- 學位論文
乳房攝影與磁振影像之乳腺緻密度相關性研究
Correlation of Breast Densities between Magnetic Resonance Image and Digital Mammography
摘要
乳腺緻密度與乳癌具有一定程度之關連性。目前放射科醫師大多利用醫學影像進行乳癌定性診斷。其中大部份採用目視法評估乳房影像之緻密程度，此法很難客觀量化且因人而異，因此本研究針對乳房攝影影像(Mammography)與磁振造影影像(Magnetic Resonance Imaging, MRI)採用客觀方法量化或估計乳腺緻密程度，同時評估二者量化乳腺緻密程度之相關性，做為未來提供乳腺緻密之參考指標。 利用回顧性分組研究實驗設計，對同一案者同時收集MRI 和Mammography影像；共計49 例，平均年齡52 歲且介於38 歲至64 歲之間，排除乳房植入物及乳房手術史。接著記錄QuantraTM 軟體對Mammography 估算之乳腺緻密度(Volume by Mammography with QuantraTM, VQ)；MRI 部份則使用K-means 估算MRI 乳腺緻密度(Volume by MRI with K-means, VM)。透過迴歸分析找出二者乳腺緻密度之相關性。樣本分析採用案例基礎(49 人)與乳房基礎(89 個乳房)。 以案例為樣本單位之迴歸分析結果為左側與右側乳房的VQ與VM相關性(r)皆為0.708、線性迴歸係數(β)為0.878 與0.758、R2 為0.522 與0.501，代表左右側之VQ 與VM 呈現高度正相關性；若按年齡分組之左右側乳房為分析樣本時，其中50 歲以上之群組其r 皆高於0.8 與0.7、R2 高於0.65 與0.44、β 介於1.104與1.538 與0.72 與1.119。以乳房個數為樣本單位之迴歸分析結果為相關性(r)為0.583、β 是0.628、R2 為0.34，代表VQ 與VM 呈中度正相關性；按年齡分組發現，年齡高於50 歲時，其r 皆高於0.75、R2 均高於0.55、β 介於0.87 與1.41。不論是以案例或乳房個數為樣本單位均顯示出以50 歲為分界之VQ 與VM 具有很高相關性。 對於不分年齡之乳腺緻密相關分析，以案例為樣本單位之迴歸分析會得較佳之結果；當女性年齡高於50 歲時，不論以案例或乳房個數為分析單位，利用Mammography 與MRI 評估乳腺緻密度均具有很高的相關性；當女性年齡低於歲時，Mammography 與MRI 評估乳腺緻密度僅具有低中度的相關性，其中又以左側乳房之Mammography 與MRI 具有較好的相關性。
並列摘要
The breast density are highly certainty degree of resistance with breast cancer. The invasive medical images are useful to diagnose breast cancer by the method of radiologists according to breast density and structural shape shown in images. However, it is difficult to provide objectively quantification and information. In this study, both mammography imaging (Mammography) and MR imaging (Magnetic Resonance Imaging, MRI) are used to generate objective quantitative information by estimated the correlation between breast densities. The retrospective study are designed to collect the effective samples with MRI and Mammography images simultaneously. Total of 49 cases are involved with average age of 52 years and between 38 to 64 years by excluding the history of breast implants and breast surgery. The Mammography breast density (Volume by QuantraTM, VQ) were estimated by QuantraTM software. The K-means were applied to estimate MRI breast density (Volume by MRI with K-means, VM). Regression analysis was used to identify both the correlation between VQ and VM. The analysis areaccording to case basis (49) and breast basis (89 breast). In the case basis, the regression analysis for the left and right breast VQ and VM are with correlations (r) 0.708 and 0.708, linear regression coefficients (β) of 0.878 and 0.758, and R2 0.522 and 0.501 respectively. It was shown that VQ and VM was high positive correlation. Meanwhile, the regression analysis for VM and VQ with left and right breast densities ware higher than 0.8 and 0.7, R2 higher than 0.65 and 0.44, β between 1.104 and 1.538 and 0.72 and 1.119 with respectively. In the breast basis, the regression analysis of the VQ and VM is with correlation (r) was 0.583, β is 0.628, and R2 of 0.34. The moderate positive correlation was shown between VQ and VM. As patient’s age over 50 years of age, the r 0.75, R2 is higher than 0.55, β is between 0.87 and 1.41 were shown after regression analysis between VQ and VM. Under patient’s age over 50 years old in both cases and breast basis, the correlation between VQ and VM was highly positive relation. The correlation between VQ and VM under all ages, the correlation in case basiswas higher than those of in breast bases. When age is over 50 years old, the VQ and VM (Mammography and MRI) are shown high relevance. As age is less than 50 years old, the correlation of breast density between Mammography and MRI was displayed moderate relevance. In the breast basis , the correlation between VQ and VM on the left side was higher than those of right side.