© FUJIFILM Corporation

Automatic Characterization of Liver Tumors from Multi-phase CT Images


The liver is the largest solid organ in the body which performs vital functions such as creating essential nutrients and removing waste products. Distinguishing the types of liver tumors is important for determining treatment strategy, and dynamic contrast-enhanced CT (DCE-CT) provides useful characteristics for it. In this paper [1], the authors propose a deep learning neural network model for classifying tumor characteristics in DCE-CT images (Figure 2). This model is trained to classify eight features, such as nonrim APHE (nonrim arterial phase hyper enhancement) and washout, which are important for differentiating liver tumors. In addition, they propose three deep neural network classification models that can deal with variable number of input images. Evaluation results showed high discrimination performance exceeding 91% in ROC-AUC on average. In the future, it is hoped that the models will help to deal with liver tumors more accurately and quickly. 

Figure 1. Characteristics of Hemangioma and hepatocellular carcinomas (HCC) in Dynamic contrast-enhanced CT (DCE-CT).

Figure 2. The proposed architectures of models. [based on [1]]
[1] Otani, K. et al. (2023). Automatic Characterization of Liver Tumors from Multi-phase CT Images. Proc. SPIE Medical Imaging (SPIE MI), San Diego, CA, February 2023. 

DOI: http://doi.org/10.1117/12.2648644

CAUTION:This is Fujifilm Global Website. Fujifilm makes no representation that products on this website are commercially available in all countries. Approved uses of products vary by country and region. Specifications and appearance of products are subject to change without notice.