Data Sheet 1_The capability of deep-radiomics to predict pathological response to neoadjuv

A retrospective multicenter study of 178 patients with non-small cell lung cancer (NSCLC) undergoing neoadjuvant immunochemotherapy. The dataset, created by Yuanxin Ye and last updated in March 2026, includes chest CT images before and after therapy, from which radiomics and deep learning features were extracted to build predictive models for pathological complete response (pCR).

Use Cases

• Training a logistic regression classifier to predict pathological complete response (pCR) based on combined radiomics, deep learning, and clinical features.
• Developing a deep learning model to extract predictive features from post-therapy chest CT images.
• Building a radiomics model using features from pre- and post-therapy CT scans to assess treatment efficacy.
• Creating a combined predictive model that integrates iRECIST criteria with radiomic and deep learning signatures.

Strengths

• Includes data from 178 patients across three medical centers, providing a multi-source cohort.
• Models were validated on internal and external testing cohorts, with reported AUCs up to 0.955.
• Combines multiple data modalities: pre- and post-therapy CT images, extracted radiomics features, deep learning features, and clinical data.

Limitations

• Row count for the underlying image and feature data is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset is small in file size (252.7 KB), suggesting it may contain summary features or a limited sample rather than raw images.

Provenance

Source

figshare, author Yuanxin Ye.

Collection Method

Retrospectively collected patient data and CT images from three medical centers.

Freshness

Last updated 2026-03-18 07:37:44; freshness should be verified.