MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel
  1. Home
  2. Meeting Program
  3. All Sessions
  4. All Multi-Disciplinary General ePoster Discussion
  5. The Stability of CT Radiomics Feature Using Repeated Scans

The Stability of CT Radiomics Feature Using Repeated Scans

C Ma1*, (1) Rutgers Cancer Institute of New Jersey

Presentations

(Tuesday, 7/16/2019) 9:30 AM - 10:00 AM

Room: Exhibit Hall | Forum 6

Purpose: To evaluate the stability of CT radiomics features with different image acquisitions in terms of dose levels and reconstruction algorithms.

Methods: A cylindrical phantom (Helical CT Phantom, CIRS Inc.) was scanned on a 128-slice Definition Flash (Siemens Healthcare) CT scanner 100 times with different effective mAs (60 mAs, 120 mAs, and 240 mAs corresponding to CTDIvol values of 4 mGy, 8 mGy, and 16 mGy) and different reconstruction methods (filter back projection as FBP, iterative reconstruction as IR with two different noise reduction levels). The phantom contains water equivalent material and liver-tissue equivalent material. Two regions of interest (ROIs) including both materials but at different physical locations were selected for analysis. A total of 64 first-order histogram-based and second-order texture-based radiomics features (Gray Level Co-occurrence Matrix as GLCM, and Gray Level Run Length Matrix as GLRLM) were extracted using an open-source CERR radiomics platform. For each dose level and reconstruction combination, the mean, standard deviation and coefficient of variation (COV) were calculated from all the repeated scans.

Results: First-order histogram-based CT radiomics features showed less COV in general compared to second-order texture-based features. Different image reconstruction algorithms were not interchangeable and FBP algorithm generated more stable radiomics features compared to IR algorithms. First-order histogram-based features were more stable with higher dose-level due to better signal-to-noise (SNR) level. But no such trend was observed for second-order texture features.

Conclusion: First-order features show less variation with repeated scans, and the COV increases with iterative reconstruction and noise reduction level. While COV for first-order features decreases with increased image dose, the effect of dose or mAs level on second-order feature stability needs further investigation.

Keywords

CT, Feature Selection, Noise

Taxonomy

IM/TH- Image Analysis (Single modality or Multi-modality): Imaging biomarkers and radiomics

Contact Email

Additional Links