時間序列之雙波段紅外線乳房影像組織分類

Abstract

乳癌是女性常見的癌症之一且發生率有逐年升高之趨勢導致，目前婦女因乳癌死亡之比率占所有癌症第二，因此這十幾年來，乳癌成為女性健康的主要議題，而早期診斷乳癌和有效治療乳癌成為保障女性健康之不二法門。乳癌之治療除了傳統乳房手術外，還有許多治療方法，如：化學治療、放射線治療和賀爾蒙治療等。為了降低手術風險，臨床上積極採用新輔助式化學治療(Neoadjuvant chemotherapy)，即手術前給予化學治療使得腫瘤縮小後再做手術，但目前臨床上尚未發展出一有效評估化療反應之工具。 本研究提出一定量型雙波段紅外線影像技術(Quantitative Dual-Spectrum Infrared, QDS-IR)做為評估化療療效之工具，因紅外線影像除了具有非侵入性、低成本以及可反覆成像等優點以外，最重要的是具有感測微小癌細胞血管增生之能力，因此紅外線影像被視為具有潛力來評估化療反應的醫學影像技術。 QDS-IR各自捕捉中波(波長為3-5μm, MIR)與長波(波長為8-9.2μm, LIR)的光子並且運用雙波段熱圖譜分離(Dual-Spectrum Heat Pattern Separation, DS-HPS)演算法，推估影像中基礎體溫組織區與高溫組織區的含量，並且採用自組織映射圖(Self-organizing map, SOM)分類演算法，針對多時間點化療前與化療後之雙波段紅外線影像做組織分類，找出隨著化療而有相似反應的腫瘤及其周邊血管組織，並針對這些區域做量化分析。初步研究結果顯示，經由SOM分類找出之腫瘤及其周邊血管組織的面積以及高溫組織平均含量，皆會隨著化療次數的增加而有下降之趨勢。本研究亦採用動態對比顯影核磁共振影像來確認本系統方法之正確性，初步發現本研究結果與MRI顯影的區域有某種程度之相關，因此認為本研究提出之QDS-IR系統可有效的偵測乳癌高溫組織的改變量，並用以初步評估化療反應。

The breast cancer is one of the common cancers for women and the incidence ration is trending up year by year. The death ratio of women who died of the breast cancer is the second of all kinds of cancers currently. So the breast cancer has become the main threat of women's health in the past decades. Early diagnosis and effective treatment of the breast cancer have become the one and only way to protect women's health. Besides the traditional breast surgery, there are many treatments, such as chemotherapy, radiotherapy and hormone therapy. Clinically, neoadjuvant chemotherapy shrinks the tumor size in advance, thus reducing the surgery risk. By far, however, an effective system for treatment response assessment of chemotherapy is not available. This study proposed a quantitative dual-spectrum infrared (QDS-IR) technology to assess the chemotherapy response. Infrared modality has such advantages as non-invasive, lower costs, and repeatable monitoring. The upmost one is that its being able to sense the growth of angiogenesis. Therefore, infrared has the potential to assess chemotherapy response. QDS-IR system captured the IR photon information from a middle-wave and a long-wave IR (MIR and LIR) cameras. To quantify the changes of the heat pattern, a dual-spectrum heat pattern separation (DS-HPS) algorithm was developed to separate the tissue area of high temperature from that of normal temperature. After classifying the longitudinal dual-spectrum infrared image by self-organizing map (SOM), the changes of tumor and its peripheral blood vessels were contoured which have the similar chemotherapy responses in the chemotherapy courses. The preliminary results showed that the area and the mean energy originated from the high temperature tissues both tended to decrease as the patients received more chemotherapy. By using DCE-MRI to confirm this system’s accuracy, the preliminary results showed that the information revealed by the QDS-IR spectrogram in accordance with that by MRI. It suggests the proposed QDS-IR system has the potential to detect the changes of the high temperature of breast cancer effectively and assess the chemotherapy response.