# Decoding the Rotation Effect: A Retrospective Analysis of Lesion Orientation and Its Impact on Wavelet-Based Radiomics Feature Extraction and Lung Cancer Classification

## TL;DR

This study shows that rotating lung cancer lesions in CT scans can change radiomic features, especially those based on wavelet decomposition, affecting model accuracy.

## Contribution

The study reveals that wavelet-based radiomic features are more sensitive to lesion orientation than non-wavelet features, impacting classification performance.

## Key findings

- Wavelet decomposition features showed significant instability with rotation, affecting 23.7% of features compared to 0.5% for non-WD features.
- Model performance for NSCLC subtype classification dropped significantly with increased rotation in WD-based models.
- Non-WD models showed no significant performance change with rotation.

## Abstract

Wavelet decomposition (WD), widely used in radiomics, redistributes information among derived wavelet components when the input is rotated. This redistribution may alter predictions for the same lesion when scanned at different angles. Despite its potential significance, this vulnerability has frequently been overlooked in radiomic studies while its impact remains poorly understood. Therefore, this study aims to investigate how variations in lesion orientation affect both WD and non-WD radiomic feature values, and subsequently, model performance. We analyzed CT radiomics of primary non-small-cell lung cancer (NSCLC). Prior to feature extraction, we introduced random rotations ranging from 5° to 80° to the tumors. Their effects were quantified by evaluating the percentage difference (\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\%\Delta$$\end{document}%Δ) between the rotated and unrotated feature values, and validated using Spearman’s rank test. Additionally, radiomics models were trained to discriminate between three histological subtypes of NSCLC using the original features, and then tested on rotated inputs. The correlation between the model accuracies and the degree of rotation was again evaluated using Spearman’s rank test. Four-hundred nineteen NSCLC patients (mean age: 68.1 ± 10.1, 289 men) were evaluated. Significant correlations between feature values and rotations (Spearman’s correlation [CC] magnitude ≥ 0.1, p < .05) were found in 23.7% (176/744) of the WD and 0.5% (5/930) of the non-WD texture features. Significant association between performance and rotation was observed in WD-based models built to discriminate between NSCLC histological subtypes (CC =  − 0.44, p < .001) but not in non-WD-based models (CC = 0.03, p = 0.07). Input lesion orientation affects radiomic feature values and model reproducibility. WD features exhibited significantly greater instability to orientation variations compared to non-WD features.

The online version contains supplementary material available at 10.1007/s10278-025-01520-8.