영역 성장법을 기반으로 자동적인 임계치 설정을 이용한 미세 석회화 추출
Microcalcification Extraction by Using Automatic Thredholding Based on Region Growing
1
조회
0
다운로드
1
인용
0
참고문헌
초록
본 논문에서는 영역 성장법을 기반으로 자동적인 임계치 설정에 의하여 미세 석회화를 추출하는 방법을 제안하였다. 미세 석회화 후보 영역에서 임계치를 반복적으로 증가시키면서 국부 최대치 화소로부터 영역을 성장시키고 명암 대비와 에지 선예도가 최대일 때 최적의 임계치가 결정됨으로써, 실제 영상에 있어서 효과적으로 미세 석회화를 추출할 수 있었다. 총 299개의 미세 석회화에 대하여 81.5%의 TP(true positive) 비율과 1.1개의 평균 FP(false positive) 개수를 가지는 만족할 만할 결과를 얻었으며, 진단 방사선 전문의의 조기 유방암 진단을 위한 보조 역할이 될 수 있음을 알 수 있었다.
Abstract
In this paper, we proposed the algorithm for detection of microcalcification by automatic threshold decision based on region growing method. The region for optimal threshold is grown from local maximum pixel by increasing repeatedly threshold in microcalcification candidate region. Then, the optimal threshold is automatically decided at the maximum value of the contrast and edge sharpness in this region. Microcalcifications could be efficiently detected as satisfied result that true positive ratio is 81.5% and average false positive numbers are 1.1 about total 299 microcalcifications in real image. In a result, we showed that this algorithm can be used to aid diagnostic-radiologist for the diagnosis of the early phase of breast cancer.
참고문헌 (20)
- "Evaluation of breast microcalcification", 33, pp.1109~1121 (1995)
- "Mammogrhphy in the diagnosis of in situ breast carcinoma", 168, pp.279~280 (1988)
- "Application of shape analysis to mammographic calcifications", 13, pp.263~274 (1994)
- "Computer assisted diagnosis for digital mammography", 10, pp.561~569 (1995)
- "Computer-aided detection of clustered microcalcifications", pp.1~11 (1993)
- "Automatic detection of calcification in mammograms", pp.139~145
- "Wavelets for contrast enhancement of digital mammography", 14, pp.536~550 (1995)
- "Detection of microcalcifications in digital mammograms using wavelets", 17(4), pp.498~509
- "Wavelet transform for detecting micro- calcification in mammograms", 15(2), pp.218~229
- "Detection of Microcalcifications in Digital mammograms using wavelets", 17(4), pp.498~509
- "Image segmentation and analysis via multiscale gradient watershed hierarchies", 8(1), pp.69~79
- "Hybrid image segmentation using watersheds and fast region merging", 7(12), pp.1684~1699
- "Digital image processing" (2001), Prentice Hall
- "Snakes:Active contour models", 1, pp.321~331 (1988)
- "Geodesic active contours", 22(1), pp.61~79 (1997)
- "Fast geodesic active contours", 10(10), pp.1467~1475
- "Detection of mammographic microcalficications by statistical pattern classification of pattern matchine", 18(4) (1997)
- "Computer- aided detection of microcalcifications"
- "Comparative evaluation of pattern recognition techniques for detection of microcalcifications in mammography", 7
- "Comparative evaluation of pattern recognition techniques for detection of microcalcifications in mammography", 7