2016/01/26 by Fausto Milletarì, Seyed‐Ahmad Ahmadi, Milletari, Fausto +19
Computer Science · Engineering · Neuroscience · #Advanced Neural Network Applications #Brain Tumor Detection and Classification #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Medical Image Segmentation Techniques #Medical Imaging and Analysis
paper · pdf · doi:10.48550/arxiv.1601.07014
openalex publication_date 2016/01/26 · openalex created_date 2022/10/03 · openalex updated_date 2026/07/28
In this work we propose a novel approach to perform segmentation by\nleveraging the abstraction capabilities of convolutional neural networks\n(CNNs). Our method is based on Hough voting, a strategy that allows for fully\nautomatic localisation and segmentation of the anatomies of interest. This\napproach does not only use the CNN classification outcomes, but it also\nimplements voting by exploiting the features produced by the deepest portion of\nthe network. We show that this learning-based segmentation method is robust,\nmulti-region, flexible and can be easily adapted to different modalities. In\nthe attempt to show the capabilities and the behaviour of CNNs when they are\napplied to medical image analysis, we perform a systematic study of the\nperformances of six different network architectures, conceived according to\nstate-of-the-art criteria, in various situations. We evaluate the impact of\nboth different amount of training data and different data dimensionality (2D,\n2.5D and 3D) on the final results. We show results on both MRI and transcranial\nUS volumes depicting respectively 26 regions of the basal ganglia and the\nmidbrain.\n