在本篇論文，我們提出一套自動的方法，可以增強乳房X光片影像中可疑乳癌腫塊的區域，並且把這些區域在乳房影像上框選出來。乳癌腫瘤和微鈣化點在X光片影像上的一個主要特徵是灰階值亮度會比其它正常組織為高。要增強影像對比，我們就針對此一項特性來把可疑區域的亮度值保持住，並且降低其它區域的亮度。
我們首先利用 Histogram Equalization & Contrast Enhancement 和 Difference of Gaussians Filter 兩種不同的方法來增強可疑腫瘤和微鈣化點的影像與背景之間的對比，讓這些可疑的位置可以很明顯的由影像中判讀出來。接著我們利用 Difference of Gaussians Filter 加上 Bezier Curve 由乳房影像中求出乳房的輪廓，再把所得到的輪廓和先前利用兩種不同方法所得到的增強影像相結合，可以幫助醫生在診斷時有參考點的依據。 我們也介紹如何利用自組織類神經網路（Self-Organization Map neural network）和Morphology的方法，來由影像中自動框選出可疑腫瘤區域的影像。

實驗結果由台北榮民總醫院小兒放射科刁翠美刁主任來評估。刁主任認為我們對可疑腫瘤區域所增強過後的影像，對醫生在診斷上有所幫助。而可疑腫瘤區域的框選結果，也和經由刁主任框選出來的結果相對照。我們發現我們所發展的演算法使用在現有醫生所提供的資料庫中，可以把真正的腫塊都框選出來，但是缺點是會額外的框選出不是腫塊的區域。

本篇論文的目的在於提供一套自動化的電腦輔助乳癌判斷系統，類似第二位專家，從旁輔助助醫生在診斷乳癌上能夠更快速與正確。

This study focuses on how to enhance the suspicious regions in the mammography and circumscribe the regions of interest (ROI) automatically. We introduce use of histogram equalization & contrast enhancement (HECE) algorithm and Difference of Gaussian (DoG) filter to enhance the suspicious masses and micro-calcifications. The DoG filter and Bezier curve are then used to find the boundary of breast in the mammography, which can provide useful orientation information to doctors during diagnosing the mammograms. Methods using self-organization map (SOM) neural network and morphological techniques are also developed to separate the suspicious circumscribed masses from the mammography.
The experimental results had been assessed by Dr. Tiu, the Chairwoman of the Pediatric Radiology Department, Veterans General Hospital, Taipei, to determine the efficacy of the proposed methods. It is concluded by Dr. Tiu that the enhancement is good and can be valuable to doctors during the mammogram diagnosis. A comparison of the suspicious mass circumscription results with that circumscribed manually by Dr. Tiu was also performed. It was found that our algorithms can detect all the masses in the mammography without any loss in our experiments, i.e., the proposed method can not only mark out the regions spotted by doctors but also those suspicious regions which might not be marked by doctors in the first glance, thus reducing the misdiagnosis probability.

[1] Bettyann Holtzmann Kevles, Naked to the bone : medical imaging in the twentieth century, Rutgers University Press, 1997.
[2] Taiwan Department of Health, Republic of China, http://www.doh.gov.tw/new/891115/891115-6.htm.
[3] I. Andersson, “Mammography in clinical practice,” Med. Radiography and Photography, vol. 62, no. 2, p.2, 1986
[4] D. Kopans, Breast Imaging, Philadelphia, PA: J. B. Lippincott Company, 1989.
[5] I. Christoyianni, E. Dermatas, and G. Kokkinakis, “Fast detection of masses in computer-aided mammography,” IEEE Signal Processing Magazine, pp.54-64, Jan 2000.
[6] M. L. Giger, “Computer-aided diagnosis,” RSNA Categorical Course Phys., pp. 283–298, 1993.
[7] W. Polakowski, D. Cournoyer, S. Rogers, M. DeSimio, D. Ruck, J. Hoffmeister, R. Raines, “Computer-aided breast cancer detection and diagnosis of masses using difference of Gaussians and derivative-based feature saliency,” IEEE Trans. On Medical Imaging, VOL. 16, NO. 6, pp. 811-819, December 1997.
[8] C. Vyborny, “Can computers help radiologists read mammograms,” Radiology, Vol. 191, pp. 315-317, 1994.
[9] Isaac Bankman, Handbook of Medical Imaging Processing and Analysis, Academic Press, 2000.
[10] Rogers, L.S.; Bowling, C, “A new technique for automatic contour tracing,” Proceedings of the 16th Annual International Conference of the IEEE , Vol.1, pp. 608-609,1994.
[11] Crawford-Hines, S.; Anderson, C, “Machined-learned contours to assist boundary tracing task, ” Image Analysis and Interpretation, 1998 IEEE Southwest Symposium, pp.229-231, 1998.
[12] Alan Watt, 3D Computer Graphics, Addison-Wesley, 2000.
[13] Rafael C. Gonzalez; Richard E. Woods, Digital Image Processing, Addison Wesley, 1993.
[14] Peter Armstrong; Martin L. Wastie, Diagnostic Imaging, Blackwell Scientific Publications,1987.
[15] Anil K. Jain;Farshid Farrokhnia, “Unsupervised texture segmentation using Gabor Filters,” Pattern Recognition, Vol.24, No.12, pp.1167-1186,1991.
[16] Carlos Avils Cruz, “Unsupervised texture segmentation using stochastic version of the EM algorithm and data fusion, ” Pattern Recognition, 1998. Proceedings. Fourteenth International Conference, Vol.2, pp.1005-1009, 1998.
[17] Jia-Lin Chen; Kundu, A, “Unsupervised texture segmentation using multichannel decomposition and hidden Markov models,” IEEE Transactions on Image Processing, Vol.4 5, pp.603-619, May 1995.
[18] Hofmann, T.; Puzicha, J.; Buhmann, J.M, “Unsupervised texture segmentation in a deterministic annealing framework,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol.20 8, pp.803-818, Aug. 1998.
[19] Simon Haykin, Neural Networks, Prentice Hall, 2000.