Development of automatized new indices for radiological assessment of chest-wall deformity and its quantitative evaluation

Abstract

Pre-operative diagnosis of chest-wall deformity is important for successful surgical correction and post-operative evaluation of funnel chest patients. However, conventional indices that define the severity of deformity have several limitations; manually calculated and cannot supply information about asymmetry. We developed four indices that can represent both the depression and the asymmetry of the chest-wall, and can automatically be extracted by computerized image processing technique. Three indices, including eccentricity index (EI), flatness index (FI), and circularity index (CI), were suggested to represent the depression of the chest-wall, and one index, rotation index (RI), to represent the asymmetry of the chest-wall. To verify the feasibility of new indices, several synthetic images and real CT images were used to analyze the performance of new indices and the statistical relationship with conventional Haller index. The experimental results showed possible application of suggested indices to the diagnosis of funnel chest patient. Suggested indices showed clear trends of change with the severity of chest-wall deformation in regards to both the depression and the asymmetry. Results of statistical analysis showed high correlation between new indices and HI, showing possibility of replacing HI.