Scope
Mammography has been shown to be the most effective imaging modality for the early detection of breast cancer; it has been shown that when mammograms are independently read by two radiologists (double reading), an increment of sensitivity of up to 15% can be obtained. Because double reading by two radiologists skilled in mammography is difficult to realize in most radiological centres, in recent years, several computerized systems have been developed to aid diagnosis by radiologists working in mammography. In such computer-aided diagnosis (CAD) systems, it is of particular interest to detect microcalcifications, due to the fact that, in studies analyzing the mammographic nature of missed cancers, clustered microcalcifications composed 19% – 31% of lesions missed in screening. In the study by Bird et al., 18% (six of 33) of the missed malignancies presented as clustered microcalcifications were simply overlooked by the radiologists. Microcalcifications may be inconspicuous, owing to their small size and/or obscuration by overlying fibroglandular tissues, and may be missed even by a diligent radiologist.
In this research, a two-stage method for detecting microcalcifications in mammograms is presented. In the first stage, the determination of the candidates for microcalcifications is performed. For this purpose, a 2D linear prediction error filter is applied, and for those pixels where the prediction error is larger than a threshold, a statistical measure is calculated to determine whether they are candidates for microcalcifications or not. In the second stage, a feature vector is derived for each candidate, and after a classification step using a support vector machine, the final detection is performed. The algorithm was tested with 40 mammographic images, from Screen Test: The Alberta Program for the Early Detection of Breast Cancer with 50 mm resolution, and the results were evaluated using a free-response receiver operating characteristics curve. Two different analyses have been performed: an individual microcalcification detection analysis and a cluster analysis. In the analysis of individual microcalcifications, detection sensitivity values of 0.75 and 0.81 were obtained at 2.6 and 6.2 false positives per image, on the average, respectively. The best performance is characterized by a sensitivity of 0.89, a specificity of 0.99, and a positive predictive value of 0.79. In cluster analysis, a sensitivity value of 0.97 was obtained at 1.77 false positives per image and a value of 0.90 was achieved at 0.94 false positive per image.
Collaborations
Department of Electrical and Computer Engineering, University of Calgary
Department of Radiological Sciences and. Diagnostic Imaging. Foothills Hospital, Calgary
Funding
Algorithms based on natural processes applied to prediction, decision and Communications, CICYT
Images
Publications
- Begoña Acha Piñero, Maria Carmen Serrano Gotarredona, R.M. Rangayyan, J.E. Leo Desautels: Detection of Microcalcifications in Mammograms. Recent Advances in Breast Imaging, Mammography and Computer-Aided Diagnosis of Breast Cancer. Bellingham, EE.UU. SPIE- the International Society for Optical Engineering. 2006. Pag. 292-308
- Jasjit S. Suri, R Chandrasekhar, Nico Lanconelli, Renato Campanini, Matteo Roffilli, et. al.: The Current Status and Likely Future of Breast Imaging CAD. Recent Advances in Breast Imaging, Mammography and Computer-Aided Diagnosis of Breast Cancer. Bellingham, EE.UU. SPIE- the International Society for Optical Engineering. 2006. Pag. 901-961
- Begoña Acha Piñero, Maria Carmen Serrano Gotarredona, R.M. Rangayyan, J.E. Leo Desautels: Detection of Microcalcifications in Mammograms Using Error of Prediction and Statistical Measures. Journal of Electronic Imaging . Vol. 18. Núm. 1. 2009. Pag. 013011-1-013011-10
