Clinical Application Value Of Cone Beam Breast Computed Tomography

Objective:To investigate the clinical application value of cone beam breast CT,based on the reliability of breast density evaluation and the accuracy of tumor sizing in invasive breast carcinoma on CBBCT comparing to mammography.Methods:（1）The preoperative CBBCT images and mammograms of 200 female patients performed from May,2012 to January,2014 in Tianjin Medical University Institute and Hospital were analyzed in this retrospective study.Breast density assessment of 130 patients who were eligible for inclusion criteria were performed by5 radiologists based on CBBCT images and mammograms referring to 5^th BI-RADS atlas.The interval evaluation time between two modalities was 2 weeks,and in random order.Evaluations were repeated by 3 radiologists 1 month later.The intra-reader agreement and the agreement between two modalities were analyzed by weighted Kappa.Intraclass correlation coefficient（ICC）was used to assess inter-reader agreement.（2）In this retrospective study,we collected 133 female patients（with 135 lesions）who underwent CBBCT and mammography and were pathologically confirmed invasive breast cancer carrying immunohistochemical results,from November,2011 to December,2017,in Tianjin Medical University Cancer Institute and Hospital and Sun Yat-sen University Cancer Center.Tumor size evaluation were performed on preoperative mammograms and CBBCT images.All the lesions were divided into concordance and discordance groups,defined as the discrepancy less than 0.5 cm referring to the pathology estimation results.ICC and Spearman’s correlation coefficient were used to analyze the reliability of tumor sizing based on CBBCT.CBBCT-pathology discordance was analyzed the difference between two groups of clinical,histopathology and CBBCT image features by using t test,chi-square and Mann-Whitney U test.ROC curve was used to analyze the cut-off value between tumor size and enhanced-degree with CBBCT-pathology discordance.Results:（1）The inter-reader agreement on CBBCT（ICC=0.781,95%CI:0.727-0.829）and mammograms（ICC=0.744,95%CI:0.687-0.797）were both on moderate agreement,but higher on CBBCT.The level of intrareader agreement was higher on CBBCT images（Kappa=0.856,95%CI:0.794-0.918）than mammograms（Kappa=0.786,95%CI:0.717-0.856）for breast density evaluation.Based on the majority report,the agreement between two modalities was on substantial agreement degree（Kappa=0.734,95%CI:0.622-0.846）.Comparing to mammogram result,15patients were evaluated as higher category on CBBCT images and only 1 as lower category.（2）The agreement between CBBCT（2.20,（1.70,2.90）cm）and pathology（2.00,（1.50,3.00）cm）measurement was higher than mammography（2.20,（1.70,3.00）cm）based on ICC value and Spearman’s correlation coefficient.On univariate analysis,CBBCT scan mode（P=0.032）,tumor size（P<0.001）,lesion type（P<0.001）,ER expression level（P=0.044）and enhanced-degree（P=0.003）were associated with CBBCT-pathology discordance.The cut-off value of ROC curve between tumor size and enhanced-degree with CBBCT-pathology discordance was 2.95cm and 33.10 HU.Conclusion:（1）Referring to the 5th BI-RADS atlas,CBBCT showed equal aptitude and better agreement on breast density assessment comparing to mammography.Hence,CBBCT could be an effective modality for breast density assessment and breast cancer risk evaluation in clinical diagnosis.（2）CBBCT has high accuracy on invasive carcinoma tumor size estimation,especially on enhanced-contrast CBBCT examination.larger tumor,ER low expression or non-mass lesions were strongly associated with CBBCT-pathology discordance,which will change the treatment planning and type decision.