A NOVEL REPRODUCIBLE RADIOMICS FRAMEWORK ON PSMA PET/CT FOR LESION BURDEN QUANTIFICATION: TOWARD CLINICALLY ROBUST IMAGING BIOMARKER IN METASTATIC PROSTATE CANCER
Keywords:
Prostate cancer; Radiomics; PSMA PET/CT; Lesion burden; Reproducible features.Abstract
Background: Radiomics is a powerful approach in modern cancer management which offers significant potential to transform cancer management through quantitative non-invasive biomarkers in oncology and its clinical application of radiomics features. In metastatic prostate cancer, assessing disease burden is crucial for prognosis and treatment planning .However, it remains unclear whether reproducible radiomic features can reliably reflect lesion burden, including lesion number and total volume. This study developed a novel reproducibility-driven radiomics framework for PSMA PET/CT which evaluates the correlation between stable radiomic features and lesion burden.
Methods: In this retrospective study, patients with metastatic prostate cancer who underwent PSMA PET/CT were included. A predefined set of reproducible radiomic features, which had previously been validated for consistency, was extracted from the metastatic lesions. Lesion burden was defined by both the number of metastatic sites and cumulative lesion volume. Correlation analysis were performed using Pearson correlation to assess associations between radiomic features and lesion burden.
Results: A total of 117 metastatic prostate cancer patients were analyzed. The median age was 67.92 years. Several reproducible radiomic features demonstrated significant correlations with both lesion number and total lesion volume. The strongest associations were observed for Energy (abdominal lymph) with correlation coefficients ranging from 0.47
Conclusion: Our results show various relations between reproducible radiomic features derived from PSMA PET/CT are associated and lesion burden in metastatic prostate cancer. These findings support the potential role of radiomics as a non-invasive biomarkers for aggressiveness of disease. Furthermore, using Pearson correlation to analyze clinical relevance of these mentioned radiomics features, is an important consideration for users conducting radiomics studies in prostate cancer researches.
