Learning objectives
Prostate cancer (PCa) is the second most commonly diagnosed cancer in men and a significant cause of cancer-related mortality worldwide [1]. Early detection is crucial, as the five-year survival rate is high when PCa is identified early [2]. However, definitive diagnosis relies on invasive biopsies, which are prone to variability. Multiparametric magnetic resonance imaging (mpMRI) has become a vital tool for initial evaluation and identifying suspicious lesions.Traditional diagnostic methods include digital rectal examination (DRE), prostate-specific antigen (PSA) testing, and transrectal ultrasound (TRUS)-guided biopsy. The Gleason...
Background
This study utilized data from two sources: a private multicenter cohort and the PROSTATEx2 dataset. These datasets included a total of 207 prostate cancer lesions, of which 152 were classified as csPCa and 55 as cinsPCa. The inclusion of multicenter data provided a robust framework for evaluating the impact of variability in imaging protocols and scanner types.Radiomics features were extracted from ADC maps using Pyradiomics after comprehensive preprocessing. Preprocessing steps included resampling voxel dimensions to 1 mm³ to ensure consistency, intensity normalization, and the exclusion...
Findings and procedure details
The radiomics-ADCratio model, which combined RFE and RF with ComBat harmonization, emerged as the best-performing model. This approach achieved an AUC-PR of 0.92±0.04 and an F1 score of 0.86±0.04. It outperformed both the radiomics-only model (AUC-PR: 0.91±0.06, F1: 0.84±0.03) and the ADC-only model (AUC-PR: 0.71±0.13, F1: 0.58±0.15).The integration of radiomics features and the ADCratio provided supplementary information, capturing both the complex textural characteristics of tumor microarchitecture and the simpler quantitative measure of tumor aggressiveness. This combination improved diagnostic accuracy and robustness, particularly in multicenter settings.ComBat...
Conclusion
This study highlights the potential of integrating radiomics features with the ADCratio for the accurate differentiation of csPCa and cinsPCa. The radiomics-ADCratio model, enhanced by robust feature selection and ComBat harmonization, demonstrated superior diagnostic performance compared to standalone radiomics or ADCratio models. These findings underscore the importance of combining complex radiomics features with simple, interpretable biomarkers to capture complementary information and improve diagnostic precision.ComBat harmonization was pivotal in addressing inter-scanner variability, ensuring model generalizability across diverse datasets. This capability is particularly critical for the deployment...
Personal information and conflict of interest
D. Samaras:
Nothing to disclose
G. Agrotis:
Nothing to disclose
A-C. Vamvakas:
Nothing to disclose
M. Vakalopoulou:
Nothing to disclose
M. Vlychou:
Nothing to disclose
K. Vassiou:
Nothing to disclose
V. Tzortzis:
Nothing to disclose
C-I. Tsougos:
Nothing to disclose
References
[1] Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229–63. https://doi.org/10.3322/caac.21834.[2] Siegel RL, Giaquinto AN, Jemal A. Cancer statistics, 2024. CA Cancer J Clin 2024;74:12–49. https://doi.org/10.3322/caac.21820.[3] Turkbey B, Brown AM, Sankineni S, Wood BJ, Pinto PA, Choyke PL. Multiparametric prostate magnetic resonance imaging in the evaluation of prostate cancer. CA Cancer J Clin 2016;66:326–36. https://doi.org/10.3322/caac.21333.[4] Kuhl CK, Bruhn...
