Combining multi-parametric MRI radiomics features with tumor abnormal protein to construct a machine learning-based predictive model for prostate cancer

Abstract

This study aims to investigate the diagnostic value of integrating multi-parametric magnetic resonance imaging (mpMRI) radiomic features with tumor abnormal protein (TAP) and clinical characteristics for diagnosing prostate cancer. A cohort of 109 patients who underwent both mpMRI and TAP assessments prior to prostate biopsy were enrolled. Radiomic features were meticulously extracted from T2-weighted imaging (T2WI) and the apparent diffusion coefficient (ADC) maps. Feature selection was performed using t-tests and the Least Absolute Shrinkage and Selection Operator (LASSO) regression, followed by model construction using the random forest algorithm. To further enhance the model's accuracy and predictive performance, this study incorporated clinical factors including age, serum prostate-specific antigen (PSA) levels, and prostate volume. By integrating these clinical indicators with radiomic features, a more comprehensive and precise predictive model was developed. Finally, the model's performance was quantified by calculating accuracy, sensitivity, specificity, precision, recall, F1 score, and the area under the curve (AUC). From mpMRI sequences of T2WI, dADC(b = 100/1000 s/mm²), and dADC(b = 100/2000 s/mm²), 8, 10, and 13 radiomic features were identified as significantly correlated with prostate cancer, respectively. Random forest models constructed based on these three sets of radiomic features achieved AUCs of 0.83, 0.86, and 0.87, respectively. When integrating all three sets of data to formulate a random forest model, an AUC of 0.84 was obtained. Additionally, a random forest model constructed on TAP and clinical characteristics achieved an AUC of 0.85. Notably, combining mpMRI radiomic features with TAP and clinical characteristics, or integrating dADC (b = 100/2000 s/mm²) sequence with TAP and clinical characteristics to construct random forest models, improved the AUCs to 0.91 and 0.92, respectively. The proposed model, which integrates radiomic features, TAP and clinical characteristics using machine learning, demonstrated high predictive efficiency in diagnosing prostate cancer.

Keywords: Machine learning; Prostate cancer; Radiomics; TAP.

Fig. 1

Diagram for inclusion of patients in the study.

Fig. 2

Flowchart of multimodal diagnostic model construction for prostate cancer.

Fig. 3

Example of ROI delineation. (A, B and C) depict the respective dADC(b = 100/1000 s/mm²) and T2WI and dADC(b = 100/2000 s/mm²) images acquired from the same patient. The green-shaded regions were delineated as ROIs.

Fig. 4

(A, B, and C) represent the effectiveness curves for T2WI, dADC (b = 100/1000 s/mm²) and dADC100/2000 s/mm²), respectively. (D, E, and F) depict the convergence graphs of the corresponding sequence feature coefficients, where Lasso model was employed along with cross-validation for feature selection.

Fig. 5

ROC curves of random forest. (A) ROC curves for radiomics-based T2WI model. (B) ROC curves for radiomics-based dADC(b = 100/1000 s/mm²) model. (C) ROC curves for radiomics-based dADC(b = 100/2000 s/mm²) model. (D) ROC curves for TAP and additional clinical characteristics model. (E) ROC curves for radiomics-based T2WI and dADC model. (F) ROC curves for radiomics-based dADC, TAP and clinical characteristics model. (G) ROC curves for radiomics-based T2WI, dADC, TAP and clinical characteristics model. (Cm: Clinical characteristics model, including TAP. T2: T2WI)

Fig. 6

Overall structure of the adopted random forest.

Fig. 7

Heatmap for normalized feature value distribution of the extracted 31 features between PCa and BPH. The red regions indicate positive correlations with PCa features, while the blue regions represent negative correlations with nPCa features. The color intensity is proportional to the strength of correlation (darker hues denote strong correlations, whereas lighter hues indicate weak correlations).