Comparison of the quality of prostate images from different diffusion-weighted imaging sequences: Single-shot echo-planar, reduced field-of-view, readout-segmented multi-shot

INTRODUCTION

The diffusion-weighted imaging (DWI) sequence is one of the most important sequences used for semiquantitative diagnosis of prostate lesions in the Prostate Imaging Reporting and Data System Version 2.1 (PI-RADS v 2.1).^[1,2] The image quality (IQ) of DWI images and the corresponding apparent diffusion coefficient (ADC) images have a significant impact on the PI-RADS v2.1 category, especially in the evaluation of the prostate nodule in the peripheral zone.^[3,4] Malignant prostate nodules exhibit markedly hyperintense signal on DWI images and markedly focal hypointense signal on ADC images due to the limited diffusion of water molecules caused by high cell density,^[5-7] but the signal-to-noise ratio (SNR) of the images is low.^[8] Therefore, high-quality DWI images have significant clinical significance in improving the diagnostic efficacy of PI-RADS v2.1 for prostate lesions.

The commonly used scanning protocols for DWI sequences include single-shot echo-planar DWI (ss-DWI), reduced field-of-view DWI (rf-DWI), and readout-segmented multi-shot DWI (rs-DWI).^[9-11] Previous studies have reported that rfDWI results in better IQ than full-field DWI or ss-DWI dose in patients with cervical and rectal cancer.^[12,13] Previous studies have reported that the feature reproducibility of rs-DWI was better than that of ss-DWI in patients with cervical cancer, especially in terms of texture features.^[14] Several previous studies reported that the ADC value has good differential diagnostic performance for prostate lesions.^[15,16] Deforche et al., reported that rs-DWI sequence better discriminates patients with clinically significant prostate cancer (csPCa) than ss-DWI sequence (area under the curve [AUC] Readout-segmented multi-shot echo-planar imaging (rsEPI) = 0.84, AUC Single-shot echo-planar echo-planar imaging (ssEPI) = 0.68, P < 0.05) with a cut-off value of 1,232 μm²/s associated with a sensitivity–specificity of 97–63%.^[17] However, the scanning protocol that can provide the optimal IQ for PIRADS v2.1 assessment and optimal ADC values for diagnosing prostate lesions has not been reported. The purpose of this study was to compare the IQ of prostate images obtained with ss-DWI, rf-DWI, and rs-DWI to optimize the prostate DWI protocol and improve the accuracy of imaging diagnosis.

MATERIAL AND METHODS

Patients

This study was approved by the local research ethics committee (Ethics Committee of Fuyang Hospital of Anhui Medical University, Fuyang, China. Code: KY202443), and informed consent was obtained from all patients. We analyzed the images of 96 patients who underwent prostate multiparametric magnetic resonance imaging (mpMRI) scans according to the PI-RADS v2.1 scanning protocols as well as rf-DWI and rs-DWI sequences (from April 2023 to March 2024). Twenty-four patients were excluded due to incomplete three types of DWI sequences, prior radiotherapy before magnetic resonance imaging (MRI) scanning, significant artifacts resulting from inadequate bowel preparation, or artifacts caused by post-operative pelvic metallic implants. Finally, 72 patients were included in the study. The flowchart for patient screening is shown in Figure 1. Among these patients, 42 patients underwent biopsy through the rectum under ultrasound guidance by experienced urologists using the same method as described in previous studies.^[18]

Close

Figure 1:

Flowchart for patient screening.

Export to PPT

MRI technique and image analysis

All mpMRI images were collected by a 3.0 T MRI scanner (Veda, Siemens Healthineers) with a pelvic surface coil and activated spine coils (4–16 channels and 8–12 channels, respectively).^[3] The DWI scanning sequences included ssDWI, rf-DWI, and rs-DWI, and the scanning protocol is provided in Table 1.

Table 1: Protocol sequence parameters for multiparametric MRI of the prostate.

Scanning sequence	T2-Weighted Imaging	DWI			DCE
		ss-EPI DWI	rf- DWI	rs-DWI
FOV sequence parameters (mm)	220×100	200×100	200×00	200×000	260×60.5
Matrix size	320×80	100×100	100×100	100×100	70×64
Slice thickness (mm)	3.0	3.0	3.0	3.0	3.0
TR/TE (ms)	3410.0/101.0	4100.0/76.0	5300.0/58.0	5730.0/53.0	3.3/1.32
Flip angle	160	100	110	180	10
Voxel (mm3)	0.3×0.3×3.0	1.0×1.0×3.0	1.0×1.0×3.0	1.0×1.0×3.0	1.0×1.0×3.0
Receiver bandwidth (Hz/voxel)	200	1724	2004	963	480
Acquisition time (min)	1:27	2:43	4:53	9:29	2:42
Number of signals averaged	3	3	6	3	1
b-value (s/mm2)	-	50, 800, 1200, 1500	50, 800, 1200, 1500	50, 800, 1200, 1500	-

DWI: Diffusion-weighted imaging, ss-EPI DWI: Single-shot echo-planar DWI, rf-DWI: Reduced field-of-view DWI, rs-DWI: Readout-segmentation multi-shot DWI, MRI: Magnetic resonance imaging, DCE: Dynamic contrast-enhanced, TR: Repetition time, TE: Echo time, FOV: Field of view

The qualitative evaluation and quantitative measurement of prostate images were performed by two urogenital radiologists from different hospitals (L.X. and A.L.Q., both with 11 years of experience in interpreting mpMRI scans of the prostate) in a double-blinded manner, and they were unaware of the patient’s clinical data and pathological results. In the qualitative evaluation for IQ, a Likert scale, where 1 is poor, i.e., severe distortion or artifacts; 2 is below average, i.e., moderate distortion or artifacts; 3 is average; 4 is above average, i.e., slight distortion or artifacts; and 5 is excellent, i.e., no distortion or artifacts, was used to assess the clarity (especially the edge clarity of prostate nodules), distortion (caused by magnetic field inhomogeneities), and artifacts (mainly composed of magnetization artifacts) of prostate ss-DWI, rf-DWI, and rsDWI images, and the corresponding ADC images generated automatically from them (b values were 1,500 s/mm² for all).^[19,20] Two urogenital radiologists trained in PI-RADS v2.1 (L.X. and A.L.Q. with 11 years of experience in interpreting mpMRI scans of the prostate) estimated the MRI-based imaging scores of prostate nodules from 46 patients separately and were blinded to the clinical and pathological data while knowing that all specimens had been obtained from biopsy following the PI-RADS v2.1 evaluation protocol.^[2]

Quantitative parameters were measured on axial images. Using T2-weighted images (T2WI) as a reference [Figure 2], due to the restricted field of view (FOV) in the rf-DWI, the SNR and contrast-to-noise ratio (CNR) in this study were calculated using the obturator muscle as the reference tissue. Signal intensities of the obturator muscle and prostate tissue were measured on identical DWI slices using regions of interest (ROIs) with identical areas. The signal intensity (SI_muscle) and standard deviation (SD_muscle) of the obturator muscle and the signal intensity (SI_prostate) and standard deviation (SD_prostate) of the prostate region were measured on ss-DWI, rf-DWI, and rs-DWI images (b = 50, 1,500 mm²/s), using the slice with the largest area of the obturator muscle. The SNR and CNR were calculated using the following equations: $SNR=S{I}_{prostate}/S{D}_{muscles},CNR=\left(S{I}_{prostate}-S{I}_{muscle}\right)/\sqrt{\left(S{D}_{muscle}{}^2+S{D}_{prostate}{}^2\right)}$. The anterior-posterior (AP) diameters and right-left (RL) diameters of the prostate were measured on T2WI images and images obtained with the three DWI sequences. For the quantitative analysis of the ADC value, ROIs were delineated on the ADC maps within discernible areas of csPCa lesions. The ROIs were placed as close as possible to the edge of the lesion without touching it.

Close

Figure 2:

A 57-year-old man’s prostate axial T2-weighted imaging image. A hypointense nodule (white arrowhead) with clear boundaries is visible in the left transition zone.

Export to PPT

RESULTS

Qualitative image analysis

The DWI images of all patients had good clarity, and the IQ met the diagnostic requirements. The consistency of the qualitative scores of the two radiologists was average or good (Kappa = 0.729, P < 0.001). There were significant differences in the qualitative scores for clarity, geometric distortion, and artifacts among the three types of DWI images and the corresponding ADC images (P < 0.05) [Table 2]. In general, among the three types of DWI sequences, rf-DWI images had the best clarity, minimal distortion, and fewest artifacts on both DWI and ADC images. ss-DWI exhibited the worst clarity and the most obvious geometric distortion on both the DWI and ADC images. Among the DWI images [Figure 3a-c], the clarity and distortion of the rf-DWI [Figure 3a] and rs-DWI [Figure 3b] were better than those of the ss-DWI [Figure 3c], and the rf-DWI images had better clarity than the rs-DWI did; there were fewer artifacts on the rf-DWI images than on the rs-DWI images (P < 0.05 for all). Among the ADC images [Figure 4a-c], the clarity of the rf-ADC [Figure 4a] and the rs-ADC [Fiure 4b] images was better than that of the ss-ADC images [Figure 4c]; the distortion of the rf-ADC and the rs-ADC images was less than that of the ss-ADC images; there were fewer artifacts on the rf-ADC images than on the rs-ADC images; however, the artifacts on the ss-ADC were fewer than that on the rs-ADC (P < 0.05 for all). A total of 156 nodules were detected in 42 patients (42 with csPCa and 114 without benign prostatic hyperplasia [BPH]) in ss-DWI, rf-DWI, and rs-DWI. Among them, 90 nodules with BPH and 26 nodules with csPCa located in the transition zone (TZ) and 24 nodules with BPH and 16 nodules with csPCa located in the peripheral zone (PZ). On the three DWI sequences, the PI-RADS v2.1 scores of nodules with csPCa were higher than those with BPH (ssDWI: 3.54 ± 1.14 for csPCa and 2.01 ± 0.89 for BPH in the TZ, 3.44 ± 1.26 for csPCa and 2.00 ± 0.93 for BPH in the PZ; rf-DWI: 3.62 ± 1.10 for csPCa and 1.90 ± 0.84 for BPH in the TZ, 3.56 ± 1.21 for csPCa and 1.83 ± 0.76 for BPH in the PZ; rs-DWI: 3.58 ± 1.10 for csPCa and 2.00 ± 0.94 for BPH in the TZ, 3.50 ± 1.21 for csPCa and 1.92 ± 0.83 for BPH in the PZ, all P < 0.001) [Table 3]. The area under the receiver operating characteristic curve (AUC) of ss-DWI, rf-DWI, and rs-DWI was 0.837, 0.868, and 0.844 in the TZ and 0.810, 0.871, 0.841 in the PZ, respectively [Figure 5 a,b]. There were statistically significant differences in the AUC between ss-DWI and rfDWI as well as between rs-DWI and rf-DWI in the TZ and between ss-DWI and rf-DWI in the PZ (P < 0.05).

Table 2: Comparison of qualitative assessment of three different diffusion weighted and apparent diffusion coefficient image (n=72).

Evaluating indicator	Clarity	Distortion	Artifacts
ss-EPI DWI	3.47±0.75	3.94±0.99	4.47±0.82
rf-DWI	4.44±0.69	4.57±0.62	4.68±0.62
rs-DWI	3.99±0.88	4.50±0.71	4.28±0.79
F-value	18.118	8.653	5.025
P-value	<0.05abc	<0.05ab	<0.05c
ss-ADC	3.82±0.74	4.21±0.65	4.57±0.78
rf-ADC	4.74±0.56	4.79±0.56	4.76±0.49
rs-ADC	4.36±0.79	4.75±0.52	4.49±0.67
F-value	18.566	17.473	3.290
P-value	<0.05abc	<0.05ab	<0.05

ss-EPI: Single-shot echo-planar DWI/ADC, rf: Reduced field-of-view DWI/ADC, rs: Readout-segmentation multi-shot DWI/ADC were compared (b=50 and 1,500 s/mm²). Comparisons: ^a(ss-EPI vs. rf), ^b(ss-EPI vs. rs), ^c(rf vs. rs) for both DWI and ADC metrics. All intergroup differences were statistically significant (P<0.05). Data are expressed as mean ± SD (x̄± s) immediately following P<0.05.

Table 3: Comparison of apparent diffusion coefficient value and PI-RADS score of three DWI sequences.

Scanning sequence	ADC (D10−3 mm2/s)		PI-RADS
	TZ	PZ	TZ	PZ
ss-EPI DWI
BPH (nPZ=24) (nTZ=90)	0.998±0.191	1.016±0.194	2.01±0.89	2.00±0.93
PCa (nPZ=16) (nTZ=26)	0.741±0.203	0.750±0.241	3.54±1.14	3.44±1.26
P-value	<0.001	<0.001	<0.001	<0.001
F-value	0.233	1.943	3.778	3.758
Sensitivity (%)	80.77	68.75	80.77	75.0
Specificity (%)	74.44	75.00	75.56	75.0
AUC	0.813	0.802	0.837a	0.810a
rf-DWI
BPH (nPZ=24) (nTZ=90)	1.039±0.178	1.070±0.204	1.90±0.84	1.83±0.76
PCa (nPZ=16) (nTZ=26)	0.783±0.194	0.779±0.249	3.62±1.10	3.56±1.21
P-value	<0.001	<0.001	<0.001	<0.001
F-value	0.931	2.508	3.198	5.742
Sensitivity (%)	84.62	68.75	84.62	81.25
Specificity (%)	76.67	79.17	83.33	87.5
AUC	0.826	0.865	0.868	0.871
rs-DWI
BPH (nPZ=24) (nTZ=90)	1.092±0.172	1.113±0.219	2.00±0.94	1.92±0.83
PCa (npz=16) (nTZ=26)	0.843±0.205	0.832±0.260	3.58±1.10	3.50±1.21
P-value	<0.001	0.001	<0.001	<0.001
F-value	3.107	2.170	1.769	4.287
Sensitivity (%)	80.77	75.0	84.62	75.00
Specificity (%)	82.22	79.2	76.67	79.17
AUC	0.824	0.828	0.844a	0.841

Data are expressed as mean ± SD (x̄± s) immediately following ^aP<0.05 when comparing to the AUC of rf-DWI. BPH: Benign prostatic hyperplasia, PCa: Prostate cancer, AUC: Area under the curve, ss-EPI DWI: Single-shot echo-planar DWI, rf-DWI: Reduced field-of-view DWI, rs-DWI: Readout-segmentation multi-shot DWI, DWI: Diffusion-weighted imaging, PZ: Peripheral zone, TZ: Transition zone, PI-RADS: Prostate Imaging Reporting and Data System, ADC: Apparent diffusion coefficient, DWI: Diffusion-weighted imaging.

Close

Figure 3:

A 57-year-old man’s prostate DWI images obtained with (a) single-shot echo-planar DWI (ss-DWI), (b) readout-segmented multi-shot DWI (rs-DWI), and (c) reduced field-of-view DWI (rfDWI) (b values were 1500s/mm² for all). Among the three types of DWI, the clarity of the prostate nodule capsule (or lesion clarity indicated by white arrowheads) is the worst on the ss-DWI images and best on rf-DWI images. Magnetic susceptibility artifacts from rectal bowel gas at the junction of the rectum and prostate (white arrow), as well as on the right wall of the rectum (yellow arrow), were smallest on the rf-DWI images, and the largest on the ss-DWI images. The rs-DWI image appeared blurry, especially the right rectal wall (yellow arrow). The rf-DWI image has the least distortion among the three types of DWI images.

Export to PPT

Close

Figure 4:

A 57-year-old man’s prostate ADC images automatically generated from (a) single-shot echo-planar DWI (ss-DWI), (b) readout-segmented multi-shot DWI (rs-DWI), and (c) reduced field-of-view DWI (rf-DWI) (b values were 1500s/mm² for all). The bilateral transition zone hypointense nodule is clearest on the rf-ADC image, and the ss-ADC image has the worst image clarity (white arrowheads). Moreover, the ss-ADC image at the junction of the rectum and prostate shows the largest artifact, which is not obvious in the other two DWI images (yellow arrow). Compared with T2WI images, ss-DWI images exhibit the most significant deformation.

Export to PPT

Close

Figure 5:

Receiver operating characteristic (ROC) curves of diagnostic performance for clinically significant prostate cancer (csPCa). (a and b) Present ROC curves of apparent diffusion coefficient values derived from single-shot echo-planar diffusion-weighted imaging (ss-DWI), readout-segmented multi-shot diffusion-weighted imaging (rs-DWI), and reduced field-of-view diffusion-weighted imaging (rf-DWI) sequences for diagnosing clinically significant prostate cancer (csPCa) in the transition zone (TZ) and peripheral zone (PZ), respectively. (c and d) Display ROC curves of Prostate Imaging Reporting and Data System Version 2.1 scores based on ss-DWI, rs-DWI, and rf-DWI images for csPCa diagnosis in the TZ and PZ, respectively.

Export to PPT

DISCUSSION

High-quality DWI/ADC images play a crucial role in obtaining precise diagnoses during the evaluation of PCa through biparametric MRI or mpMRI.^[23,24] To optimize the quality of DWI images of the prostate, this study compared qualitative and quantitative indicators of images obtained with three different DWI scanning protocols, namely ss-DWI, rf-DWI, and rs-DWI. For diagnosing prostate lesions, PI-RADS v2.1 recommends a b-value of not less than 1,400 s/mm² for DWI.^[1,2] Tamada et al. reported that a very high b-value (3,000 s/mm²) did not improve the diagnostic performance for PCa. Therefore, in this study, the b-value was 1,500 s/mm² for ss-DWI, rf-DWI, and rs-DWI.^[25]

In the qualitative evaluation of images, our study provided evidence that among the three types of DWI scanning protocols, rf-DWI images had the best clarity, least image distortion, and the fewest artifacts. On the corresponding automatically generated ADC images, the rf-ADC images also presented the best clarity, minimal distortion, and fewest artifacts. In this study, rf-DWI was conducted through a technique called “2D spatially selective radiofrequency (RF) excitation”, in which only a small specific ROI is excited, producing images with higher resolution and reduced geometric distortion and susceptibility artifacts.^[26] Compared with ss-DWI images, rf-DWI images had better clarity (4.44 ± 0.69 vs. 3.47 ± 0.75, P < 0.001) and less distortion (4.57 ± 0.62 vs. 3.94 ± 0.99, P < 0.001). Similar results were also observed in the ADC images (clarity, 4.74 ± 0.56 vs. 3.82 ± 0.74; distortion, 4.79 ± 0.56 vs. 4.21 ± 0.65, P < 0.001). This result was consistent with those of other recently published studies. Lawrence et al. reported that rf-DWI images had less geometric distortion than ss-DWI images (3.2 ± 0.2 mm vs. 3.4 ± 0.3 mm, respectively; P = 0.033).^[27] Moreover, our research suggested that rf-DWI demonstrated superior clarity, less geometric distortion, and fewer artifacts compared to rs-DWI. Klingebiel et al. assessed the DWI IQ on a 5-point scale, and the results showed that the IQ of rf-DWI images was significantly superior to that of rs-DWI images. This result is consistent with the findings of our study.^[28] Inoue et al. compared the IQ of rectal cancer images between rf-DWI and rs-DWI with a full-size FOV and reported that rf-DWI had better IQ, higher rectal lesion conspicuity, and fewer artifacts than rs-DWI images did.^[29] The reasons for this result may be that the larger number of excitation times for rf-DWI (6 times) than for rs-DWI (3 times) and a longer echo time (TE) (58 ms for rf-DWI vs. 53 ms for rs-DWI) improved the SNR of rf-DWI images.^[30]

Klingebiel et al. reported that the subjective imaging parameters of rs-DWI were more superior to those of ss-DWI images but at the expense of a longer acquisition time (7:33 min).^[28] In addition, Liney et al. reported that, compared with ss-DWI, rs-DWI exhibited better performance in minimizing geometric artifacts.^[30] The reason for this difference was that a single-shot echo-planar echo planar imaging (ss-EPI) sequence fills the k-space in a single shot, whereas a readout-segmented multi-shot EPI (rs-EPI) sequence partitions the k-space into non-overlapping segments in the readout direction. This partitioning reduces the echo spacing and the echo train length, thus minimizing the susceptibility-induced distortions.^[31] Our study also revealed that clarity and geometric distortion of rs-DWI images were superior to those of ss-DWI images but at the cost of a longer acquisition time (9:29 min), which may limit its application in clinical practice. The superior IQ of DWI imaging is essential for ensuring accurate semi-quantitative PI-RADS scoring of prostatic nodules. In this study, the differential performance on rf-DWI imaging was the optimum among three DWI sequences (AUC: 0.868 in the TZ and 0.871 in the PZ), the following was rs-DWI sequence (AUC: 0.844 in the TZ and 0.841 in the PZ). The reason for this was due to the fact that the best IQ of rf-DWI images contributes to the precise semi-quantitative assessment of PI-RADS v2.1 for prostate nodules.

The quantitative evaluation results indicated that the SNR of rf-DWI was the lowest among the three types of DWI images. The reason for the decreased SNR in the rf-DWI images is complex. The increased spatial resolution and long acquisition time of rf-DWI compared to those of ss-DWI may contribute to the reduced SNR; however, the potential reasons may be the reduction in the FOV or different b-factor fitting compared to those of rs-DWI.^[13,32] The CNR of rf-DWI was the greatest among the three DWI images because the reduced FOV allows for higher-resolution imaging of the targeted area and fewer k-space lines, which leads to shorter echo train lengths and a reduction in susceptibility artifacts.^[33] The higher CNR of rs-DWI images compared to that of ssDWI images may be due to the greater number of excitations or higher scanning resolution, which increase the underlying T2 weighting and might therefore increase the difference in the tissue signal.^[29] The RL diameter of the prostate on the ss-DWI images was smaller than those on the T2WI images (P < 0.05). The discrepancy in RL diameters between ss-DWI and T2WI was attributed to the lower spatial resolution, greater geometric distortion, and increased artifacts in the DWI images.

Our study demonstrated that ADC values from ss-DWI, rf-DWI, and rs-DWI sequences all demonstrated significant diagnostic performance for csPCa (all P <0.001, except for P = 0.001 in the PZ of the rs-DWI sequence). Notably, the rf-DWI sequence yielded the highest AUC values in both the TZ (0.826) and PZ (0.865) of the prostate, the following sequence was rs-DWI (TZ:0.824, PZ:0.828). Although the three DWI sequences showed good diagnostic efficacy, statistical analysis confirmed no meaningful differences in their performance for prostate cancer. Deforche et al. demonstrated that rs-DWI ADC values had significantly higher diagnostic utility for prostate cancer (AUCrsEPI = 0.84, AUCssEPI = 0.68, P < 0.05) than ss-DWI (AUC 0.85 vs. 0.72), with an optimal cutoff at 1,232 × 10^-6 mm²/s, despite no statistically significant difference in ADC values (P > 0.05).^[17] Tavakoli et al. demonstrated that quantitative ADC measurements significantly increased upgrade category for ADC PI-RADS 3 (P = 0.04) and ADC PI-RADS 4 (P < 0.001) lesions when applying a cutoff of ≤0.90 × 10^-3 mm²/s.^[34] They emphasized the clinical utility of integrating quantitative ADC analysis into the decision-making process for PI-RADS 3 lesions. However, the authors did not specify DWI parameters or discuss protocol optimality in the literature. Our study demonstrates that the rf-DWI sequence with the highest diagnostic efficacy for prostate cancer exhibits an ADC threshold of 0.958 × 10⁻³ mm²/s in the TZ and 0.769 × 10⁻³ mm²/s in the PZ at a b-value of 1,500 s/mm², which may serve as clinical reference for deciding whether a prostate biopsy was necessary. Nevertheless, the study had a small sample size of pathologically confirmed prostatic nodules, but validation was needed with larger, multi-center data.

This study has several limitations. First, this study exclusively investigated a single b-value of 1,500 s/mm² in the DWI sequence. DWI images or ADC values obtained at other b-values may exhibit distinct characteristics. Second, the lack of a unified MRI scanning protocol (including TR/TE and excitation frequency) may have an impact on determining the optimal DWI scanning protocol. Second, this study exclusively utilized biopsy-derived pathological findings for prostate nodules, which inherently carry the risk of prostate cancer underdiagnosis due to known false-negative rates of needle biopsies. In addition, the limited sample size of pathologically confirmed prostatic lesions introduces statistical uncertainty that may compromise the reliability of quantitative analytical outcomes. Third, this investigation was conducted as a single-center study utilizing MRI scanners from a single vendor. The generalizability of these findings necessitates further validation through multi-institutional trials incorporating MRI systems from multiple manufacturers to assess cross-platform reproducibility.

CONCLUSION

In summary, compared to ss-DWI and rs-DWI sequences, the rf-DWI sequence demonstrates superior IQ. PI-RADS v2.1 category assessments on rf-DWI images and the ADC value derived from corresponding ADC maps exhibited optimal diagnostic performance for csPCa. Biopsy should be considered when ADC values is below 0.958 × 10³ mm²/s in the TZ or 0.769 × 10³ mm²/s in the PZ. These findings require multicenter validation with multivendor MRI scanners and larger sample sizes to confirm clinical generalizability.