Independent Component Analysis to Detect Clustered Microcalcification Breast Cancers

• Published in: TheScientificWorld Vol. 2012; no. 2012; pp. 1 - 6
• Main Authors: Gallardo-Caballero, R., García-Orellana, C. J., García-Manso, A., González-Velasco, H. M., Macías-Macías, M.
• Format: Journal Article
• Language: English
• Published: Cairo, Egypt Hindawi Puplishing Corporation 01.01.2012; The Scientific World Journal; John Wiley & Sons, Inc; Hindawi Limited
• Subjects: Algorithms; Breast cancer; Breast Neoplasms - diagnosis; Broaching; Calcification; Calcinosis - diagnosis; Datasets; Decision theory; Diagnosis; Digital computers; Feature extraction; Female; Humans; Independent component analysis; International conferences; Mammography; Mammography - methods; Online data bases; Prototypes; Radiographic Image Interpretation, Computer-Assisted - methods; Sensitivity; Sensitivity and Specificity; Studies; Spain; United States > US
• ISSN: 2356-6140; 1537-744X; 1537-744X
• DOI: 10.1100/2012/540457

The presence of clustered microcalcifications is one of the earliest signs in breast cancer detection. Although there exist many studies broaching this problem, most of them are nonreproducible due to the use of proprietary image datasets. We use a known subset of the currently largest publicly available mammography database, the Digital Database for Screening Mammography (DDSM), to develop a computer-aided detection system that outperforms the current reproducible studies on the same mammogram set. This proposal is mainly based on the use of extracted image features obtained by independent component analysis, but we also study the inclusion of the patient’s age as a nonimage feature which requires no human expertise. Our system achieves an average of 2.55 false positives per image at a sensitivity of 81.8% and 4.45 at a sensitivity of 91.8% in diagnosing the BCRP_CALC_1 subset of DDSM.