Purpose
Magnetic resonance imaging is increasingly used in the diagnostic workup of prostate cancer. The is based on at least T2-weighted and diffusion weighted imaging (biparametric MRI, bpMRI) and lesions are characterised accordind to the Prostate Imaging-Reporting and Data System (PI-RADS). However, a significant proportion of lesion with intermediate or high probability of the presence of malignancy represent benign conditions such as chronic prostatitis. This study aimed toevaluate if texture analysis (TA) can improve the prediction of intermediate and high grade prostate cancer using biparametric MRI...
Methods and materials
The study group consisted of 405 retrospectively included patients who 2015-2018 underwent bpMRI (3T or 1,5T) of the prostate followed by MR/US software fusion guided biopsi of 566 PI-RADS (version 2) 3-5 lesions. A region of interest was delineated for each lesion in T2-weighted images and ADC-maps enabling extraction of data for TA (area, entropy, maximum, minimum, Median, mean, kurtosis, contrast, energy, homogeneity). Logistic regression models for prediction of the presence of intermediate and high grade prostate (Gleason ≥ 7) were determined using clinical data...
Results
PI-RADS,PSA-density and age were independent predictors ofthe presence of intermediate and high grade prostate cancer (Gleason ≥ 7 ) for the prostate as a whole and PZ, whilePI-RADS andPSA-density were independent predictors for TZ. No single texture feature showed to be an independent predictor.
ROC analysis showed good performance of the models using clinical routine information only (prostate AUC 0.81, PZ AUC 0.80,TZ AUC 0.86). By adding TA an improved performance was seen for the whole prostate and PZ, but not for TZ(prostate AUC 0.84,...
Conclusion
Prediction ofGleason ≥ 7 prostate cancer using bpMRI in combination with clinically available data can be improved by adding TA. However, no single TA component could be identified as an independent predictor.
Personal information and conflict of interest
W. Krauss; Örebro/SE - nothing to disclose P. Thunberg; Örebro/SE - nothing to disclose J. Frey; Örebro/SE - nothing to disclose J. Heydorn Lagerlöf; Örebro/SE - nothing to disclose M. Lidén; Örebro/SE - nothing to disclose
References
1.Weinreb JC, Barentsz JO, Choyke PL, Cornud F, Haider MA, Macura KJ, Margolis D, Schnall MD, Shtern F, Tempany CM, Thoeny HC, Verma S. (2016)PI-RADS Prostate Imaging - Reporting and Data System: 2015, Version 2.ur Urol. 2016 Jan;69(1):16-40.