What is STIC-model?

STIC-model is an automatic diagnostic system for differentiating malignant hepatic tumors based on patients’ multimodal medical data including multi-phase contrast-enhanced computed tomography and clinical features. The model has modular design of SpatialExtractor-TemporalEncoder-Integration-Classifier (STIC), which takes the preprocessed multi-phase CECT images and corresponding encoded clinical features as input, and finally outputs the score for each type of malignant hepatic tumors. SpatialExtractor module is a deep CNN that uses convolutional layers to extract detailed spatial features of CECT images. TemporalEncoder module uses gated RNN to mine the changing pattern among different CECT phases. In the Integration module, the TemporalEncoder module is concatenated with the vector of encoded dummy clinical variables. Finally, in the Classifier module, the Integration output is passed through the softmax activation function to implement the classification task.

How to use STIC-model?

This AI-assisted diagnostic system aims to provide an easy-to-use user interface. The user only needs to upload three phase CECT images (DICOM format) and input corresponding clinical features. STIC-model online tool automatically preprocess input CECT images, including CT HU value conversion and multi-phase CT affine registration. STIC-model finally outputs the score for each type of malignant hepatic tumors.

1.Upload

(a) Three phase CECT DICOM files: non-contrast-enhanced phase (NC phase), arterial phase (ART phase) and portal venous phase (PV phase)

(b) Clinical features: age, gender, platelet (PLT), total bilirubin (TBIL), alpha fetoprotein (AFP), carbohydrate antigen 19-9 (CA19-9), carcinoembryonic antigen (CEA), carbohydrate antigen 125 (CA125) and hepatitis B surface antigen (HBsAg).

2.Analyzing

STIC-model automatedly analyzes uploaded data (Three phase CECT DICOM files and coresspoding clinical features).

3.Results

STIC-model outputs the predicted score for each type of malignant hepatic tumors, including hepatocellular carcinoma (HCC), intrahepatic cholangiocarcinoma (ICC) and metastatic liver cancer (metastasis)

Examples of STIC-assisted Diagnosis

Patients ID NC phase ART phase PV phase Predicted score of STIC-model
test006


HCC score: 0.774
ICC score: 0.102
Metastasis score: 0.124

Diagnosis result: HCC
test008


HCC score: 0.067
ICC score: 0.646
Metastasis score: 0.287

Diagnosis result: ICC
test019


HCC score: 0.173
ICC score: 0.176
Metastasis score: 0.651

Diagnosis result: Metastasis

Citation

Please cite our work if you find this STIC model helpful:

Gao, R., Zhao, S., Aishanjiang, K. et al. Deep learning for differential diagnosis of malignant hepatic tumors based on multi-phase contrast-enhanced CT and clinical data. J Hematol Oncol 14, 154 (2021). [Find the article]

Contact us

Department of Bioinformatics and Biostatistics, School of Life Sciences and Biotechnology, Shanghai Jiao Tong University, Shanghai, China

Ye Jiequan Building, Shanghai Jiao Tong University. 800 Dongchuan Road, Shanghai 200240, China

Author: Ruitian Gao (r.t.gao@sjtu.edu.cn)

Correspondence: Zhangsheng Yu (yuzhangsheng@sjtu.edu.cn)