Automated Measurement of Skin Thickness in Computed Tomography Images for Unilateral Lower Extremity Lymphedema

Background: Lower extremity lymphedema (LEL) is a common complication following gynecological cancer treatment, characterized by the accumulation of protein-rich fluid owing to lymphatic system insufficiency. This condition manifests as in skin thickening, soft tissue swelling, and other complications. Therefore, this study proposes an automatic method for measuring skin thickness in lower extremity computed tomography (CT) images and assessing its effectiveness in diagnosing LEL. Methods and Results: We selected 56 patients with unilateral LEL based on specific diagnostic criteria, including clinical evaluation, lymphoscintigraphy, indocyanine green lymphography, and a noncontrast CT scan, which included the entire lower extremities. The CT images had a pixel spacing ranging from 0.723 to 0.976 mm, with a slice thickness of 10.0 mm. Our skin thickness measurement method involves preprocessing, such as extracting the leg, bone, muscle, and subcutaneous fat regions, defining the target slice range, and measuring the skin thickness. The lower extremity was divided into eight subregions, and the skin thickness was measured across these subregions. The overall median skin thicknesses were 0.883 ± 0.201 and 1.536 ± 0.487 mm in normal and positive subregions, respectively. The classification of LEL was based on the Z-score calculated for each subregion, with the median skin thickness from normal legs serving as a reference. Our classification method demonstrated an overall accuracy of 0.839, sensitivity of 0.703, and specificity of 0.937. Conclusions: Our automated method for measuring skin thickness in CT images shows promise in diagnosing LEL, with high accuracy and specificity. This approach enables a comprehensive evaluation of the entire leg, potentially enhancing the diagnostic process for LEL.