Minimizing acquisition-related radiomics variability by image resampling and batch effect correction to allow for large-scale data analysis

Abstract

Objective To identify CT-acquisition parameters accounting for radiomics variability and to develop a post-acquisition CT-image correction method to reduce variability and improve radiomics classification in both phantom and clinical applications. Methods CT-acquisition protocols were prospectively tested in a phantom. The multi-centric retrospective clinical study included CT scans of patients with colorectal/renal cancer liver metastases. Ninety-three radiomics features of first order and texture were extracted. Intraclass correlation coefficients (ICCs) between CT-acquisition protocols were evaluated to define sources of variability. Voxel size, ComBat, and singular value decomposition (SVD) compensation methods were explored for reducing the radiomics variability. The number of robust features was compared before and after correction using two-proportion z test. The radiomics classification accuracy (K-means purity) was assessed before and after ComBat- and SVD-based correction. Results Fifty-three acquisition protocols in 13 tissue densities were analyzed. Ninety-seven liver metastases from 43 patients with CT from two vendors were included. Pixel size, reconstruction slice spacing, convolution kernel, and acquisition slice thickness are relevant sources of radiomics variability with a percentage of robust features lower than 80%. Resampling to isometric voxels increased the number of robust features when images were acquired with different pixel sizes (p < 0.05). SVD-based for thickness correction and ComBat correction for thickness and combined thickness–kernel increased the number of reproducible features (p < 0.05). ComBat showed the highest improvement of radiomics-based classification in both the phantom and clinical applications (K-means purity 65.98 vs 73.20). Conclusion CT-image post-acquisition processing and radiomics normalization by means of batch effect correction allow for standardization of large-scale data analysis and improve the classification accuracy. Key Points • The voxel size (accounting for the pixel size and slice spacing), slice thickness, and convolution kernel are relevant sources of CT-radiomics variability. • Voxel size resampling increased the mean percentage of robust CT-radiomics features from 59.50 to 89.25% when comparing CT scans acquired with different pixel sizes and from 71.62 to 82.58% when the scans were acquired with different slice spacings. • ComBat batch effect correction reduced the CT-radiomics variability secondary to the slice thickness and convolution kernel, improving the capacity of CT-radiomics to differentiate tissues (in the phantom application) and the primary tumor type from liver metastases (in the clinical application).

MRI-Based Radiomics Analysis for the Pretreatment Prediction of Pathologic Complete Tumor Response to Neoadjuvant Systemic Therapy in Breast Cancer Patients: A Multicenter Study

“…The in-house developed pipeline started first by applying bias field correction to every image using MIM software (version 6.9.4, Cleveland, OH, USA) to correct for nonuniform grayscale intensities in the MRI caused by field inhomogeneities. Second, in order to minimize acquisition-related radiomics variability, voxel dimensions were standardized across the cohorts to arrive at an isotropic voxel resolution of 1 mm3 by means of cubic interpolation [ 37 ]. Third, to homogenize arbitrary MRI units and clip image intensities to a certain range, a histogram matching technique was applied, adjusting the pixel values of the MR image such that its histogram matched that of the target MR image from the training data cohort [ 38 , 39 , 40 ].…”

Development of CT-Based Imaging Signature for Preoperative Prediction of Invasive Behavior in Pancreatic Solid Pseudopapillary Neoplasm

“…The CT images in arterial and venous phases were firstly resampled isotropically into 1 mm ×1 mm × 1 mm voxel size by using trilinear interpolation, to reduce the heterogeneity resulted from different scanner (24,25). Then the CT images in respective phases were sequentially imported into A.K.…”

“…Considering the radiomics feature could be extracted from the single cross section (two dimensional, 2D) or multi-slices (three dimensional, 3D) of the tumor in CT images, the reported radiomics-based pancreatic cancer studies have either applied 2D segmentation (20) or 3D whole-tumor segmentation (14,(21)(22)(23)(24)(25). However, whether to select 2D regions of interest (ROIs) or 3D ROIs still remains unclear for invasive behavior prediction in pSPN.…”

The application of a workflow integrating the variable reproducibility and harmonizability of radiomic features on a phantom dataset

“…These sources of variation are further referred to as batch effects. ComBat was subsequently adopted in radiomics analysis, and some studies reported that ComBat outperforms other harmonization methods (e.g, histogram-matching, voxel size normalization, and singular value decomposition) in radiomics analyses [15,16]. Several radiomics studies have reported on the successful application of ComBat in removing the differences in RFs introduced by different vendors and acquisition protocols [17][18][19][20][21].…”