Texture Analysis Of Synthetic MRI Combined With DWI In The Diagnosis Of Prostate Cancer

Purpose:To evaluate the diagnostic ability of texture analysis of synthetic MRI(SyMRI)combined with diffusion-weighted imaging(DWI)in prostate cancer(PCa)and clinically significant PCa(csPCa).Materials and Methods:This prospective study included 124 consecutive patients with suspected PCa and pathological results between April 2018 and October 2020 at our institution.All patients underwent prostate MRI examinations on a 3.0 Tesla MR scanner.The MRI exams included T1-weighted imaging,T2-weighted imaging,DWI with b value=50,1400 s/mm2 and SyMRI.Apparent diffusion coefficient(ADC)maps were derived from DWI,and quantitative maps including T1,T2,PD maps were derived from SyMRI.We built and compared the following models in PCa and csPCa diagnosis:texture analysis of SyMRI quantitative maps,DWI and ADC,SyMRI quantitative maps combined with DWI and ADC,and mean values of SyMRI quantitative maps,ADC,SyMRI quantitative maps combined with ADC.The diagnostic performance of each model was evaluated and compared by using receiver operating characteristic analysis.Results:A total of 271 lesions were analyzed in this study,including 58 PCa in the peripheral zone(PZ),50 PCa in the transition zone,59 noncancerous peripheral zone(PZ),57 stromal hyperplasia(SH),and 47 glandular hyperplasia(GH).Compared with other models,texture analysis models based on SyMRI quantitative maps combined with DWI and ADC had the highest diagnostic performance.The area under the curves(AUCs)for discriminating PCa from noncancerous PZ,SH,GH were 0.883,0.911,0.965,respectively.The AUCs for discriminating csPCa from noncancerous PZ and low grade(LG)PCa,SH and LG PCa,GH and LG PCa were 0.909,0.885,0.958,respectively.The diagnostic performance of the texture analysis models was similar or higher than the mean value models in the diagnosis of PCa and csPCa.The diagnostic performance of texture analysis models of SyMRI quantitative maps was similar to that of texture analysis models of DWI and ADC in discriminating PCa or csPCa from other lesions,except that the texture analysis model of SyMRI quantitative maps had higher diagnostic performance than the texture analysis model of DWI and ADC in discriminating PCa with GH.Conclusions:The diagnostic performance of texture analysis models of SyMRI quantitative maps for PCa and csPCa was better than or equal to texture analysis models of DWI and ADC.Texture analysis models based on SyMRI quantitative maps combined with DWI and ADC had the highest diagnostic performance and contributed to a more accurate diagnosis of PCa and csPCa.