TY - JOUR AU - Saidu, C.I. AU - Csato, L. PY - 2019/06/17 TI - Medical Image Analysis with Semantic Segmentation and Active Learning JF - Studia Universitatis Babeș-Bolyai Informatica; Vol 64 No 1 (2019)DO - 10.24193/subbi.2019.1.03 KW - N2 - We address object detection using semantic segmentation and apply it for prostate detection in an MRI data-set. Our detection pipeline uses first a segmentation step followed by a classifier with a convolutional neural network (CNN). Since the segmentation provides a set of unbalanced data-sets – where a high accuracy is difficult to obtain – we leverage the prospect of improving detection accuracy using a Bayesian treatment of deep networks and the possibility of better exploiting the data using active learning. The resulting algorithm is both adaptive and data-efficient: by assuming that from a large pool of data only a few are segmented, the active learning module of the algorithm finds the image that improves most detection accuracy. We test our algorithm on a prostate medical image data-set and show that the active learning-based algorithm performs well in the prostate detection class. The resulting system is invariant to translations within the image and the results show improvements when using the pipeline that includes active learning and CNNs. UR - https://www.cs.ubbcluj.ro/~studia-i/journal/journal/article/view/34