Work

Single center experience of the impact of artificial intelligence image analysis software on short-term prognosis of non-small cell lung cancer

Year: 2025

Type: article

Abstract: Objective: To explore the single center experience of the value of artificial intelligence image analysis software in short-term prognostic assessment of non-small cell lung cancer. Methods: Artificial intelligence image analysis software was used to analyze typical cases of NSCLC in our hospital; 450 patients diagnosed with NSCLC were selected as research subjects, Single-factor and multi-factor COX proportional hazards regression were used to analyze the imaging features that affect the short-term survival prognosis (progression/death within 12 months) of NSCLC patients, and the short-term prognostic predictive value of each independent predictor factor was analyzed through the receiver operating characteristic (ROC) curve. Results: The artificial intelligence image analysis software can accurately identify and segment tumor areas, extract key features such as tumor size, shape, and texture, and help doctors diagnose and treat patients more efficiently and accurately. COX regression analysis showed that the maximum diameter of the tumor, spiculation sign, vascular bundle sign, pleural indentation sign, calcification and lymph node metastasis are all imaging features that affect the prognosis of NSCLC patients. The ROC curve shows that the areas under the curve (AUC) of the six factors are 0.676, 0.768, 0.689, 0.696, 0.713, 0.810, respectively, with 95% confidence intervals (95%CI) are =0.576~0.740, 0.663~0.847, 0.610~0.763, 0.590~0.781, 0.614~0.808, 0.716~0.886 respectively. The precision-recall curve of lymph node metastasis and spiculation sign performed best. Even under high recall rate, the precision rate remained above 0.7. The model quality score showed that lymph node metastasis had the highest score (0.74) and spiculation sign was 0.66. Conclusion: The imaging analysis software based on artificial intelligence can significantly improve the accuracy of assessment of NSCLC patients, help improve the short-term prognosis of patients, and has short-term clinical application value.

Source: Frontiers in Oncology

DOI: 10.3389/fonc.2025.1633035

Authors: Luyuan Chang, Siyu Dong, Ailijiang Kadeer, Shilong Song, Tingting Guo, Fei Liu

Institutions: Bengbu Medical College, First Affiliated Hospital of Bengbu Medical College, The Second Affiliated Hospital of Bengbu Medical College

Language: English

Topic: Radiomics and Machine Learning in Medical Imaging

Subfield: Radiology, Nuclear Medicine and Imaging

Field: Medicine

Domain: Health Sciences

SDG: Good health and well-being

Open Access status: gold

Estimated APC paid: $3,801