Standardization of imaging features for radiomics analysis

Area of Science:

• Radiology
• Medical Imaging
• Oncology

Background:

• Radiomics offers a noninvasive method to characterize tumors.
• Accurate prediction of non-small cell lung cancer (NSCLC) histology is crucial for treatment planning.

Purpose of the Study:

• To evaluate the impact of data standardization on radiomics analysis for NSCLC.
• To assess the effectiveness of different standardization techniques in improving prediction accuracy.

Main Methods:

• Extracted 476 imaging features from two CT databases: 40 NSCLC patients (training) and 29 early-stage NSCLC datasets (testing).
• Applied min-max normalization, z-score normalization, and principal component analysis (PCA) whitening for data standardization.
• Evaluated prediction model performance using the area under the receiver operating characteristic (ROC) curve.

Main Results:

• All standardization strategies enhanced the accuracy of NSCLC histology prediction.
• The area under the ROC curve improved to 0.789 with min-max normalization and 0.785 with PCA whitening.
• Radiomics analysis confirmed that standardized, robust features possess high prognostic power for early-stage NSCLC.

Conclusions:

• Data standardization in the feature space is effective in improving radiomics analysis performance.
• Standardized radiomics features can reliably predict early-stage NSCLC histology subtypes.
• This study highlights the importance of data preprocessing for robust radiomics applications in oncology.