Prediction of the Presence of Fluid Accumulation in the Subcutaneous Tissue in BCRL Using Texture Analysis of Ultrasound Images.

Abstract

Background: Breast cancer-related lymphedema (BCRL) is a chronic swelling of the arm due to breast cancer treatment. Lymphedema is diagnosed and staged on the basis of limb circumference measurements and the patient's subjective symptoms, which have poor reproducibility and objectivity: these cannot detect any fluid accumulation in the tissue. Ultrasonography is a feasible noninvasive technique that can be used to evaluate tissue structure in real time. This study aimed to assess the ability of texture features for discriminating the presence of accumulated fluid within the subcutaneous tissue of BCRL using ultrasound (US) imaging. Methods and Results: This study included 20 women who were treated for unilateral breast cancer and who subsequently developed BCRL (International Society of Lymphology stage II). Subcutaneous tissue was scanned through an US system (Sonosite Edge II; Sonosite, Inc., FUJIFILM) using a 6- to 15-MHz linear transducer to assess the ability of texture features for discriminating the presence of accumulated fluid within the subcutaneous tissue of BCRL. Fluid accumulation was observed using a 3-Tesla MR system under double-echo steady-state conditions. There was a significant difference among the three groups (with hyperintense area, without hyperintense area, and unaffected side) in 11 of 14 textural features (p < 0.05). Post hoc analysis (Mann-Whitney U test; Bonferroni correction p < 0.0167) revealed significant differences in seven textural features within the hyperintense area. Conclusions: This study revealed that seven texture features quantified by US imaging data can provide information regarding fluid accumulation in the subcutaneous tissue of lymphedema.