Meeting Banner
Abstract #1615

Application of hierarchical clustering to multi-parametric MR in prostate: Differentiation of tumor and normal tissue with high accuracy

Yuta Akamine1, Yu Ueda1, Yoshiko Ueno2, Takamichi Murakami3, Masami Yoneyama1, Makoto Obara1, and Marc Van Cauteren4

1Philips Japan, Tokyo, Japan, 2Department of Radiology, Kobe University Graduate School of Medicine, Hyogo, Japan, 3Department of Radiology, Kobe University Hospital, Hyogo, Japan, 4Asia Pacific, Philips Japan, Tokyo, Japan

Recently, machine learning (ML) or deep learning (DL) techniques has gain more attention for prostate cancer (PCa) detection. However, DL is often described as “black boxes” and difficult to explain results. In this study, hierarchical clustering (HC),an unsupervised ML technique, was applied to multi-parametric MR to differentiate PCa. DWI (IVIM and DKI) and permeability parameters were used for HC. Comparison of HC methods was conducted. We demonstrated that HC can accurately differentiate PCa and normal tissue (PZ: 97.5%, TZ: 95.7%), with an comparable to state-of-the-art D and K. Contrary to DL, HC produces results that can be interpreted (heatmaps).

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

Join Here