Larue RTHM, et al. – Researchers explored 4D CT-scans (4DCT) imaging to assess radiomics feature stability. As an alternative to eliminate unstable radiomics features as first step in a feature selection procedure, they suggested the use of 4DCT–scans. Moreover, feature robustness was tumour–site specific and independent of prognostic value.

Methods

• In test–retest CT-scans of 26 non-small cell lung cancer (NSCLC) patients and 4DCT-scans (8 breathing phases) of 20 NSCLC and 20 oesophageal cancer patients, they estimated the 1045 radiomics features of the primary tumours.
• Robust features were determined by using a concordance correlation coefficient (CCC) >0.85.
• By using univariate cox regression, correlation with prognostic value was tested in 120 oesophageal cancer patients.

Results

• As compared to wavelet-filtered features, features based on unfiltered images demonstrated greater robustness.
• As per observation, in total 63/74 (85%) unfiltered features and 268/299 (90%) wavelet features stable in the 4D-lung dataset were also stable in the test–retest dataset.
• 397/1045 (38%) features were robust in oesophageal cancer.
• 108 features were significantly associated with overall-survival, among those.