Purpose <p>To investigate the impact of deep learning reconstruction (DLR) on the subjective and objective image quality of Synthetic diffusion-weighted imaging (sDWI) for prostate lesion detection.</p> Methods <p>47 patients underwent prostate magnetic resonance imaging (MRI) with Low-b-value DWI (i.e., aB50/800_DLR and aB50/800_ConR) reconstructed using DLR and conventional reconstruction (ConR). Additionally, sDWIs at b-value of 1500&#xa0;s/mm² (i.e., sB1500_DLR and sB1500_ConR) were computed from DWI_DLR and DWI_ConR. The imaging datasets was evaluated using 5-point scoring scales for subjective evaluation. Quantitative analysis included signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR) and apparent diffusion coefficient (ADC) values were assessed for benign lesions and prostate cancer (PCa). We compared the consistency and differences in subjective and quantitative evaluations.</p> Results <p>All assessed variables presented good inter-observer consistency {all κ and interclass correlation coefficient (ICC) &gt; 0.60}. Compared with ConR imagings, DLR demonstrated better image quality, including significantly higher 5-point Likert scale score (<i>P</i> &lt; 0.05). The score of sB1500-DLR is highest for overall image quality, artifact, benign lesion conspicuity and PCa conspicuity. For prostate lesions, the SNR and CNR of aB50/800-DLR, sB1500-DLR, and ADC-DLR images were significantly higher than those of ConR (<i>P</i> &lt; 0.05). For benign lesions and PCa, the SNR of aB50-DLR images was higher than that of aB800-DLR and sB1500_DLR (mean ± standard deviation: 56.49 ± 20.63 and 47.44 ± 14.32 respectively), the CNR of sB1500-DLR images was higher than that of aB50-DLR and aB800_DLR {median (interquartile range): 2.09 (4.07) and 4.54 (4.81) respectively}.</p> Conclusion <p>Relative to ConR, sDWI generated from DWI_DLR showed improved image quality and enhanced prostate lesion detection.</p>