Canbilen, Ayse ElifCeylan, Murat2020-03-262020-03-262018978-3-319-65981-7; 978-3-319-65980-02212-9391https://dx.doi.org/10.1007/978-3-319-65981-7_2https://hdl.handle.net/20.500.12395/36274This study presents a decision support system aid to radiologists for defining focal lesions and making diagnosis more accurate by using liver magnetic resonance images. A new method called the complex orthogonal Ripplet-II transform is proposed as a feature extraction procedure. Artificial neural network is utilized to classify the obtained features as a hemangioma or cyst. The results are evaluated with the results of the systems using Ridgelet, Ripplet type-II and orthogonal Ripplet type-II transforms. The highest accuracy ratio (85.3%) and area under curve value (0.92) are achived by the complex orthogonal Ripplet-II transform. The accuracy of the classification procedure is increased up to 95.6% by a combined system that collectively analyzes the results obtained from the artificial neural network outputs of the two methods (Ridgelet and complex orthogonal Ripplet-II transforms). While this combined system is built up of three methods (adding Ripplet type-II), the accuracy rate reaches 97.06% and the area under curve value to 0.99.en10.1007/978-3-319-65981-7_2info:eu-repo/semantics/closedAccessArtificial neural networkBiomedical image classificationComplex orthogonal Ripplet-II transformComplex wavelet transformLiver MR imagingRipplet type-II transformBook Chapter263356N/AWOS:000460338400003N/A