Decomposition of DCE-MRI Data for the Classification of Benign and Malignant Prostates

Aijie Hou,Xinchun Li,Weiwei Du,Yahui Peng

DOI: https://doi.org/10.1145/3468081.3471137

2021-01-01

Abstract:The purpose of the study is to analyze whether certain components can be extracted from dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) for the quantitative classification of prostate with and without prostatic adenocarcinoma. Nonnegative matrix factorization (NMF) was used to separate the components of DCE-MRI images, and the curve with fast wash-in and wash-out features were identified as the characteristic curve. The full width at half maximum (FWHM) of the characteristic curve was calculated as the feature reflecting the sharpness of the curve, and the effectiveness of the feature on the classification of prostates with and without cancer was analyzed. The number of the decomposed components, total imaging time and temporal resolution of DCE-MRI image sequence were investigated to identify their impact on the classification performance. Results showed that the feature of FWHM of the characteristic curve was significantly different in prostates with and without cancer (p = 0.03) and the area under the receiver operating characteristic curve was up to 0.86 ± 0.09, under the optimal number of components, original imaging time and temporal resolution. We conclude that NMF can be used to analyze the components in DCE-MRI data of the prostate. The characteristic curve with the earliest peak rises rapidly and declines rapidly. The FWHM of the characteristic curve can be used to classify the malignant and benign prostates accurately.