Automated detection of the anterior cruciate ligament tear from MRI
Prof. Dr. Ivan Stajduhar
A radiologists work in detecting various injuries or pathologies from radiological scans can be tiresome, time consuming and prone to errors. The field of computer-aided diagnosis aims to reduce these factors by introducing a level of automation in the process. In this seminar we deal with the problem of detecting the presence of the anterior cruciate ligament (ACL) injury in a human knee. We examine the possibility of aiding the diagnosis process by building a decision-support model for detecting the presence of the injury from sagittal plane magnetic resonance (MR) volumes of human knees. Two approaches are under consideration: 1) a linear-kernel support vector machine (SVM) model is learnt from histogram of oriented gradients (HoG) descriptors obtained from the region of interest, enveloping the cruciate ligament area, and 2) a deep convolutional neural network model is learnt form the region of interest.
Ivan Štajduhar obtained his Ph.D. in 2010 at the Faculty of Electrical Engineering and Computing, University of Zagreb, Croatia. His research was in the past primarily focused on developing and enhancing machine learning procedures, mostly to be applied to the field of automated medical diagnosis. He was mostly involved with learning probabilistic graphical models (e.g. Bayesian networks) from various non-trivial problems, including censored data, which is abundant in medicine. He is holding a position of assistant professor at the Department of Computer Engineering, University of Rijeka – Faculty of Engineering, Rijeka, Croatia. He is currently working with Prof. Gozde Unal at solving certain medical image analysis problems under the TUBITAK visiting scholar program at VPALAB, FENS, Sabanci University.