Abstract:Purpose: This study aims to compare the performance of radiomics features and deep learning (DL) features in predicting the prognostic of stage II colorectal cancer (CRC) patients, and fuse these two types of imaging features of primary lesion and peripheral lymph nodes (LNs) to improve the prediction performance. Materials and Methods: We first collected 230 patients (170 in the training set; 60 in the testing set) with clinicopathologically confirmed stage II CRC. Then, we built two CT radiomics models by using primary tumor and LN image features, respectively. The radiomics model development process included tumor or LN segmentation, feature extraction and selection, synthetic minority oversampling technique (SMOTE) oversampling, and support vector machine classifier training/testing. In order to improve the radiomics model performance, we used an information fusion method to build a fusion radiomics model by combining tumor and LN image features. Meanwhile, we applied a transfer learning method to build a deep convolutional neural network (CNN) model to predict the prognosis of CRC patients. Finally, we fused the prediction scores generated by radiomics model and CNN model to improve the prognosis prediction performance. Results: The DFS and OS prediction AUC values generated by fusion model improved to 0.76±0.08 (95% CI: [0.57, 0.89]) and 0.91±0.05 (95% CI: [0.77, 0.97]), which were significantly higher than the AUC values generated by the models using CT radiomics feature and deep image feature individually. Applying survival analysis method, the DFS fusion model and OS fusion model yielded a concordance index (C-index) value of 0.73 (95% CI: [0.56, 0.89]), and 0.9 (95% CI: [0.83, 0.97]), respectively. Conclusion: The present study developed an artificial intelligence (AI) model to predict the prognosis of stage II CRC by radiomics and DL analysis of primary lesion and peripheral LNs in CT scan that can assist clinical screening of high-risk patients with poor prognosis to achieve precision medicine. Funding Statement: National Natural Science Foundation of China (Grant No. 81971687). Declaration of Interests: None of the investigators has any affiliations or financial involvement that conflicts with the material presented in this report. Ethics Approval Statement: Our Institutional Review Board (Fudan University Shanghai Cancer Center Medical Ethics Committee) approved this study and waived the need for informed consent from patients.

MTSC2: Multi-task Survival Prediction on Colorectal CT Images with Clinical Information

A CT-based multitask deep learning model for predicting tumor stroma ratio and treatment outcomes in patients with colorectal cancer: a multicenter cohort study

Multi-Size Deep Learning Based Preoperative Computed Tomography Signature for Prognosis Prediction of Colorectal Cancer

A computed tomography-based multitask deep learning model for predicting tumour stroma ratio and treatment outcomes in patients with colorectal cancer: a multicentre cohort study

Diagnosis and Survival Prediction of Colorectal Cancer Using Deep Learning from the Histopathological HE Images

Predicting Lymph Node Metastasis of Colorectal Cancer in CT Scans Using Attention-Based Multiple Instance Learning.

Predicting rectal cancer prognosis from histopathological images and clinical information using multi-modal deep learning

MPBD-LSTM: A Predictive Model for Colorectal Liver Metastases Using Time Series Multi-phase Contrast-Enhanced CT Scans

A New Automated Prognostic Prediction Method Based on Multi-Sequence Magnetic Resonance Imaging for Hepatic Resection of Colorectal Cancer Liver Metastases

Exploring prognostic biomarkers in pathological images of colorectal cancer patients via deep learning

Special issue "The advance of solid tumor research in China": Prognosis prediction for stage II colorectal cancer by fusing computed tomography radiomics and deep-learning features of primary lesions and peripheral lymph nodes

Assessment of Primary Colorectal Cancer CT Radiomics to Predict Metachronous Liver Metastasis

Knowledge-guided Multi-Task Attention Network for Survival Risk Prediction Using Multi-Center Computed Tomography Images

SurvivalCNN: A deep learning-based method for gastric cancer survival prediction using radiological imaging data and clinicopathological variables

Prognosis Prediction for Stage II Colorectal Cancer by Fusing CT Radiomics and Deep Learning Features of Primary Lesion and Peripheral Lymph Nodes

Adaptive segmentation-to-survival learning for survival prediction from multi-modality medical images

Multi-channel Multi-Task Deep Learning for Predicting EGFR and KRAS Mutations of Non-Small Cell Lung Cancer on CT Images.

Prediction of Microsatellite Instability of Colorectal Cancer Using Multi-Scale Pathological Images Based on Deep Learning

An interpretable integration model improving disease-free survival prediction for gastric cancer based on CT images and clinical parameters

MS-ResNet: disease-specific survival prediction using longitudinal CT images and clinical data

A radiomics and genomics-derived model for predicting metastasis and prognosis in colorectal cancer