Prediction of ERCC1 Expression Level in Nasopharyngeal Carcinoma Based on Multi-sequence MRI Radiomics and Habitat Analysis

Objective To explore the value of multi-sequence MRI-derived radiomics and habitat analysis for predicting excision repair cross-complementation group 1 (ERCC1) expression in nasopharyngeal carcinoma (NPC). Methods In this retrospective study, 171 patients with non-keratinizing undifferentiated NPC without distant metastasis were enrolled. Patients were stratified into high-ERCC1 (n = 75) and low-ERCC1 (n = 96) groups and randomly split into training (n = 136) and test (n = 35) sets at an 8:2 ratio. Radiomic and habitat features were extracted from T1WI, CE-T1WI and T2WI to calculate radscore and habitat radscore. We constructed radiomics, habitat, clinical, radiomics-clinical and habitat-clinical combined models. Results A total of 14 radiomic features and 20 habitat features were finally selected. Clinical stage (OR = 1.828, 95%CI: 1.011–3.305, P = 0.046) was identified as an independent predictor of high ERCC1 expression. Based on this predictor, five models were constructed: radiomics model, habitat model, radiomics-clinical model, habitat-clinical model and clinical model. Their area under the receiver operating characteristic curve (AUC) values in the training set were 0.825, 0.917, 0.830, 0.922 and 0.605, respectively; the corresponding AUC values in the test set were 0.770, 0.787, 0.770, 0.807 and 0.537. All radiomics-based models yielded higher AUC values than the clinical model (De Long test, all P < 0.05). The habitat model outperformed the conventional radiomics model in the training set (De Long test, P < 0.05)). Conclusions The Multi-sequence MRI-based radiomics and habitat analysis enable effective prediction of ERCC1 expression in NPC patients. The habitat-clinical model achieved the best predictive performance and may support clinical decision-making and prognostic evaluation. The habitat model was superior to conventional radiomics models, suggesting that refined sub-regional segmentation can help boost radiomics prediction accuracy.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-19
DOI
https://doi.org/10.21203/rs.3.rs-11465121/v1
Primary Topic
Radiomics and Machine Learning in Medical Imaging
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Prediction of ERCC1 Expression Level in Nasopharyngeal Carcinoma Based on Multi-sequence MRI Radiomics and Habitat Analysis

Xishan Chen, haixin huang, Ying Lu, Wenjie Zhang
Research Square
Radiomics and Machine Learning in Medical Imaging
preprint

Prediction of ERCC1 Expression Level in Nasopharyngeal Carcinoma Based on Multi-sequence MRI Radiomics and Habitat Analysis

Xishan Chen, haixin huang, Ying Lu, Wenjie Zhang
preprint en

Abstract

Objective To explore the value of multi-sequence MRI-derived radiomics and habitat analysis for predicting excision repair cross-complementation group 1 (ERCC1) expression in nasopharyngeal carcinoma (NPC). Methods In this retrospective study, 171 patients with non-keratinizing undifferentiated NPC without distant metastasis were enrolled. Patients were stratified into high-ERCC1 (n = 75) and low-ERCC1 (n = 96) groups and randomly split into training (n = 136) and test (n = 35) sets at an 8:2 ratio. Radiomic and habitat features were extracted from T1WI, CE-T1WI and T2WI to calculate radscore and habitat radscore. We constructed radiomics, habitat, clinical, radiomics-clinical and habitat-clinical combined models. Results A total of 14 radiomic features and 20 habitat features were finally selected. Clinical stage (OR = 1.828, 95%CI: 1.011–3.305, P = 0.046) was identified as an independent predictor of high ERCC1 expression. Based on this predictor, five models were constructed: radiomics model, habitat model, radiomics-clinical model, habitat-clinical model and clinical model. Their area under the receiver operating characteristic curve (AUC) values in the training set were 0.825, 0.917, 0.830, 0.922 and 0.605, respectively; the corresponding AUC values in the test set were 0.770, 0.787, 0.770, 0.807 and 0.537. All radiomics-based models yielded higher AUC values than the clinical model (De Long test, all P < 0.05). The habitat model outperformed the conventional radiomics model in the training set (De Long test, P < 0.05)). Conclusions The Multi-sequence MRI-based radiomics and habitat analysis enable effective prediction of ERCC1 expression in NPC patients. The habitat-clinical model achieved the best predictive performance and may support clinical decision-making and prognostic evaluation. The habitat model was superior to conventional radiomics models, suggesting that refined sub-regional segmentation can help boost radiomics prediction accuracy.

Research Square
Fourth Affiliated Hospital of Guangxi Medical University (CN)
Radiomics and Machine Learning in Medical Imaging
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon