A dynamic-static combination model based on radiomics features for prostate cancer using multiparametric MRI

Area of Science:

• Radiology
• Medical Imaging
• Oncology

Background:

• Prostate cancer (PCa) detection relies on accurate imaging techniques.
• Multiparametric magnetic resonance imaging (mpMRI) is crucial for PCa diagnosis.
• Radiomics offers advanced feature extraction from medical images.

Purpose of the Study:

• To develop a novel dynamic mpMRI radiomics method for PCa detection.
• To establish a combined model integrating dynamic and static radiomics features.
• To evaluate the diagnostic performance of the proposed models for PCa.

Main Methods:

• A radiomics workflow was applied to 31,872 mpMRI images from 166 patients.
• Dynamic radiomics features (standard discrete, parameter, relative change rate) were constructed from intravoxel incoherent motion diffusion-weighted imaging (IVIM-DWI) at various b-values.
• Combined models integrated dynamic features with static features from apparent diffusion coefficient (ADC) and T2-weighted imaging (T2WI).

Main Results:

• Dynamic radiomics models demonstrated strong PCa identification potential, with AUCs up to 90.78% (Db_SD model).
• The proposed dynamic models outperformed the traditional ADC model (AUC 75.48%).
• Combined models integrating T2WI static features and dynamic features achieved the highest performance, with AUCs reaching 92.90% (Db_SD + T2WI).

Conclusions:

• The dynamic-static combination model based on the novel mpMRI radiomics method is effective for PCa identification.
• This approach shows significant potential for personalized PCa diagnosis and management.
• The developed method offers a promising tool for improving prostate cancer detection accuracy.