utility of White Light Scanning as an Alternative to Computed Tomography to Evaluate Severity of Pectus Excavatum Deformity

Background: Pectus excavatum (PE) severity and surgical candidacy are determined by computed tomography (CT)-delineated Haller Index (HI) and Correction Index (CI). White light scanning (WLS) has been proposed as a non-ionizing alternative. The purpose of this retrospective study is to create models to determine PE severity using WLS as a non-ionizing alternative to CT.

Methods: Between November 2015 and February 2023, CT and WLS were performed for children ≤18 years undergoing evaluation at a high-volume, chest-wall deformity clinic. Separate quadratic discriminate analysis models were developed to predict CT HI ≥ 3.25 and CT CI ≥ 28% indicating surgical candidacy. Two bootstrap forest models were trained on WLS measurements and patient demographics to predict CT HI and CT CI values then compared to actual index values by intraclass correlation coefficient (ICC).

Results: In total, 242 patients were enrolled (86.4% male, mean [SD] age 15.2 [1.3] years). Quadratic discriminate analysis models predicted CT HI ≥ 3.25 with specificity = 91.7%, PPV = 97.7% (AUC = 0.91), and CT CI ≥ 28% with specificity = 92.3%, PPV = 93.5% (AUC = 0.84). Bootstrap forest model predicted CT HI with training dataset ICC (95% CI) = 0.91 (0.88-0.93, R² = 0.85) and test dataset ICC (95% CI) = 0.86 (0.71-0.94, R² = 0.77). For CT CI, training dataset ICC (95% CI) = 0.91 (0.81-0.93, R² = 0.86) and test dataset ICC (95% CI) = 0.75 (0.50-0.88, R² = 0.63).

Conclusions: Using noninvasive and nonionizing WLS imaging, we can predict PE severity at surgical threshold with high specificity obviating the need for CT. Furthermore, we can predict actual CT HI and CI with moderate-excellent reliability. We anticipate this point-of-care tool to obviate the need for most cross-sectional imaging during surgical evaluation of PE.

Level of evidence: Level III.

Study type: Study of Diagnostic Test.