RaFIST: A Radiomics-Based Tool for Intratumoral Fibrosis Stratification in Non-Small Cell Lung Cancer

Development and Prognostic Validation of RaFIST: A Radiomics-Based Tool for Intratumoral Fibrosis Stratification in Non-Small Cell Lung Cancer

Patients with suspected Lung cancer who underwent contrast-enhanced CT and pathological examinations at Center 1 between January 2011 and January 2020 and Center 2 between September 2017 and Januray 2020 were eligible for inclusion in this study.The Clinical data, preoperative clinical information, laboratory results, CT images and pathological sections were collected. The investigators also collected the disease-free survival and overall survival time. On the Deepwise multi-modal research platform, the images were semi-automatically segmented and PyRadiomics was used to extract the radiomic features. Fibrosis quantification was performed on picrosirius red (PSR)-stained tissue sections using color deconvolution and binarized collagen signal analysis in ImageJ software. The investigators developed RaFIST, a radiomics-based stratification tool, to noninvasively quantify intratumoral fibrosis in non-small cell lung cancer (NSCLC) using contrast-enhanced CT imaging. And it was further tested on the held-out external test cohort. Discrimination was assessed by using the C-index and area under the receiver operating characteristic curve (AUC).

Inclusion Criteria:

• Solid pulmonary nodule detected on CECT within 3 weeks pre-resection
• No history of previous treatment such as chemotherapy, or radiotherapy

Exclusion Criteria:

• Unavailable pathological sections
• Missing CT images or hard-to-annotate CT images
• Undergone anticancer therapy before CT
• Loss to follow-up
• Concurrent other malignant tumors