Sparse-view Cone-beam Breast CT Reconstruction via cGAN Constrained by Image Edges / 中国医疗器械杂志

Clinical applications of cone-beam breast CT(CBBCT) are hindered by relatively higher radiation dose and longer scan time. This study proposes sparse-view CBBCT, i.e. with a small number of projections, to overcome the above bottlenecks. A deep learning method - conditional generative adversarial network constrained by image edges (ECGAN) - is proposed to suppress artifacts on sparse-view CBBCT images reconstructed by filtered backprojection (FBP). The discriminator of the ECGAN is the combination of patchGAN and LSGAN for preserving high frequency information, with a modified U-net as the generator. To further preserve subtle structures and micro calcifications which are particularly important for breast cancer screening and diagnosis, edge images of CBBCT are added to both the generator and the discriminator to guide the learning. The proposed algorithm has been evaluated on 20 clinical raw datasets of CBBCT. ECGAN substantially improves the image qualities of sparse-view CBBCT, with a performance superior to those of total variation (TV) based iterative reconstruction and FBPConvNet based post-processing. On one CBBCT case with the projection number reduced from 300 to 100, ECGAN enhances peak-signal-to-noise ratio (PSNR) and structural similarity (SSIM) on FBP reconstruction from 24.26 and 0.812 to 37.78 and 0.963, respectively. These results indicate that ECGAN successfully reduces radiation dose and scan time of CBBCT by 1/3 with only small image degradations.

Reliability of breast density estimation based on cone beam breast CT / 中国肿瘤临床

Objective:To investigate the accuracy of a threshold-based segmentation method based on cone beam breast CT(CBBCT) images in breast density measurement,and its value for breast-type classification and breast cancer screening.Methods:A retrospec-tive analysis of 195 patients who had undergone CBBCT examination at Tianjin Medical University Cancer Institute and Hospital be-tween May 2012 and August 2014 was performed.A total of 64 breasts were analyzed.On the basis of the classification criteria for breast density in BI-RADS,they were classified into four types and the majority report was reported.Breast density was measured by the threshold-based segmentation method based on CBBCT images and corrected manually to obtain the corrected breast density.A month later,the procedure was repeated.Intra-class correlation coefficients(ICCs)were used to compare the intra-observer and inter-observer consistencies of threshold-based segmentation and manually corrected breast density measurement results for non-dense and dense breasts.Results:For threshold-based segmentation measurements the intra-observer and inter-observer ICC values were 0.0.9624(95% CI:0.9388~0.9770)and 0.9666(95% CI:0.9500~0.9785).For manually corrected measurements,the intra-observer and inter-observer ICC values were 0.9750 (95% CI: 0.9592~0.9847) and 0.9775 (95% CI: 0.9661~0.9855). The ICC between the threshold-based segmentation method and manual correction was 0.9962 (95% CI: 0.9983~0.9977). The ICC values of threshold-based and manually corrected measurement in non-dense and dense breasts were 0.9497(95% CI:0.7072-0.9914)and 0.9983(95% CI:0.9971-0.9990),respectively.Conclusions:The threshold-based segmentation method based on CBBCT is a reliable and accurate com-puter-aided method of measuring breast density.It is expected to be applied in large-scale screening of breast cancer and to provide more information for predicting the risk of breast cancer.

Comparison of the diagnostic efficiency in breast malignancy between cone beam breast CT and mammography in dense breast / 中华肿瘤杂志

Objective@#To compare the diagnostic efficiency of lesion in dense breast between cone beam breast computer tomography (CBBCT) and mammography.@*Methods@#From May 2012 to August 2014, 160 patients with 165 breasts who underwent mammography and CBBCT examinations were included in this study. The diagnostic results of CBBCT and mammography were reviewed and compared with pathological results.@*Results@#In the 165 breast, 24 were dense breasts and 141 were dense breasts. The diagnostic results were similar in 109 lesions, but different in 56 lesions. According to the analysis of the 165 breasts using receiver operation characteristics (ROC) curve, the area under the ROC curves (AUC) of CBBCT and mammography were 0.923 (95%CI: 0.878-0.967, P<0.05) and 0.959 (95%CI: 0.926-0.992, P<0.05), respectively. With Breast Imaging-Reporting and Data System (BI-RADS) 4b as the critical value, the sensitivity and specificity were 70.0% and 98.7% using mammography, and 83.3% and 97.3% using CBBCT, respectively. The AUC of CBBCT and mammography of the 141 dense breasts was 0.919(95%CI: 0.868-0.969, P<0.05) and 0.973(95%CI: 0.947-0.999, P<0.05), respectively. With BI-RADS 4b as the critical value, the sensitivity and specificity were 69.0% and 98.6% by mammography, and 83.1% and 98.6% by CBBCT, respectively.@*Conclusions@#CBBCT showed higher sensitivity and specificity in the diagnosis of breast malignant tumors compared to mammography. It is expected to be applied to breast cancer detection in the future, especially in dense breast.