Search

Stability analysis remains a fundamental step in developing a successful imaging biomarker to personalize oncological strategies. This study proposes an in silico contour generation method for simulating segmentation variations to identify stable radiomic features. Ground-truth annotation provided for the whole prostate gland on the multi-parametric MRI sequences (T2w, ADC, and SUB-DCE) were perturbed to mimic segmentation differences observed among human annotators. In total, we generated 15 synthetic contours for a given image-segmentation pair. One thousand two hundred twenty-four unfiltered/filtered radiomic features were extracted applying Pyradiomics, followed by stability assessment using ICC(1,1). Stable features identified in the internal population were then compared with an external population to discover and report robust features. Finally, we also investigated the impact of a wide range of filtering strategies on the stability of features. The percentage of unfiltered (filtered) features that remained robust subjected to segmentation variations were T2w-36% (81%), ADC-36% (94%), and SUB-43% (93%). Our findings suggest that segmentation variations can significantly impact radiomic feature stability but can be mitigated by including pre-filtering strategies as part of the feature extraction pipeline.

Objectives: To evaluate the inter-observer reliability of modified Response Evaluation Criteria In Solid Tumours (mRECIST) of patients with hepatocellular carcinoma (HCC) undergoing neo-adjuvant treatments before liver transplant (LT). The agreement of tumor number, size, transplant criteria, and the radiological-pathological concordance were also assessed., Methods: A total of 180 radiological studies before/after neo-adjuvant therapies performed on 90 patients prior to LT were reviewed from three expert centers. Kappa-statistic and intraclass correlation (ICC) were evaluated on mRECIST and on tumoral features. Complete radiological response (CR) was compared with complete pathological response (CPR)., Results: Before neo-adjuvant therapies, the agreement on tumor number, size, and transplant criteria ranged from moderate (defined as ICC of 0.41-0.60) to almost perfect (ICC of 0.81-0.99), being higher with magnetic resonance imaging (MRI) than CT (0.657-0.899 and 0.422-0.776, respectively). After neo-adjuvant therapies, the agreement decreased, as ICCs ranged between 0.518 and 0.663 with MRI and between 0.508 and 0.677 with CT. Concordant mRECIST pairs were 201 of 270 reviews (76.3%) with a kappa of 0.648 indicating substantial agreement. When the three observers completely agreed on CR, the positive predictive value for CPR was 51.6%. The negative predictive value was 94.2% with a kappa of 0.512 indicating fair agreement between radiology and pathology., Conclusions: mRECIST agreement was substantial among the three observers involved. The agreement on tumor number, size, and transplant criteria ranged from moderate to almost perfect, with the highest ICCs obtained with MRI before neo-adjuvant therapies. Finally, the predictive value of mRECIST in the diagnosis of CPR was only fair., Key Points: • The review of 180 radiological exams of patients with hepatocellular carcinoma before and after neo-adjuvant therapies showed that the concordance among three different raters on mRECIST diagnosis was substantial. • The inter-observer reliability on fulfilment of transplant criteria slightly decreased when evaluated through CT and after loco-regional therapies. • The radiological diagnosis of complete response after neo-adjuvant therapies was predictive of complete pathological response in only 51.6% of cases., (© 2021. European Society of Radiology.)