Objectives: Mammography has an important role in the diagnosis of breast cancer. computer-aided detection (CADe) or computeraided diagnosis (CADx) systems have been developed to improve the capability of radiologists to interpret medical images and to di?erentiate benign and malignant lesions. We studied the capability of ICA as a technique used in CAD systems.Methods: An observational study involving 20 participants (5 radiologists and 15 last year radiology residents) has been initiated. First, they performed assessment of the 60 digitized flm mammograms consisting of 27 normal and 33 abnormal mammograms. Subsequently, from the 33 mammograms with pathological massed, they were asked to di?erentiate benign and malignant lesions (15 and 18 mammograms, respectively).Results: The sensitivity to detect the abnormal lesions in the mammograms was 70% and the specifcity was 72.78%, increased to 74% and 88%, respectively when using ICA. The sensitivity and specifcity to di?erentiate between benign and malignant lesions turned from 40.8 % and 80.3% to 62.5% and 68% after using ICA. The missed diagnostic rate in mammography increased especially in dense breast.Conclusion: ICA improves the radiologists detection of abnormal lesions in mammograms so that it can properly be used as CADe. However, ICA might not be used as a CADx.
Computer-Aided Detection (CAD) Berbasis Intensity Correlation Analysis (ICA)
