乳癌化療病患之多時間點紅外線影像對位

Image Registration of Longitudinal Infrared Images for Breast Cancer Patients Receiving Chemotherapy

張哲瑋(Che-Wei Chang)

指導教授 ： 陳中明

國立臺灣大學/工學院/醫學工程學研究所/碩士(2010年)

https://doi.org/10.6342/NTU.2010.10355

摘要

根據行政院衛生署統計，乳癌一直都是女性主要的致死病症之一，唯有早期的發現及完善的治療監控才能降低乳癌對於台灣女性的威脅。然而，至今仍然沒有一種檢查方法可有效的達到此目標。所以本實驗室近年來著重於發展一套量化型雙波段紅外線影像分析系統，嘗試追蹤高溫組織隨著時間變化的資訊以量化分析化學治療之效果。本研究之主要目的即為發展其中關鍵技術—多時間之雙波段紅外線影像對位演算法。
本實驗中，影像對位演算法是要將經過雙波段對位後的多時間點紅外線乳房影像形變為同一幾何結構，使相對應之組織移動到相同位置以取得更多有用的資訊。其演算法大約分為三個步驟，首先利用Harris角落點偵測演算法(Harris corner detector)偵測出目標影像(target image)及欲對位之來源影像(source image)中的角落點，以及分析Hessian矩陣找出血管中線之交叉點。接著利用手動選擇的方式建立兩張影像中角落點及血管交叉點的相互關係，以形成一組應用於薄板仿樣法(Thin Plane Spline, TPS)之相對應點(corresponding points)。最後利用Nelder-Mead單體法(simplex method)來修正來源影像中控制點的位置，並使得共同訊息(mutual information, MI)達到最大值，以得到最佳化對位結果。
為了評估多時間點紅外線影像對位演算法的準確性，本研究拍攝了十位正常受測者並模擬不同時間點造成的乳房形狀及熱圖譜的改變。實驗結果顯示，本研究所提出的對位演算法具有一定的精準度，其最大誤差值約為1.57個pixels。我們也將對位演算法應用在化療病患的雙波段紅外線序列影像中，並利用DS-HPS演算法計算出有效高溫組織含量。其結果顯示，乳癌部位之高溫組織含量會隨著時間下降，這也證明了結合多時間點紅外線影像對位演算法的雙波段紅外線系統，對於乳癌的偵測及化療的追蹤評估具有一定的潛力。

關鍵字

乳癌 ； 影像對位 ； Harris角落點偵測演算法 ； 血管中線 ； 薄板仿樣法 ； Nelder-Mead單體法 ； 共同訊息

並列摘要

According to the statistical data from Department of Health, breast cancer has been one of major cancer diseases causing death for a long time. Detecting breast cancer early and monitoring the effect of chemotherapy completely could reduce the death rate of breast cancers. However, there were few methods to achieve the goal effectively. Therefore, recently, our team has been developing the Quantified Dual Spectrum-Infrared (QDS-IR) system to quantify the effect of chemotherapy by adding the information of heat changing with time. The aim of this study was to develop one of the key technologies of the QDS-IR system, i.e., the registration algorithm for longitudinal dual spectrum infrared images.
The proposed image registration algorithm was composed of three steps. First, the corner points on the heat pattern were detected by Harris corner detector algorithm and the intersections of blood vessels were found via the Hessian matrix. Second, the relationships of the corner points and the intersections of blood vessels were established between the source image and the target image manually to get a set of corresponding point for Thin Plane Spline (TPS) model. Third, the Nelder-Mead simplex method was used to modify the locations of control points on the source image via maximizing the mutual information (MI) value. In the mean time, optimal and sub-pixel registration could be achieved.
To evaluate the accuracy of the longitudinal infrared image registration algorithm, we took the sequences infrared images of ten normal subjects, and simulated the changes of the breast shape caused by poses and the variation of the heat patterns at different time points. The experiment results showed that the maximal registered error was about 1.57 pixels. We applied the proposed registration algorithm to the breast cancer patients who need to do chemotherapy, and used DS-HPS algorithm to quantify the effective volume of high temperature tissues. The results showed that the effective volume of tumor tissues would decrease with time. The QDS-IR system with longitudinal image registration had potential to assess the effect of chemotherapy and further to detect breast cancers.

並列關鍵字

Breast cancer ； image registration ； Harris corner detector ； middle lines ； Thin Plane Spline ； Nelder-Mead simplex method ； mutual information

參考文獻

[1] Albanes, D., Weinberg, G. B., Boss, L. et al., “A survey of physician’s breast cancer early detection practices,” Prev Med. 1988; 17: 643–52.

[3] Rieber, A., Brambs, H. J., Gabelmann, A., et al. “Breast MRI for monitoring response of primary breast cancer to neo-adjuvant chemotherapy,” Eur Radiol. 2002; 12: 1711-1719.

[5] Qi, H., Kuruganti, P. T., Snyder, W. E., “Detecting breast cancer from thermal infrared images by asymmetry analysis," In Biomedical Engineering Handbook. 2006, CRC Press.

[6] Ng, E. Y. K., Ung, L. N., Ng, F. C. et al., “Statistical analysis of healthy and malignant breast thermography,” Journal of Medical Engineering and Technology. 2001; 6: 253–263.

[8] Zylberberg, B., Salat-Baroux, J., Ravina, J. H. et al., “Initial chemoimmunotherapy in inflammatory carcinoma of the breast,” Cancer. 1982; 49: 1537-1543.