Multimodal data integration for predicting progression risk in castration-resistant prostate cancer using deep learning: a multicenter retrospective study

Frontiers in Oncology, Journal Year: 2024, Volume and Issue: 14

Published: March 14, 2024

Purpose Patients with advanced prostate cancer (PCa) often develop castration-resistant PCa (CRPC) poor prognosis. Prognostic information obtained from multiparametric magnetic resonance imaging (mpMRI) and histopathology specimens can be effectively utilized through artificial intelligence (AI) techniques. The objective of this study is to construct an AI-based CRPC progress prediction model by integrating multimodal data. Methods materials Data 399 patients diagnosed at three medical centers between January 2018 2021 were collected retrospectively. We delineated regions interest (ROIs) 3 MRI sequences viz, T2WI, DWI, ADC a cropping tool extract the largest section each ROI. selected representative pathological hematoxylin eosin (H&E) slides for deep-learning training. A joint combined nomogram was constructed. ROC curves calibration plotted assess predictive performance goodness fit model. generated decision curve analysis (DCA) Kaplan–Meier (KM) survival evaluate clinical net benefit its association progression-free (PFS). Results AUC machine learning (ML) 0.755. best deep (DL) radiomics pathomics ResNet-50 model, 0.768 0.752, respectively. graph showed that DL contributed most, 0.86. DCA indicate had good ability benefit. KM indicated data guide patient prognosis management strategies. Conclusion integration improves risk progression CRPC.

Language: Английский

---

Does Size Matter? A Retrospective Study Analysing the Size of PI-RADS 4 Lesions and Its Associated Prostate Cancer Positivity with Transperineal Prostate Biopsy

Research and Reports in Urology, Journal Year: 2025, Volume and Issue: Volume 17, P. 49 - 57

Published: Feb. 1, 2025

Magnetic resonance imaging (MRI) is an essential tool in Prostate Cancer (PCa) diagnosis. PI-RADS v2.1 score correlates with clinically significant prostate cancer (CSPCa) and according to the most recent guidelines, prevalence of CSPCa 4 33-41%, while 5 62-79%. These groups are separated only by a size 15 mm yet difference risk significant. This study aims find threshold associated within group, which may be used combination other prostatic parameters, such as PSA density order help stratification patient counselling pre-biopsy setting. also aid surveillance smaller lesions setting negative biopsy avoid unnecessary repeat biopsies unless triggered threshold. A retrospective was performed data from 407 patients undergoing transperineal (TPPB) between April 2022 November 2023. subgroup included for analysis. ROC-AUC obtained. Median age 67 (interquartile range: 61-71) 0.20 range 0.13-0.28). correlated CSPCa: 1 2, frequency 10%; 3, it 20%; 4, 60%; 5, 80%, Pearson correlation = 0.51, p < 0.001. The Receiver Operating Characteristic Area Under Curve (ROC-AUC) determined 0.664 [0.579-0.7499]. optimal cut-off point 8.5 mm. Patients larger than had 2.31 times higher CSPCa. does matter useful predictor In our study, identified. provides specificity 80% detection.

Language: Английский

---

A retrospective study on predicting clinically significant prostate cancer via a bi-parametric ultrasound-based deep learning radiomics model

Frontiers in Oncology, Journal Year: 2025, Volume and Issue: 15

Published: April 8, 2025

This study aimed to establish and evaluate a model utilizing bi-parametric ultrasound-based deep learning radiomics (DLR) in conjunction with clinical factors anticipate clinically significant prostate cancer (csPCa). We retrospectively analyzed 232 participants from our institution who underwent both B-mode ultrasound shear wave elastography (SWE) prior biopsy between June 2022 December 2023. A random allocation placed the into training test cohorts 7:3 distribution. developed nomogram that integrates DLR within cohort, which was subsequently validated using cohort. The diagnostic performance applicability were evaluated receiver operating characteristic (ROC) curve analysis decision analysis. In study, demonstrated an area under (AUC) of 0.80 (95%CI: 0.70-0.91) set, surpassing models individually. By integrating factors, composite model, presented as nomogram, exhibited superior performance, achieving AUC 0.87 0.77-0.95) set. exceeded (P = 0.049) (AUC 0.79, 95%CI: 0.69-0.86, P 0.041). Furthermore, indicated provided greater net benefit across various high-risk threshold than or alone. To knowledge, this is first proposal indicators for predicting csPCa. can improve accuracy csPCa prediction may help physicians make more confident decisions regarding interventions, particularly settings where MRI unavailable.

Language: Английский

---

Impact of Nerve-Sparing Techniques on Prostate-Specific Antigen Persistence Following Robot-Assisted Radical Prostatectomy: A Multivariable Analysis of Clinical and Pathological Predictors

Diagnostics, Journal Year: 2025, Volume and Issue: 15(8), P. 987 - 987

Published: April 13, 2025

Background/Objectives: Prostate-specific antigen (PSA) persistence, defined as a postoperative PSA level ≥ 0.1 ng/mL measured within 4-8 weeks after radical prostatectomy (RP), predicts biochemical recurrence (BCR) and adverse oncological outcomes. The influence of nerve-sparing (NS) surgical techniques on persistence remains debated, especially among patients with high-risk pathological features. This study aimed to evaluate the impact NS following robot-assisted (RARP), considering tumor characteristics, parameters, patient-specific factors. Methods: A retrospective cohort analysis was performed 779 who underwent RARP at single institution between January 2002 December 2015. inclusion criteria consisted histologically confirmed prostate cancer available preoperative data, including measurements taken surgery. served primary outcome. Statistical analyses included descriptive statistics, univariate multivariable logistic regression models identify predictors Spearman's correlation along Kruskal-Wallis H test associations. Results: Of included, 55% surgery (51% unilateral, 49% bilateral). mean 11.85 (SD: 7.63), while 0.70 4.42). An elevated associated larger size (r = 0.1285, p < 0.001), advanced stages (χ2 45.10, 3.79 × 10-9), higher Gleason scores 24.74, 1.57 10-4). correlated lower (mean: 0.20 ng/mL) compared non-NS procedures 0.65 ng/mL), slight differences unilateral 0.30 bilateral 0.35 approaches. Multivariable identified stage (coefficient 1.16, 0.04) an independent predictor had no significant effect -0.01, 0.99). Conclusions: Nerve-sparing do not independently predict when adjusting for tumor-related factors confounders. Advanced stage, particularly pT3b, primarily determines persistence. These findings highlight necessity personalized planning informed by imaging patient-centered decision making optimize functional

Language: Английский

---

Interpretable multiparametric MRI radiomics-based machine learning model for preoperative differentiation between benign and malignant prostate masses: a diagnostic, multicenter study

Frontiers in Oncology, Journal Year: 2025, Volume and Issue: 15

Published: May 5, 2025

The study aimed to develop and externally validate multiparametric MRI (mpMRI) radiomics-based interpretable machine learning (ML) model for preoperative differentiating between benign malignant prostate masses. Patients who underwent mpMRI with suspected masses were retrospectively recruited from two independent hospitals May 2016 2023. mass regions in T2-weighted imaging (T2WI) diffusion-weighted (DWI) images segmented by ITK-SNAP. PyRadiomics was utilized extract radiomic features. Inter- intraobserver correlation analysis, t-test, Spearman the least absolute shrinkage selection operator (LASSO) algorithm a five-fold cross-validation applied feature selection. Five ML models built using chosen Model performance evaluated internal external validation, area under curve (AUC), calibration curves, decision analysis select optimal model. interpretability of most robust conducted via SHapley Additive exPlanation (SHAP). A total 567 patients enrolled, consisting training (n = 352), test 152), 63) sets. In total, 2,632 features extracted interest (ROIs) T2WI DWI images, which reduced 18 LASSO. established, among random forest (RF) presented best predictive ability, AUCs 0.929 (95% confidential interval [CI]: 0.885-0.963) 0.852 CI: 0.758-0.934) sets, respectively. analyses confirmed excellent clinical usefulness RF Besides, contributing relations uncovered SHAP. Radiomic combined facilitate accurate evaluation malignancy SHAP can disclose underlying prediction process model, may promote its applications.

Language: Английский

---

The detection rate for prostate cancer in systematic and targeted prostate biopsy in biopsy‐naive patients, according to the localization of the lesion at the mpMRI: A single‐center retrospective observational study

The Prostate, Journal Year: 2024, Volume and Issue: 84(13), P. 1234 - 1243

Published: June 25, 2024

Evaluate the detection rates of systematic, targeted and combined cores at biopsy according to tumor positions in biopsy-naïve patients.

Language: Английский

---

The Correlation Between Body Mass Index and Prostate Volume: A Retrospective Analysis of Pre and Postoperative Measurements in Prostate Cancer Patients

The Prostate, Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 23, 2024

ABSTRACT Background This study aims to assess the relationship between body mass index (BMI) and prostate volume, utilizing pre postoperative measurements. Methods A retrospective, observational was conducted at a single site using data from an institutional database. Medical records of patients who underwent robot‐assisted radical prostatectomy were reviewed. Data included age, BMI, volumes measured through digital rectal exam (DRE), transrectal ultrasound (TRUS), magnetic resonance imaging (MRI), surgical specimen weight (SPW). Results total 168 identified in analysis. Spearman's correlation test revealed significant association BMI volume for all measurement methods, reporting r = 0.146 ( p 0.047) DRE, 0.268 < 0.0001) TRUS, 0.177 0.021) MRI 0.234 0.002) SPW. Linear regression analysis confirmed reporting, respectively, R 2 0.026 0.036) 0.076 0.038 0.011) 0.040 0.009) Notably, considering SPW best way estimate every increase predicted is 0.865gr. Conclusions demonstrates positive linear highlighting importance assessments.

Language: Английский

---

The Role of Multiparametric MRI in the Local Staging of Prostate Cancer

Frontiers in Bioscience-Elite, Journal Year: 2023, Volume and Issue: 15(3)

Published: Sept. 20, 2023

Prostate cancer ranks as the second most frequently diagnosed globally among men and stands fifth leading cause of cancer-related death in males. Hence, an early precise diagnosis staging are critical. Traditional is based on clinical nomograms but presents a lower performance than prostate multiparametric magnetic resonance imaging (mpMRI). Since tumor serves basis for risk stratification, prognosis, treatment decision-making, primary objective mpMRI to distinguish between organ-confined locally advanced diseases. Therefore, this modality has emerged optimal selection local cancer, offering incremental value evaluating pelvic nodal disease bone involvement, supplying supplementary insights regarding location extension. As per Imaging Reporting & Data System v2.1 guideline, comprehensive accurate requires several key sequences, which include T1-weighted (T1WI) T2-weighted (T2WI) morphological assessment, with T2WI serving cornerstone staging. Additionally, diffusion-weighted (DWI) dynamic sequences acquired intravenous administration paramagnetic contrast medium (DCE) crucial components. It worth noting that while MRI exhibits high specificity, its sensitivity diagnosing extracapsular extension, seminal vesicle invasion, lymph node metastases limited. Moreover, own constraints not effective detecting distant or nodes, extended dissection remains gold standard. This review aims highlight significance provide practical approach assessing invasions, involvement adjacent organs nodes.

Language: Английский

---

iPCa-Net: A CNN-based framework for predicting incidental prostate cancer using multiparametric MRI

Computerized Medical Imaging and Graphics, Journal Year: 2023, Volume and Issue: 110, P. 102309 - 102309

Published: Nov. 1, 2023

Language: Английский

---

Risk Biomarkers for Biochemical Recurrence after Radical Prostatectomy for Prostate Cancer Using Clinical and MRI-Derived Semantic Features

Cancers, Journal Year: 2023, Volume and Issue: 15(21), P. 5296 - 5296

Published: Nov. 5, 2023

Objectives: This study aimed to assess the impact of covariates derived from a predictive model for detecting extracapsular extension on pathology (pECE+) biochemical recurrence-free survival (BCRFS) within 4 years after robotic-assisted radical prostatectomy (RARP). Methods: Retrospective data analysis was conducted single center between 2015 and 2022. Variables under consideration included prostate-specific antigen (PSA) levels, patient age, prostate volume, MRI semantic features, Grade Group (GG). We also assessed influence pECE+ positive surgical margins BCRFS. To attain these goals, we used Kaplan–Meier function multivariable Cox regression model. Additionally, analyzed features BCR (biochemical recurrence) in low/intermediate risk patients. Results: A total 177 participants with follow-up exceeding 6 months post-RARP were included. The 1-year, 2-year, 4-year risks 5%, 13%, 21%, respectively. non-parametric approach showed that adverse such as macroscopic ECE (mECE+), capsular disruption, high tumor contact length (TCCL), GG ≥ 4, (PSM), factors BCR. In low/intermediate-risk patients (pECE− < 4), presence has been shown increase Conclusions: highlights importance incorporating pre-surgery influencing early outcomes clinical decision making; mECE+, TCCL, based pre-surgical biopsy independent prognostic can assist identifying who will benefit closer monitoring.

Language: Английский

---