Meeting Banner
Abstract #3835

Unsupervised quality control of prostate MRSI using Non Negative Matrix Factorization

Nassim Tayari 1 , Anca R. Croitor Sava 2 , Diana M. Sima 2 , Sabine Van Huffel 2 , and Arend Heerschap 1

1 Department of Radiology and Nuclear Medicine, Radboud University Medical Center, Nijmegen, Netherlands, 2 Department of Electrical Engineering, Katholieke Universiteit Leuven, Leuven, Belgium

An automated quality control algorithm plays an important role in automation of the analysis of MRSI data of the prostate cancer patients. In this work we present an automated unsupervised quality control algorithm for 3D 1H MRSI data sets. The method is based on feature extraction using Non Negative Matrix Factorization(NNMF). Consensus decisions of spectral quality judged by four experts is used as a Gold Standard for performance evaluation showing that the algorithm can separate good quality from bad quality spectra with 90% sensitivity and 90% specificity.

This abstract and the presentation materials are available to members only; a login is required.

Join Here