Diagnostic performance of MRI in detecting prostate cancer in patients with prostate-specific antigen levels of 4-10 ng/mL: a systematic review and meta-analysis.

OBJECTIVE: To investigate the diagnostic performance of MRI in detecting clinically significant prostate cancer (csPCa) and prostate cancer (PCa) in patients with prostate-specific antigen (PSA) levels of 4-10 ng/mL. METHODS: A computerized search of PubMed, Embase, Cochrane Library, Medline, and Web of Science was conducted from inception until October 31, 2023. We included articles on the use of MRI to detect csPCa or PCa at 4-10 ng/mL PSA. The primary and secondary outcomes were MRI performance in csPCa and PCa detection, respectively; the estimates of sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were pooled in a bivariate random-effects model. RESULTS: Among the 19 studies (3879 patients), there were 10 (2205 patients) and 13 studies (2965 patients) that reported MRI for detecting csPCa or PCa, respectively. The pooled sensitivity and specificity for csPCa detection were 0.84 (95% confidence interval [CI], 0.79-0.88) and 0.76 (95%CI, 0.65-0.84), respectively, for PCa detection were 0.82 (95%CI, 0.75-0.87) and 0.74 (95%CI, 0.65-0.82), respectively. The pooled NPV for csPCa detection was 0.91 (0.87-0.93). Biparametric magnetic resonance imaging also showed a significantly higher sensitivity and specificity relative to multiparametric magnetic resonance imaging (both p < 0.01). CONCLUSION: Prostate MRI enables the detection of csPCa and PCa with satisfactory performance in the PSA gray zone. The excellent NPV for csPCa detection indicates the possibility of biopsy decision-making in patients in the PSA gray zone, but substantial heterogeneity among the included studies should be taken into account. CLINICAL RELEVANCE STATEMENT: Prostate MRI can be considered a reliable and satisfactory tool for detecting csPCa and PCa in patients with PSA in the "gray zone", allowing for reducing unnecessary biopsy and optimizing the overall examination process. KEY POINTS: Prostate-specific antigen (PSA) is a common screening tool for prostate cancer but risks overdiagnosis. MRI demonstrated excellent negative predictive value for prostate cancer in the PSA gray zone. MRI can influence decision-making for these patients, and biparametric MRI should be further evaluated.

Exploring a simplified way to diagnose pelvic lipomatosis: prediction of pelvic fat volume using a single cross-sectional image.

Background: Pelvic lipomatosis (PL) is a rare disease characterized by the overgrowth of pelvic adipose tissue (AT). We investigated the relationships between areas of subcutaneous adipose tissue (SAT) and visceral adipose tissue (VAT) and pelvic fat volume (PFV), and analyzed the feasibility of diagnosing PL from a single cross-sectional image. Methods: The study included 50 patients and 50 controls. We used nnU-Net to segment SAT and VAT automatically. L3 vertebra was set as the zero point (L0), and a total of 201 slices were obtained with a 1 mm interval (L-100 - L+100). We selected 5 pelvic slices, including slices of the anterior superior margin of the S1-S4 vertebrae and the slice above the bilateral femoral head (FH). SAT areas, VAT areas, and PFVs were calculated by computational software. Areas and volumes of 2 groups were compared by t-test or rank-sum test. The correlations among areas and PFV were calculated. Logistic regression models were developed to identify the best slice for predicting PL. Receiver operating characteristic (ROC) curves were performed, and the area under the curve (AUC) and thresholds [with sensitivity (SEN) and specificity (SPE)] were calculated. Results: VAT areas of L-94 - L-100, L+79 - L+100, S1-S4, and FH indicated statistical differences between patients and controls (P<0.05). The linear regression model with VAT area as the independent variable was established to estimate PFV (FH level: r=0.745, P<0.001, R2=0.555). Among the univariate logistic regression models, VAT area at FH as the independent variable had the highest performance in predicting PL (AUC: 0.893, SEN: 74%, SPE: 94%), followed by S4 level (AUC: 0.800, SEN: 88%, SPE: 66%). The overall accuracy of the logistic regression model including VAT areas at S4 and FH in predicting PL was 88% (AUC: 0.927, SEN: 90%, SPE: 88%). Conclusions: VAT areas at the level of FH can help estimate the value of PFV. VAT areas of S4 and FH provide greater power than a single image for the diagnosis of PL.

Development of a multimodal kidney age prediction based on automatic segmentation CT image in patients with normal renal function.

Background: For decades, description of renal function has been of interest to clinicians and researchers. Serum creatinine (Scr) and estimated glomerular filtration rate (eGFR) are familiar but also limited in many circumstances. Meanwhile, the physiological volumes of the kidney cortex and medulla are presumed to change with age and have been proven to change with decreasing kidney function. Methods: We recruited 182 patients with normal Scr levels between October 2021 and February 2022 in Peking Union Medical College Hospital (PUMCH) with demographic and clinical data. A 3D U-Net architecture is used for both cortex and medullary separation, and volume calculation. In addition, we included patients with the same inclusion criteria but with diabetes (PUMCH-DM test set) and diabetic nephropathy (PUMCH-DN test set) for internal comparison to verify the possible clinical value of "kidney age" (K-AGE). Results: The PUMCH training set included 146 participants with a mean age of 47.5 ± 7.4 years and mean Scr 63.5 ± 12.3 µmol/L. The PUMCH test set included 36 participants with a mean age of 47.1 ± 7.9 years and mean Scr 66.9 ± 13.0 µmol/L. The multimodal method predicted K-AGE approximately close to the patient's actual physiological age, with 92% prediction within the 95% confidential interval. The mean absolute error increases with disease progression (PUMCH 5.00, PUMCH-DM 6.99, PUMCH-DN 9.32). Conclusion: We established a machine learning model for predicting the K-AGE, which offered the possibility of evaluating the whole kidney health in normal kidney aging and in disease conditions.

Effect of preoperative PI-RADS assessment on pathological outcomes in patients who underwent radical prostatectomy.

OBJECTIVE: To assess the effect of preoperative MRI with standardized Prostate Imaging-Reporting and Data System (PI-RADS) assessment on pathological outcomes in prostate cancer (PCa) patients who underwent radical prostatectomy (RP). PATIENTS AND METHODS: This retrospective cohort study included patients who had undergone prostate MRI and subsequent RP for PCa between January 2017 and December 2022. The patients were divided into the PI-RADS group and the non-PI-RADS group according to evaluation scheme of presurgery MRI. The preoperative characteristics and postoperative outcomes were retrieved and analyzed. The pathological outcomes included pathological T stage (pT2 vs. pT3-4) and positive surgical margins (PSMs). Patients were further stratified according to statistically significant preoperative variables to assess the difference in pathological outcomes. A propensity score matching based on the above preoperative characteristics was additionally performed. RESULTS: A total of 380 patients were included in this study, with 201 patients in the PI-RADS group and 179 in the non-PI-RADS group. The two groups had similar preoperative characteristics, except for clinical T stage (cT). As for pathological outcomes, the PI-RADS group showed a significantly lower percentage of pT3-4 (21.4% vs. 48.0%, p < 0.001), a lower percentage of PSMs (31.3% vs. 40.9%, p = 0.055), and a higher concordance between the cT and pT (79.1% vs. 64.8%, p = 0.003). The PI-RADS group also showed a lower proportion of pT3-4 (p < 0.001) in the cT1-2 subgroup and the cohort after propensity score matching. The PSM rate of cT3 patients was reduced by 39.2% in the PI-RADS group but without statistical significance (p = 0.089). CONCLUSIONS: Preoperative MRI with standardized PI-RADS assessment could benefit the decision-making of patients by reducing the rate of pathologically confirmed non-organ-confined PCa after RP and slightly reducing the PSM rate compared with non-PI-RADS assessment.

An MRI-based grading system for preoperative risk estimation of positive surgical margin after radical prostatectomy.

OBJECTIVE: To construct a simplified grading system based on MRI features to predict positive surgical margin (PSM) after radical prostatectomy (RP). METHODS: Patients who had undergone prostate MRI followed by RP between January 2017 and January 2021 were retrospectively enrolled as the derivation group, and those between February 2021 and November 2022 were enrolled as the validation group. One radiologist evaluated tumor-related MRI features, including the capsule contact length (CCL) of lesions, frank extraprostatic extension (EPE), apex abutting, etc. Binary logistic regression and decision tree analysis were used to select risk features for PSM. The area under the curve (AUC), sensitivity, and specificity of different systems were calculated. The interreader agreement of the scoring systems was evaluated using the kappa statistic. RESULTS: There were 29.8% (42/141) and 36.4% (32/88) of patients who had PSM in the derivation and validation cohorts, respectively. The first grading system was proposed (mrPSM1) using two imaging features, namely, CCL ≥ 20 mm and apex abutting, and then updated by adding frank EPE (mrPSM2). In the derivation group, the AUC was 0.705 for mrPSM1 and 0.713 for mrPSM2. In the validation group, our grading systems showed comparable AUC with Park et al.'s model (0.672-0.686 vs. 0.646, p > 0.05) and significantly higher specificity (0.732-0.750 vs. 0.411, p < 0.001). The kappa value was 0.764 for mrPSM1 and 0.776 for mrPSM2. Decision curve analysis showed a higher net benefit for mrPSM2. CONCLUSION: The proposed grading systems based on MRI could benefit the risk stratification of PSM and are easily interpretable. CRITICAL RELEVANCE STATEMENT: The proposed mrPSM grading systems for preoperative prediction of surgical margin status after radical prostatectomy are simplified compared to a previous model and show high specificity for identifying the risk of positive surgical margin, which might benefit the management of prostate cancer. KEY POINTS: • CCL ≥ 20 mm, apex abutting, and EPE were important MRI features for PSM. • Our proposed MRI-based grading systems showed the possibility to predict PSM with high specificity. • The MRI-based grading systems might facilitate a structured risk evaluation of PSM.

The feasibility of contrast-enhanced CT to identify the adhesive renal venous tumor thrombus of renal cell carcinoma.

OBJECTIVE: To identify adhesive renal venous tumor thrombus (RVTT) of renal cell carcinoma (RCC) by contrast-enhancement CT (CECT). MATERIALS AND METHODS: Our retrospective study included 53 patients who underwent preoperative CECT and pathologically confirmed RCC combined with RVTT. They were divided into two groups based on the intra-operative findings of RVTT adhesion to the venous wall, with 26 cases in the adhesive RVTT group (ARVTT) and 27 cases in the non-adhesive group (NRVTT). The location, maximum diameter (MD) and CT values of tumors, the maximum length (ML) and width (MW) of RVTT, and length of inferior vena cava tumor thrombus were compared between the two groups. The presence of renal venous wall involvement, renal venous wall inflammation, and enlarged retroperitoneal lymph node was compared between the two groups. A receiver operating characteristic curve was used to analyze the diagnostic performance. RESULTS: The MD of RCC and the ML and MW of the RVTT were all larger in the ARVTT group than in the NRVTT group (p = 0.042, p < 0.001, and p = 0.002). The proportion of renal vein wall involvement and renal vein wall inflammation were higher in the ARVTT group than in NRVTT groups (both p < 0.001). The multivariable model including ML and vascular wall inflammation to predict ARVTT could achieve the best diagnostic performance with the area under the curve, sensitivity, specificity, and accuracy of 0.91, 88.5%, 96.3%, and 92.5%, respectively. CONCLUSION: The multivariable model acquired by CECT images could be used to predict RVTT adhesion. CLINICAL RELEVANCE STATEMENT: For RCC patients with tumor thrombus, contrast-enhanced CT could noninvasively predict the adhesion of tumor thrombus, thus predicting the difficulty of surgery and contributing to the selection of an appropriate treatment plan. KEY POINTS: • The length and width of the tumor thrombus could be used to predict its adhesion to the vessel wall. • Adhesion of the tumor thrombus can be reflected by inflammation of the renal vein wall. • The multivariable model from CECT can well predict whether the tumor thrombus adhered to the vein wall.

Development and acceptability validation of a deep learning-based tool for whole-prostate segmentation on multiparametric MRI: a multicenter study.

Background: Accurate whole prostate segmentation on magnetic resonance imaging (MRI) is important in the management of prostatic diseases. In this multicenter study, we aimed to develop and evaluate a clinically applicable deep learning-based tool for automatic whole prostate segmentation on T2-weighted imaging (T2WI) and diffusion-weighted imaging (DWI). Methods: In this retrospective study, 3-dimensional (3D) U-Net-based models in the segmentation tool were trained with 223 patients who underwent prostate MRI and subsequent biopsy from 1 hospital and validated in 1 internal testing cohort (n=95) and 3 external testing cohorts: PROSTATEx Challenge for T2WI and DWI (n=141), Tongji Hospital (n=30), and Beijing Hospital for T2WI (n=29). Patients from the latter 2 centers were diagnosed with advanced prostate cancer. The DWI model was further fine-tuned to compensate for the scanner variety in external testing. A quantitative evaluation, including Dice similarity coefficients (DSCs), 95% Hausdorff distance (95HD), and average boundary distance (ABD), and a qualitative analysis were used to evaluate the clinical usefulness. Results: The segmentation tool showed good performance in the testing cohorts on T2WI (DSC: 0.922 for internal testing and 0.897-0.947 for external testing) and DWI (DSC: 0.914 for internal testing and 0.815 for external testing with fine-tuning). The fine-tuning process significantly improved the DWI model's performance in the external testing dataset (DSC: 0.275 vs. 0.815; P<0.01). Across all testing cohorts, the 95HD was <8 mm, and the ABD was <3 mm. The DSCs in the prostate midgland (T2WI: 0.949-0.976; DWI: 0.843-0.942) were significantly higher than those in the apex (T2WI: 0.833-0.926; DWI: 0.755-0.821) and base (T2WI: 0.851-0.922; DWI: 0.810-0.929) (all P values <0.01). The qualitative analysis showed that 98.6% of T2WI and 72.3% of DWI autosegmentation results in the external testing cohort were clinically acceptable. Conclusions: The 3D U-Net-based segmentation tool can automatically segment the prostate on T2WI with good and robust performance, especially in the prostate midgland. Segmentation on DWI was feasible, but fine-tuning might be needed for different scanners.

Development and clinical utility analysis of a prostate zonal segmentation model on T2-weighted imaging: a multicenter study.

OBJECTIVES: To automatically segment prostate central gland (CG) and peripheral zone (PZ) on T2-weighted imaging using deep learning and assess the model's clinical utility by comparing it with a radiologist annotation and analyzing relevant influencing factors, especially the prostate zonal volume. METHODS: A 3D U-Net-based model was trained with 223 patients from one institution and tested using one internal testing group (n = 93) and two external testing datasets, including one public dataset (ETDpub, n = 141) and one private dataset from two centers (ETDpri, n = 59). The Dice similarity coefficients (DSCs), 95th Hausdorff distance (95HD), and average boundary distance (ABD) were calculated to evaluate the model's performance and further compared with a junior radiologist's performance in ETDpub. To investigate factors influencing the model performance, patients' clinical characteristics, prostate morphology, and image parameters in ETDpri were collected and analyzed using beta regression. RESULTS: The DSCs in the internal testing group, ETDpub, and ETDpri were 0.909, 0.889, and 0.869 for CG, and 0.844, 0.755, and 0.764 for PZ, respectively. The mean 95HD and ABD were less than 7.0 and 1.3 for both zones. The U-Net model outperformed the junior radiologist, having a higher DSC (0.769 vs. 0.706) and higher intraclass correlation coefficient for volume estimation in PZ (0.836 vs. 0.668). CG volume and Magnetic Resonance (MR) vendor were significant influencing factors for CG and PZ segmentation. CONCLUSIONS: The 3D U-Net model showed good performance for CG and PZ auto-segmentation in all the testing groups and outperformed the junior radiologist for PZ segmentation. The model performance was susceptible to prostate morphology and MR scanner parameters.

Effects of dynamic contrast enhancement on transition zone prostate cancer in Prostate Imaging Reporting and Data System Version 2.1.

BACKGROUND: The aim of the study was to analyse the effects of dynamic contrast enhanced (DCE)-MRI on transitional-zone prostate cancer (tzPCa) and clinically significant transitional-zone prostate cancer (cs-tzPCa) in Prostate Imaging Reporting and Data System (PI-RADS) Version 2.1. PATIENTS AND METHODS: The diagnostic efficiencies of T2-weighted imaging (T2WI) + diffusion-weighted imaging (DWI), T2WI + dynamic contrast-enhancement (DCE), and T2WI + DWI + DCE in tzPCa and cs-tzPCa were compared using the score of ≥ 4 as the positive threshold and prostate biopsy as the reference standard. RESULTS: A total of 425 prostate cases were included in the study: 203 cases in the tzPCa group, and 146 in the cs-tzPCa group. The three sequence combinations had the similar areas under the curves in diagnosing tzPCa and cs-tzPCa (all P < 0.05). The sensitivity of T2WI + DCE and T2WI + DWI + DCE (84.7% and 85.7% for tzPCa; 88.4% and 89.7% for cs-tzPCa, respectively) in diagnosing tzPCa and cs-tzPCa was significantly greater than that of T2WI + DWI (79.3% for tzPCa; 82.9% for cs-tzPCa). The specificity of T2WI + DWI (86.5% for tzPCa; 74.9% for cs-tzPCa) were significantly greater than those of T2WI + DCE and T2WI + DWI + DCE (68.0% and 68.5% for tzPCa; 59.1% and 59.5% for cs-tzPCa, respectively) (all P > 0.05). The diagnostic efficacies of T2WI + DCE and T2WI + DWI + DCE had no significant differences (all P < 0.05). CONCLUSIONS: DCE can improve the sensitivity of diagnosis for tzPCa and cs-tzPCa, and it is useful for small PCa lesion diagnosis.

Preoperative CT features to predict risk stratification of non-muscle invasive bladder cancer.

PURPOSE: To investigate whether preoperative CT features can be used to predict risk stratification of non-muscle invasive bladder cancer (NMIBC). METHODS: The 168 patients with pathologically confirmed NMIBC who underwent preoperative CT urography were retrospectively analyzed and were divided into training (n = 117) and testing (n = 51) sets. According to the European Association of Urology Guidelines, patients were classified into low-risk (n = 50), medium-risk (n = 23), and high-risk (n = 95) groups. A random over-sample was performed to handle the offset caused by the unbalanced groups. We measured some CT features that may help stratify which for modeling were determined using an F-test-based feature selection with a tenfold cross-validation procedure, and the Gaussian Naive Bayes model was trained on the entire training set. In the testing set, the performance of the model was evaluated. RESULTS: The selected CT features were the maximum and the minimum diameter of the largest tumor, whether the largest tumor is located at the trigone, and tumor number. In the testing set, the model reached a macro- and micro- AUC of 0.783 and 0.745 with an accuracy of 0.529. As for the one-vs-rest problem, the model was most effective in identifying low-risk individuals, with an AUC, accuracy, sensitivity, and specificity of 0.870, 0.647, 1.000, and 0.438, respectively; the medium-risk group reached 0.814, 0.882, 0.250, and 0.936, respectively; the identification of the high-risk group was harder, going 0.665, 0.529, 0.250, and 0.870, respectively. CONCLUSION: It is feasible to predict the risk stratification of NMIBC using preoperative CT features.

Prediction of World Health Organization /International Society of Urological Pathology (WHO/ISUP) Pathological Grading of Clear Cell Renal Cell Carcinoma by Dual-Layer Spectral CT.

RATIONALE AND OBJECTIVES: To evaluate whether the dual-layer spectral computed tomography urography (DL-CTU) images could predict WHO/ISUP pathological grading of clear cell renal cell carcinoma (ccRCC). MATERIALS AND METHODS: We retrospectively included patients (n = 50) with pathologically confirmed ccRCC who underwent preoperative DL-CTU (from October 2017 to February 2021). They were divided into low-grade (WHO/ISUP 1/2, n = 30) and high-grade groups (WHO/ISUP 3/4, n = 20). The lesion size, attenuation (HU), iodine concentration (IC), normalized IC(NIC), and other quantitative characteristics were compared between the two groups. HU, IC, and NIC were obtained by plotting ROI with two different methods (circular ROI in the solid component or irregular ROI along the tumor edge containing tumor necrotic components). Receiver operating characteristic curves and multivariable model were used to evaluate the ability of parameters to predict WHO/ISUP grade. RESULTS: Transverse diameter (TD) of low-grade tumors was smaller, and HU in the non-contrast phase of the second method (HU-U-2) was lower than that of high-grade tumors (34.21±15.14 mm vs. 46.50 ± 20.68 mm, 27.33 ± 6.65 HU vs. 31.36 ± 6.09 HU, p< 0.05). The NIC in the nephrographic phase by the two methods (NIC-N-1 and NIC-N-2) of low-grade was higher than that of the high-grade group (0.78± 0.19 vs.0.58 ± 0.22, 0.73 ± 0.42 vs. 0.46 ± 0.22, p< 0.05). The final multivariable model composed of TD, HU-U-2, and NIC-N-1 could predict ccRCC grade with the area under the curve, sensitivity, specificity, and accuracy of 0.852, 70%, 90%, and 82%. CONCLUSION: Quantitative indicators in DL-CTU images could help predict the WHO/ISUP grade of ccRCC.

Application of deep learning reconstruction of ultra-low-dose abdominal CT in the diagnosis of renal calculi.

BACKGROUND: Renal calculi are a common and recurrent urological disease and are usually detected by CT. In this study, we evaluated the diagnostic capability, image quality, and radiation dose of abdominal ultra-low-dose CT (ULDCT) with deep learning reconstruction (DLR) for detecting renal calculi. METHODS: Sixty patients with suspected renal calculi were prospectively enrolled. Low-dose CT (LDCT) images were reconstructed with hybrid iterative reconstruction (LD-HIR) and was regarded as the standard for stone and lesion detection. ULDCT images were reconstructed with HIR (ULD-HIR) and DLR (ULD-DLR). We then compared stone detection rate, abdominal lesion detection rate, image quality and radiation dose between LDCT and ULDCT. RESULTS: A total of 130 calculi were observed on LD-HIR images. Stone detection rates of ULD-HIR and ULD-DLR images were 93.1% (121/130) and 95.4% (124/130). A total of 129 lesions were detected on the LD-HIR images. The lesion detection rate on ULD-DLR images was 92.2%, with 10 cysts < 5 mm in diameter missed. The CT values of organs on ULD-DLR were similar to those on LD-HIR and lower than those on ULD-HIR. Signal-to-noise ratio was highest and noise lowest on ULD-DLR. The subjective image quality of ULD-DLR was similar to that of LD-HIR and better than that of ULD-HIR. The effective radiation dose of ULDCT (0.64 ± 0.17 mSv) was 77% lower than that of LDCT (2.75 ± 0.50 mSv). CONCLUSION: ULDCT combined with DLR could significantly reduce radiation dose while maintaining suitable image quality and stone detection rate in the diagnosis of renal calculi.

Comparison between biparametric and multiparametric MRI diagnosis strategy for prostate cancer in the peripheral zone using PI-RADS version 2.1.

PURPOSE: To compare and analyse the diagnostic value of PI-RADS v2.1 when used with biparametric MRI (bpMRI) versus multiparametric MRI (mpMRI), DWI versus T2WI to detect peripheral-zone prostate cancer (pzPCa) and clinically significant peripheral-zone prostate cancer (cs-pzPCa). METHODS: The diagnostic efficiencies of mpMRI and bpMRI as well as DWI and T2WI in pzPCa and cs-pzPCa were compared using a PI-RADS score of ≥ 4 as the positive threshold and prostate biopsy and radical prostatectomy as the reference standards. RESULTS: A total of 307 prostate cases were included in the study, including 142 in the non-pzPCa group, 165 in the pzPCa group, and 130 in the cs-pzPCa group. The AUCs of mpMRI and bpMRI were 0.717 and 0.733 (P = 0.317), respectively, for the diagnosis of pzPCa (sensitivities: 89.1% and 81.8%; specificities: 54.2% and 64.8%, both P < 0.001) and 0.594 and 0.602 (P = 0.756), respectively, for the diagnosis of cs-pzPCa (sensitivities: 93.1% and 86.2%, P = 0.004; specificities: 25.7% and 34.3%, P = 0.250). The AUCs of DWI and T2WI were 0.733 and 0.749 (P = 0.308), respectively, for the diagnosis of pzPCa (sensitivities: 81.8% and 84.2%; specificities: 64.8% and 66.2%, both P > 0.05) and 0.602 and 0.581 (P = 0.371), respectively, for the diagnosis of cs-pzPCa (sensitivities: 86.2% and 87.7%; specificities: 34.3% and 28.6%, both P > 0.05). CONCLUSION: mpMRI and bpMRI as well as DWI and T2WI using PI-RADS v2.1 exhibited similar diagnostic efficiency in pzPCa and cs-pzPCa.

Value of deep learning reconstruction at ultra-low-dose CT for evaluation of urolithiasis.

OBJECTIVES: To determine the diagnostic accuracy and image quality of ultra-low-dose computed tomography (ULDCT) with deep learning reconstruction (DLR) to evaluate patients with suspected urolithiasis, compared with ULDCT with hybrid iterative reconstruction (HIR) by using low-dose CT (LDCT) with HIR as the reference standard. METHODS: Patients with suspected urolithiasis were prospectively enrolled and underwent abdominopelvic LDCT, followed by ULDCT if any urinary stone was observed. Radiation exposure, stone characteristics, image noise, signal-to-noise ratio (SNR), and subjective image quality on a 5-point Likert scale were evaluated and compared. RESULTS: The average effective radiation dose of ULDCT was significantly lower than that of LDCT (1.28 ± 0.34 vs. 5.49 ± 1.00 mSv, p < 0.001). According to the reference standard (LDCT-HIR), 148 urinary stones were observed in 85.0% (51/60) of patients. ULDCT-DLR detected 143 stones with a rate of 96.6%, and ULDCT-HIR detected 142 stones with a rate of 95.9%. The urinary stones that were not observed with ULDCT-DLR or ULDCT-HIR were renal calculi smaller than 3 mm. There were no significant differences in the detection of clinically significant calculi (≥ 3 mm) or stone size estimation among ULDCT-DLR, ULDCT-HIR, and LDCT-HIR. The image quality of ULDCT-DLR was better than that of ULDCT-HIR and LDCT-HIR with lower image noise, higher SNR, and higher average subjective score. CONCLUSIONS: ULDCT-DLR performed comparably to LDCT-HIR in urinary stone detection and size estimation with better image quality and decreased radiation exposure. ULDCT-DLR may have potential to be considered the first-line choice to evaluate urolithiasis in practice. KEY POINTS: • Ultra-low-dose computed tomography (ULDCT) has been investigated for diagnosis of urolithiasis, but stone evaluation may be adversely impacted by compromised image quality. • This study evaluated the value of novel deep learning reconstruction (DLR) at ULDCT by comparing the stone evaluation and image quality of ULDCT-DLR to the reference standard of low-dose CT (LDCT) with hybrid iterative reconstruction (HIR). • ULDCT-DLR performed comparably to LDCT-HIR in urinary stone detection and size estimation with better image quality and reduced radiation exposure.

[Research Progress of Magnetic Resonance Imaging-based Radiomics in Prostate Cancer].

Radiomics can extract high-throughput and quantitative image features from medical images and mine the information related to the pathophysiology of tumors,which can help clinical decision-making and improve the diagnostic and predictive performance.Radiomics has been widely used in the study of prostate cancer (PCa),demonstrating application values in the diagnosis and differential diagnosis,pathology classification,invasion assessment,efficacy prediction,and prognosis analysis of PCa.Here we reviewed the recent research progress of magnetic resonance imaging-based radiomics in PCa.

CT-based radiomics to predict muscle invasion in bladder cancer.

OBJECTIVES: This study investigated the feasibility of a computed tomography (CT)-based radiomics prediction model to evaluate muscle invasive status in bladder cancer. METHODS: Patients who underwent CT urography at two medical centers from October 2014 to May 2020 and had bladder urothelial carcinoma confirmed by postoperative histopathology were retrospectively enrolled. In total, 441 cases were collected and randomized into a training cohort (n = 293), an internal testing cohort (n = 73), and an external testing cohort (n = 75). The images were first filtered, and then, 1218 features were extracted. The best features related to muscle invasiveness of bladder cancer were identified by ANOVA. A prediction model was built by using the logistic regression method. Statistical analysis was performed by plotting the receiver operating characteristic curve. Indicators of the diagnostic performance of the prediction model, including sensitivity, specificity, accuracy, and area under curve (AUC), were evaluated. RESULTS: In the training, internal testing, and external testing cohorts, the prediction model diagnosed muscle-invasive bladder cancer with AUCs of 0.885 (95% confidence interval [95% CI] 0.841-0.929), 0.820 (95% CI 0.698-0.941), and 0.784 (95% CI 0.674-0.893), respectively. In the internal testing cohort, the sensitivity, specificity, and accuracy of the model were 0.667 (95% CI 0.387-0.870), 0.845 (95% CI 0.721-0.922), and 0.782 (95% CI 0.729-0.827), respectively. In the external testing cohort, the sensitivity, specificity, and accuracy of the model were 0.742 (95% CI 0.551-0.873), 0.750 (95% CI 0.594-0.863), and 0.782 (95% CI 0.729-0.827), respectively. CONCLUSIONS: CT-based radiomics prediction model can evaluate muscle invasiveness of bladder cancer before surgery with a good diagnostic performance. KEY POINTS: • CT-based radiomics model can evaluate muscle invasive status in bladder cancer. • The radiomics model shows good diagnostic performance to differentiate muscle-invasive bladder cancer from non-muscle-invasive bladder cancer. • This preoperative CT-based prediction method might complement MR evaluation of bladder cancer and supplement biopsy.

Utilisation of virtual non-contrast images and virtual mono-energetic images acquired from dual-layer spectral CT for renal cell carcinoma: image quality and radiation dose.

BACKGROUND: Renal cell carcinoma (RCC) is the most common renal malignant tumour. We evaluated the potential value and dose reduction of virtual non-contrast (VNC) images and virtual monoenergetic images (VMIs) from dual-layer spectral CT (DL-CT) in the diagnosis of RCC. RESULTS: Sixty-two patients with pathologically confirmed RCC who underwent contrast-enhanced DL-CT were retrospectively analysed. For the comparison between true non-contrast (TNC) and VNC images of the excretory phase, the attenuation, image noise, signal-to-noise ratio (SNR) and subjective image quality of tumours and different abdominal organs and tissues were evaluated. To compare corticomedullary phase images and low keV VMIs (40 to 100 keV) from the nephrographic phase, the attenuation, image noise, SNR and subjective lesion visibility of the tumours and renal arteries were evaluated. For the tumours, significant differences were not observed in attenuation, noise or SNR between TNC and VNC images (p > 0.05). For the abdominal organs and tissues, except for fat, the difference in attenuation was 100% within 15 HU and 96.78% within 10 HU. The subjective image quality of TNC and VNC images was equivalent (p > 0.05). The attenuation of lesions in 40 keV VMIs and renal arteries in 60 keV VMIs were similar to those in the corticomedullary images (p > 0.05). The subjective lesion visibility in low keV VMIs is slightly lower than that in the corticomedullary images (p < 0.05). Using VNC and VMIs instead of TNC and corticomedullary phase images could decrease the radiation dose by 50.5%. CONCLUSION: VNC images and VMIs acquired from DL-CT can maintain good image quality and decrease the radiation dose for diagnosis of RCC.

Diffusion kurtosis imaging as an imaging biomarker for predicting prognosis in chronic kidney disease patients.

BACKGROUND: Renal fibrosis is the strongest prognostic predictor of end-stage renal disease (ESRD) in chronic kidney disease (CKD). Diffusion kurtosis imaging (DKI) is a promising method of magnetic resonance imaging successfully used to assess renal fibrosis in immunoglobulin A nephropathy. This study aimed to be the first to evaluate the long-term prognostic value of DKI in CKD patients. METHODS: Forty-two patients with CKD were prospectively enrolled, and underwent DKI on a clinical 3T MR scanner. We excluded patients with comorbidities that could affect the volume or the components of the kidney. DKI parameters, including mean Kurtosis (K), mean diffusivity and apparent diffusion coefficient (ADC) of kidney cortex were obtained by region-of-interest measurement. We followed up these patients for a median of 43 months and investigated the correlations between each DKI parameter and overall renal prognosis. RESULTS: Both K and ADC values were correlated well with the estimated glomerular filtration rate (eGFR) on recruitment and the eGFR of the last visit in follow-up (P ˂ 0.001). K and ADC values were also well associated with the eGFR slopes in CKD patients, both with the first-last time point slope (P = 0.011 and P ˂ 0.001, respectively) and with the regression slope (P = 0.010 and P ˂ 0.001, respectively). Cox proportional hazard regression indicated that lower eGFR and ADC values independently predicted eGFR loss of ˃30% and ESRD. The receiver operating characteristic analysis showed that K and ADC values were predictable for renal prognosis, and ADC displayed better capabilities for both ESRD [area under the curve (AUC) 0.936, sensitivity 92.31%, specificity 82.76%] and the composite endpoint (eGFR loss ˃30% or ESRD) (AUC 0.881, sensitivity 66.67%, specificity 96.3%). CONCLUSIONS: Renal ADC values obtained from DKI showed significant predictive value for the prognosis of CKD patients, which could be a promising noninvasive technique in follow-up.

Deep Learning on Enhanced CT Images Can Predict the Muscular Invasiveness of Bladder Cancer.

BACKGROUND: Clinical treatment decision making of bladder cancer (BCa) relies on the absence or presence of muscle invasion and tumor staging. Deep learning (DL) is a novel technique in image analysis, but its potential for evaluating the muscular invasiveness of bladder cancer remains unclear. The purpose of this study was to develop and validate a DL model based on computed tomography (CT) images for prediction of muscle-invasive status of BCa. METHODS: A total of 441 BCa patients were retrospectively enrolled from two centers and were divided into development (n=183), tuning (n=110), internal validation (n=73) and external validation (n=75) cohorts. The model was built based on nephrographic phase images of preoperative CT urography. Receiver operating characteristic (ROC) curves were performed and the area under the ROC curve (AUC) for discrimination between muscle-invasive BCa and non-muscle-invasive BCa was calculated. The performance of the model was evaluated and compared with that of the subjective assessment by two radiologists. RESULTS: The DL model exhibited relatively good performance in all cohorts [AUC: 0.861 in the internal validation cohort, 0.791 in the external validation cohort] and outperformed the two radiologists. The model yielded a sensitivity of 0.733, a specificity of 0.810 in the internal validation cohort and a sensitivity of 0.710 and a specificity of 0.773 in the external validation cohort. CONCLUSION: The proposed DL model based on CT images exhibited relatively good prediction ability of muscle-invasive status of BCa preoperatively, which may improve individual treatment of BCa.

Texture analysis based on quantitative magnetic resonance imaging to assess kidney function: a preliminary study.

BACKGROUND: Magnetic resonance imaging (MRI) has demonstrated its potential in the evaluation of renal function. Texture analysis (TA) is a novel technique to quantify tissue heterogeneity. We aim to investigate the feasibility of using TA based on the apparent diffusion coefficient (ADC), as well as T1 and T2 maps to evaluate renal function. METHODS: Patients with impaired renal function and subjects with a normal renal function who underwent renal diffusion weighted imaging (DWI), as well as T1 and T2 mapping at 3T, were prospectively enrolled. The participants were classified into four groups according to the estimated glomerular filtration rate (eGFR, mL/min/1.73 m2): normal (eGFR ≥90), mildly impaired (60≤ eGFR <90), moderately impaired (30≤ eGFR <60), and severely impaired (eGFR <30) renal function groups. Texture features quantified from the renal cortex or medulla were selected to build classifiers to discriminate different renal function groups by plotting receiver operating characteristic (ROC) curves and calculating the area under the curve (AUC), sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). RESULTS: In total, 116 candidates were included (94 patients and 22 healthy volunteers, mean age 37.9±14.9 years). There were 46 participants in the normal renal function group, 14 in the mildly impaired renal function group, 27 in the moderately impaired renal function group, and 29 in the severely impaired renal function group. Texture features from the ADC and T1 maps exhibited a good correlation to eGFR. The AUC, sensitivity, specificity, PPV, and NPV to differentiate between the normal and impaired renal function groups were 0.835, 0.792, 0.867, 0.905, and 0.722, respectively; to differentiate between the mildly impaired and moderately impaired groups were 0.937, 0.889, 0.857, 0.923, and 0.800, respectively; and to differentiate between the moderately impaired and severely impaired groups was 0.940, 0.759, 0.889, 0.880, and 0.774, respectively. CONCLUSIONS: TA based on ADC and T1 maps is feasible for evaluating renal function with relatively good accuracy.