Improvement of Microcalcification Visualization Using Dual-Energy Approach in Digital Mammography

• Main Authors: Chia-Jung Tsai, 蔡佳容
• Other Authors: Jason JS Lee
• Format: Others
• Language: en_US
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/54640733167804699321

博士 === 國立陽明大學 === 生物醫學影像暨放射科學系 === 100 === Objectives: As microcalcifications are one of the earliest indicators of breast cancer, their detection at an early stage plays a decisive role in reducing breast cancer mortality. The first part of this study is to improve microcalcification detectability using a dual-energy digital mammographic (DEDM) technique. Secondary, this study is to explore the possibility of dose reduction using the DEDM technique, while maintaining robust detection ability. Materials and methods: A plate consisting of five different sized pre-sifted calcium carbonate (CaCO3) grains was used to construct simulated microcalcifications between 0.16-0.39 mm, and each cluster contained five microcalcifications of the same size. Slabs of breast-equivalent plastics with three simulated tissue densities were used to achieve various breast thicknesses from 3 to 7 cm. Three DEDM protocols were performed by adjusting the effective tube current time product (mAs) of LE image at the same (100%), one half (50%) and one quadrant (25%) of that used in HE image acquisition, named DEDM100%, DEDM50% and DEDM25%, respectively. A single-energy digital mammography (SEDM) method was also used as the control group. A total of 525 ROIs were used to compare the performance of the DEDM v.s. DEDM and DEDM v.s. SEDM by using free-response receiver operating characteristic (FROC) and areas under the FROC curve (Az). Results: For the feasibility study, using the DEDM technique, microcalcifications were successfully imaged using a breast thickness less than 5 cm when microcalcifications were larger than 0.2mm. The average number of false-positive (FP) per image at a 90% true-positive (TP) fraction was 0.0198 (Az = 0.956±0.027) using DEDM compared to 0.292 (Az = 0.681±0.235) using SEDM. Lower radiation dose for the DEDM technique could be potentially observed when the DEDM with a thickness of less than 5 cm, compared to SEDM method with a thickness larger than 5 cm. Regarding low-dose DEDM technique, for each tissue density, estimated microcalcifications using the three DEDM protocols demonstrated no significant differences; nevertheless, for each microcalcification cluster, there was an observed statistical underestimation of 0.16 mm sized microcalcifications for DEDM25% as compared to either DEDM100% or DEDM50% protocols (p=0.027). The TP fraction was 0.90 for an average of 0.017-0.042 FP per image using the DEDM100%, 0.017-0.114 using the DEDM50%, 0.021-0.148 using the DEDM25%, and 0.134-0.422 using the SEDM, respectively. The estimated Az values were 0.915-0.940, 0.867-0.935, 0.824-0.930 and 0.567-0.673, respectively. All DEDM protocols ranked significantly higher than the SEDM method (p<0.001). Conclusion: The DEDM50% protocol could provide a tradeoff benefit between accurate microcalcification detectability and lower radiation dose for any tissue density. If it was imaged under fatty conditions, the DEDM25% protocol could offer the best dose reduction while still maintaining detection accuracy. There is a potential to reduce radiation dose in DEDM without a negative impact on image quality which could improve compliance and provide earlier diagnosis of breast cancer.