Optimizing radiomics for prostate cancer diagnosis: feature selection strategies, machine learning classifiers, and MRI sequences.

OBJECTIVES: Radiomics-based analyses encompass multiple steps, leading to ambiguity regarding the optimal approaches for enhancing model performance. This study compares the effect of several feature selection methods, machine learning (ML) classifiers, and sources of radiomic features, on models' performance for the diagnosis of clinically significant prostate cancer (csPCa) from bi-parametric MRI. METHODS: Two multi-centric datasets, with 465 and 204 patients each, were used to extract 1246 radiomic features per patient and MRI sequence. Ten feature selection methods, such as Boruta, mRMRe, ReliefF, recursive feature elimination (RFE), random forest (RF) variable importance, L1-lasso, etc., four ML classifiers, namely SVM, RF, LASSO, and boosted generalized linear model (GLM), and three sets of radiomics features, derived from T2w images, ADC maps, and their combination, were used to develop predictive models of csPCa. Their performance was evaluated in a nested cross-validation and externally, using seven performance metrics. RESULTS: In total, 480 models were developed. In nested cross-validation, the best model combined Boruta with Boosted GLM (AUC = 0.71, F1 = 0.76). In external validation, the best model combined L1-lasso with boosted GLM (AUC = 0.71, F1 = 0.47). Overall, Boruta, RFE, L1-lasso, and RF variable importance were the top-performing feature selection methods, while the choice of ML classifier didn't significantly affect the results. The ADC-derived features showed the highest discriminatory power with T2w-derived features being less informative, while their combination did not lead to improved performance. CONCLUSION: The choice of feature selection method and the source of radiomic features have a profound effect on the models' performance for csPCa diagnosis. CRITICAL RELEVANCE STATEMENT: This work may guide future radiomic research, paving the way for the development of more effective and reliable radiomic models; not only for advancing prostate cancer diagnostic strategies, but also for informing broader applications of radiomics in different medical contexts. KEY POINTS: Radiomics is a growing field that can still be optimized. Feature selection method impacts radiomics models' performance more than ML algorithms. Best feature selection methods: RFE, LASSO, RF, and Boruta. ADC-derived radiomic features yield more robust models compared to T2w-derived radiomic features.

Synthetic data generation methods in healthcare: A review on open-source tools and methods.

Synthetic data generation has emerged as a promising solution to overcome the challenges which are posed by data scarcity and privacy concerns, as well as, to address the need for training artificial intelligence (AI) algorithms on unbiased data with sufficient sample size and statistical power. Our review explores the application and efficacy of synthetic data methods in healthcare considering the diversity of medical data. To this end, we systematically searched the PubMed and Scopus databases with a great focus on tabular, imaging, radiomics, time-series, and omics data. Studies involving multi-modal synthetic data generation were also explored. The type of method used for the synthetic data generation process was identified in each study and was categorized into statistical, probabilistic, machine learning, and deep learning. Emphasis was given to the programming languages used for the implementation of each method. Our evaluation revealed that the majority of the studies utilize synthetic data generators to: (i) reduce the cost and time required for clinical trials for rare diseases and conditions, (ii) enhance the predictive power of AI models in personalized medicine, (iii) ensure the delivery of fair treatment recommendations across diverse patient populations, and (iv) enable researchers to access high-quality, representative multimodal datasets without exposing sensitive patient information, among others. We underline the wide use of deep learning based synthetic data generators in 72.6 % of the included studies, with 75.3 % of the generators being implemented in Python. A thorough documentation of open-source repositories is finally provided to accelerate research in the field.

ProLesA-Net: A multi-channel 3D architecture for prostate MRI lesion segmentation with multi-scale channel and spatial attentions.

Prostate cancer diagnosis and treatment relies on precise MRI lesion segmentation, a challenge notably for small (<15 mm) and intermediate (15-30 mm) lesions. Our study introduces ProLesA-Net, a multi-channel 3D deep-learning architecture with multi-scale squeeze and excitation and attention gate mechanisms. Tested against six models across two datasets, ProLesA-Net significantly outperformed in key metrics: Dice score increased by 2.2%, and Hausdorff distance and average surface distance improved by 0.5 mm, with recall and precision also undergoing enhancements. Specifically, for lesions under 15 mm, our model showed a notable increase in five key metrics. In summary, ProLesA-Net consistently ranked at the top, demonstrating enhanced performance and stability. This advancement addresses crucial challenges in prostate lesion segmentation, enhancing clinical decision making and expediting treatment processes.

Well-being trajectories in breast cancer and their predictors: A machine-learning approach.

OBJECTIVE: This study aimed to describe distinct trajectories of anxiety/depression symptoms and overall health status/quality of life over a period of 18 months following a breast cancer diagnosis, and identify the medical, socio-demographic, lifestyle, and psychological factors that predict these trajectories. METHODS: 474 females (mean age = 55.79 years) were enrolled in the first weeks after surgery or biopsy. Data from seven assessment points over 18 months, at 3-month intervals, were used. The two outcomes were assessed at all points. Potential predictors were assessed at baseline and the first follow-up. Machine-Learning techniques were used to detect latent patterns of change and identify the most important predictors. RESULTS: Five trajectories were identified for each outcome: stably high, high with fluctuations, recovery, deteriorating/delayed response, and stably poor well-being (chronic distress). Psychological factors (i.e., negative affect, coping, sense of control, social support), age, and a few medical variables (e.g., symptoms, immune-related inflammation) predicted patients' participation in the delayed response and the chronic distress trajectories versus all other trajectories. CONCLUSIONS: There is a strong possibility that resilience does not always reflect a stable response pattern, as there might be some interim fluctuations. The use of machine-learning techniques provides a unique opportunity for the identification of illness trajectories and a shortlist of major bio/behavioral predictors. This will facilitate the development of early interventions to prevent a significant deterioration in patient well-being.

Deep learning-based segmentation of prostatic urethra on computed tomography scans for treatment planning.

Background and purpose: The intraprostatic urethra is an organ at risk in prostate cancer radiotherapy, but its segmentation in computed tomography (CT) is challenging. This work sought to: i) propose an automatic pipeline for intraprostatic urethra segmentation in CT, ii) analyze the dose to the urethra, iii) compare the predictions to magnetic resonance (MR) contours. Materials and methods: First, we trained Deep Learning networks to segment the rectum, bladder, prostate, and seminal vesicles. Then, the proposed Deep Learning Urethra Segmentation model was trained with the bladder and prostate distance transforms and 44 labeled CT with visible catheters. The evaluation was performed on 11 datasets, calculating centerline distance (CLD) and percentage of centerline within 3.5 and 5 mm. We applied this method to a dataset of 32 patients treated with intensity-modulated radiation therapy (IMRT) to quantify the urethral dose. Finally, we compared predicted intraprostatic urethra contours to manual delineations in MR for 15 patients without catheter. Results: A mean CLD of 1.6 ± 0.8 mm for the whole urethra and 1.7 ± 1.4, 1.5 ± 0.9, and 1.7 ± 0.9 mm for the top, middle, and bottom thirds were obtained in CT. On average, 94% and 97% of the segmented centerlines were within a 3.5 mm and 5 mm radius, respectively. In IMRT, the urethra received a higher dose than the overall prostate. We also found a slight deviation between the predicted and manual MR delineations. Conclusion: A fully-automatic segmentation pipeline was validated to delineate the intraprostatic urethra in CT images.

A segmentation-based method improving the performance of N4 bias field correction on T2weighted MR imaging data of the prostate.

Magnetic Resonance (MR) images suffer from spatial inhomogeneity, known as bias field corruption. The N4ITK filter is a state-of-the-art method used for correcting the bias field to optimize MR-based quantification. In this study, a novel approach is presented to quantitatively evaluate the performance of N4 bias field correction for pelvic prostate imaging. An exploratory analysis, regarding the different values of convergence threshold, shrink factor, fitting level, number of iterations and use of mask, is performed to quantify the performance of N4 filter in pelvic MR images. The performance of a total of 240 different N4 configurations is examined using the Full Width at Half Maximum (FWHM) of the segmented periprostatic fat distribution as evaluation metric. Phantom T2weighted images were used to assess the performance of N4 for a uniform test tissue mimicking material, excluding factors such as patient related susceptibility and anatomy heterogeneity. Moreover, 89 and 204 T2weighted patient images from two public datasets acquired by scanners with a combined surface and endorectal coil at 1.5 T and a surface coil at 3 T, respectively, were utilized and corrected with a variable set of N4 parameters. Furthermore, two external public datasets were used to validate the performance of the N4 filter in T2weighted patient images acquired by various scanning conditions with different magnetic field strengths and coils. The results show that the set of N4 parameters, converging to optimal representations of fat in the image, were: convergence threshold 0.001, shrink factor 2, fitting level 6, number of iterations 100 and the use of default mask for prostate images acquired by a combined surface and endorectal coil at both 1.5 T and 3 T. The corresponding optimal N4 configuration for MR prostate images acquired by a surface coil at 1.5 T or 3 T was: convergence threshold 0.001, shrink factor 2, fitting level 5, number of iterations 25 and the use of default mask. Hence, periprostatic fat segmentation can be used to define the optimal settings for achieving T2weighted prostate images free from bias field corruption to provide robust input for further analysis.

Personalized prediction of one-year mental health deterioration using adaptive learning algorithms: a multicenter breast cancer prospective study.

Identifying individual patient characteristics that contribute to long-term mental health deterioration following diagnosis of breast cancer (BC) is critical in clinical practice. The present study employed a supervised machine learning pipeline to address this issue in a subset of data from a prospective, multinational cohort of women diagnosed with stage I-III BC with a curative treatment intention. Patients were classified as displaying stable HADS scores (Stable Group; n = 328) or reporting a significant increase in symptomatology between BC diagnosis and 12 months later (Deteriorated Group; n = 50). Sociodemographic, life-style, psychosocial, and medical variables collected on the first visit to their oncologist and three months later served as potential predictors of patient risk stratification. The flexible and comprehensive machine learning (ML) pipeline used entailed feature selection, model training, validation and testing. Model-agnostic analyses aided interpretation of model results at the variable- and patient-level. The two groups were discriminated with a high degree of accuracy (Area Under the Curve = 0.864) and a fair balance of sensitivity (0.85) and specificity (0.87). Both psychological (negative affect, certain coping with cancer reactions, lack of sense of control/positive expectations, and difficulties in regulating negative emotions) and biological variables (baseline percentage of neutrophils, thrombocyte count) emerged as important predictors of mental health deterioration in the long run. Personalized break-down profiles revealed the relative impact of specific variables toward successful model predictions for each patient. Identifying key risk factors for mental health deterioration is an essential first step toward prevention. Supervised ML models may guide clinical recommendations toward successful illness adaptation.

Region-adaptive magnetic resonance image enhancement for improving CNN-based segmentation of the prostate and prostatic zones.

Automatic segmentation of the prostate of and the prostatic zones on MRI remains one of the most compelling research areas. While different image enhancement techniques are emerging as powerful tools for improving the performance of segmentation algorithms, their application still lacks consensus due to contrasting evidence regarding performance improvement and cross-model stability, further hampered by the inability to explain models' predictions. Particularly, for prostate segmentation, the effectiveness of image enhancement on different Convolutional Neural Networks (CNN) remains largely unexplored. The present work introduces a novel image enhancement method, named RACLAHE, to enhance the performance of CNN models for segmenting the prostate's gland and the prostatic zones. The improvement in performance and consistency across five CNN models (U-Net, U-Net++, U-Net3+, ResU-net and USE-NET) is compared against four popular image enhancement methods. Additionally, a methodology is proposed to explain, both quantitatively and qualitatively, the relation between saliency maps and ground truth probability maps. Overall, RACLAHE was the most consistent image enhancement algorithm in terms of performance improvement across CNN models with the mean increase in Dice Score ranging from 3 to 9% for the different prostatic regions, while achieving minimal inter-model variability. The integration of a feature driven methodology to explain the predictions after applying image enhancement methods, enables the development of a concrete, trustworthy automated pipeline for prostate segmentation on MR images.

Trajectories and Predictors of Depression After Breast Cancer Diagnosis: A 1-year longitudinal study.

Being diagnosed with breast cancer (BC) can be a traumatic experience for patients who may experience symptoms of depression. In order to facilitate the prevention of such symptoms, it is crucial to understand how and why depressive symptoms emerge and evolve for each individual, from diagnosis through treatment and recovery. In the present work, data from a multicentric study of 706 BC patients followed for 12 months are analyzed. First, a trajectory-based unsupervised clustering based on K-means is performed to capture the dynamic patterns of change in patients' depressive symptoms after BC diagnosis and to identify distinct trajectory clusters. Then a supervised learning approach was employed to build a classification model of depression progression and to identify potential predictors. Patients were clustered into 4 groups: stable low, stable high, improving, and worsening depressive symptoms. In a nested cross-validation pipeline, the performance of the Support Vector Machine model for discriminating between "good" and "poor" progression was 0.78±0.05 in terms of AUC. Several psychological variables emerged as highly predictive of the evolution of depressive symptoms with the most important ones being negative affectivity and anxious preoccupation. Clinical Relevance-The findings of the present study may help clinicians tailor individualized psychological interventions aiming at alleviating the burden of these symptoms in women with breast cancer and improving their overall well-being.

Predicting the need for mechanical ventilation and mortality in hospitalized COVID-19 patients who received heparin.

Although several studies have utilized AI (artificial intelligence)-based solutions to enhance the decision making for mechanical ventilation, as well as, for mortality in COVID-19, the extraction of explainable predictors regarding heparin's effect in intensive care and mortality has been left unresolved. In the present study, we developed an explainable AI (XAI) workflow to shed light into predictors for admission in the intensive care unit (ICU), as well as, for mortality across those hospitalized COVID-19 patients who received heparin. AI empowered classifiers, such as, the hybrid Extreme gradient boosting (HXGBoost) with customized loss functions were trained on time-series curated clinical data to develop robust AI models. Shapley additive explanation analysis (SHAP) was conducted to determine the positive or negative impact of the predictors in the model's output. The HXGBoost predicted the risk for intensive care and mortality with 0.84 and 0.85 accuracy, respectively. SHAP analysis indicated that the low percentage of lymphocytes at day 7 along with increased FiO2 at days 1 and 5, low SatO2 at days 3 and 7 increase the probability for mortality and highlight the positive effect of heparin administration at the early days of hospitalization for reducing mortality.

ICU admission and mortality classifiers for COVID-19 patients based on subgroups of dynamically associated profiles across multiple timepoints.

The coronavirus disease 2019 (COVID-19) which is caused by severe acute respiratory syndrome coronavirus type 2 (SARS-CoV-2) is consistently causing profound wounds in the global healthcare system due to its increased transmissibility. Currently, there is an urgent unmet need to identify the underlying dynamic associations among COVID-19 patients and distinguish patient subgroups with common clinical profiles towards the development of robust classifiers for ICU admission and mortality. To address this need, we propose a four step pipeline which: (i) enhances the quality of multiple timeseries clinical data through an automated data curation workflow, (ii) deploys Dynamic Bayesian Networks (DBNs) for the detection of features with increased connectivity based on dynamic association analysis across multiple points, (iii) utilizes Self Organizing Maps (SOMs) and trajectory analysis for the early identification of COVID-19 patients with common clinical profiles, and (iv) trains robust multiple additive regression trees (MART) for ICU admission and mortality classification based on the extracted homogeneous clusters, to identify risk factors and biomarkers for disease progression. The contribution of the extracted clusters and the dynamically associated clinical data improved the classification performance for ICU admission to sensitivity 0.83 and specificity 0.83, and for mortality to sensitivity 0.74 and specificity 0.76. Additional information was included to enhance the performance of the classifiers yielding an increase by 4% in sensitivity and specificity for mortality. According to the risk factor analysis, the number of lymphocytes, SatO2, PO2/FiO2, and O2 supply type were highlighted as risk factors for ICU admission and the percentage of neutrophils and lymphocytes, PO2/FiO2, LDH, and ALP for mortality, among others. To our knowledge, this is the first study that combines dynamic modeling with clustering analysis to identify homogeneous groups of COVID-19 patients towards the development of robust classifiers for ICU admission and mortality.

Prediction of Poor Mental Health Following Breast Cancer Diagnosis Using Random Forests1.

Breast cancer diagnosis has been associated with poor mental health, with significant impairment of quality of life. In order to ensure support for successful adaptation to this illness, it is of paramount importance to identify the most prominent factors affecting well-being that allow for accurate prediction of mental health status across time. Here we exploit a rich set of clinical, psychological, socio-demographic and lifestyle data from a large multicentre study of patients recently diagnosed with breast cancer, in order to classify patients based on their mental health status and further identify potential predictors of such status. For this purpose, a supervised learning pipeline using cross-sectional data was implemented for the formulation of a classification scheme of mental health status 6 months after diagnosis. Model performance in terms of AUC ranged from 0.81± 0.04 to 0.90± 0.03. Several psychological variables, including initial levels of anxiety and depression, emerged as highly predictive of short-term mental health status of women diagnosed with breast cancer.

Planning With Patient-Specific Rectal Sub-Region Constraints Decreases Probability of Toxicity in Prostate Cancer Radiotherapy.

Background: A rectal sub-region (SRR) has been previously identified by voxel-wise analysis in the inferior-anterior part of the rectum as highly predictive of rectal bleeding (RB) in prostate cancer radiotherapy. Translating the SRR to patient-specific radiotherapy planning is challenging as new constraints have to be defined. A recent geometry-based model proposed to optimize the planning by determining the achievable mean doses (AMDs) to the organs at risk (OARs), taking into account the overlap between the planning target volume (PTV) and OAR. The aim of this study was to quantify the SRR dose sparing by using the AMD model in the planning, while preserving the dose to the prostate. Material and Methods: Three-dimensional volumetric modulated arc therapy (VMAT) planning dose distributions for 60 patients were computed following four different strategies, delivering 78 Gy to the prostate, while meeting the genitourinary group dose constraints to the OAR: (i) a standard plan corresponding to the standard practice for rectum sparing (STDpl), (ii) a plan adding constraints to SRR (SRRpl), (iii) a plan using the AMD model applied to the rectum only (AMD_RECTpl), and (iv) a final plan using the AMD model applied to both the rectum and the SRR (AMD_RECT_SRRpl). After PTV dose normalization, plans were compared with regard to dose distributions, quality, and estimated risk of RB using a normal tissue complication probability model. Results: AMD_RECT_SRRpl showed the largest SRR dose sparing, with significant mean dose reductions of 7.7, 3, and 2.3 Gy, with respect to the STDpl, SRRpl, and AMD_RECTpl, respectively. AMD_RECT_SRRpl also decreased the mean rectal dose by 3.6 Gy relative to STDpl and by 3.3 Gy relative to SRRpl. The absolute risk of grade ≥1 RB decreased from 22.8% using STDpl planning to 17.6% using AMD_RECT_SRRpl considering SRR volume. AMD_RECT_SRRpl plans, however, showed slightly less dose homogeneity and significant increase of the number of monitor units, compared to the three other strategies. Conclusion: Compared to a standard prostate planning, applying dose constraints to a patient-specific SRR by using the achievable mean dose model decreased the mean dose by 7.7 Gy to the SRR and may decrease the relative risk of RB by 22%.

Comparison of CBCT-based dose calculation methods in head and neck cancer radiotherapy: from Hounsfield unit to density calibration curve to deep learning.

PURPOSE: Anatomical variations occur during head and neck (H&N) radiotherapy treatment. kV cone-beam computed tomography (CBCT) images can be used for daily dose monitoring to assess dose variations owing to anatomic changes. Deep learning methods (DLMs) have recently been proposed to generate pseudo-CT (pCT) from CBCT to perform dose calculation. This study aims to evaluate the accuracy of a DLM and to compare this method with three existing methods of dose calculation from CBCT in H&N cancer radiotherapy. METHODS: Forty-four patients received VMAT for H&N cancer (70-63-56 Gy). For each patient, reference CT (Bigbore, Philips) and CBCT images (XVI, Elekta) were acquired. The DLM was based on a generative adversarial network. The three compared methods were: (a) a method using a density to Hounsfield Unit (HU) relation from phantom CBCT image (HU-D curve method), (b) a water-air-bone density assignment method (DAM), and iii) a method using deformable image registration (DIR). The imaging endpoints were the mean absolute error (MAE) and mean error (ME) of HU from pCT and reference CT (CTref ). The dosimetric endpoints were dose discrepancies and 3D gamma analyses (local, 2%/2 mm, 30% dose threshold). Dose discrepancies were defined as the mean absolute differences between DVHs calculated from the CTref and pCT of each method. RESULTS: In the entire body, the MAEs and MEs of the DLM, HU-D curve method, DAM, and DIR method were 82.4 and 17.1 HU, 266.6 and 208.9 HU, 113.2 and 14.2 HU, and 95.5 and -36.6 HU, respectively. The MAE obtained using the DLM differed significantly from those of other methods (Wilcoxon, P ≤ 0.05). The DLM dose discrepancies were 7 ± 8 cGy (maximum = 44 cGy) for the ipsilateral parotid gland Dmean and 5 ± 6 cGy (max = 26 cGy) for the contralateral parotid gland mean dose (Dmean ). For the parotid gland Dmean , no significant dose difference was observed between the DLM and other methods. The mean 3D gamma pass rate ± standard deviation was 98.1 ± 1.2%, 91.0 ± 5.3%, 97.9 ± 1.6%, and 98.8 ± 0.7% for the DLM, HU-D method, DAM, and DIR method, respectively. The gamma pass rates and mean gamma results of the HU-D curve method, DAM, and DIR method differed significantly from those of the DLM. CONCLUSIONS: For H&N radiotherapy, DIR method and DLM appears as the most appealing CBCT-based dose calculation methods among the four methods in terms of dose accuracy as well as calculation time. Using the DIR method or DLM with CBCT images enables dose monitoring in the parotid glands during the treatment course and may be used to trigger replanning.

Rectal and Urethro-Vesical Subregions for Toxicity Prediction After Prostate Cancer Radiation Therapy: Validation of Voxel-Based Models in an Independent Population.

PURPOSE: Recent voxel-based studies have shown that the dose to specific rectal and urethro-vesical subregions is predictive of toxicities after prostate cancer intensity modulated radiation therapy. The objective of this study was to validate the discriminatory power of these subregions with respect to the whole organs in a large independent population. METHODS AND MATERIALS: The validation cohort consisted of 450 patients from the TROG03.04-RADAR trial treated with 3-dimensional conformal radiation therapy at 66 to 74 Gy. Previous voxel-based analyses identified an inferoanterior rectal subregion as predictive of rectal bleeding and 5 subregions in the urethra and the posterior and superior part of the bladder as predictive of urinary incontinence, dysuria, retention, and hematuria. In the validation cohort, these subregions were segmented in each patient's anatomy. Dose-volume histograms (DVHs) of the whole organs and the 6 subregions were compared bin-wise between patients with and without toxicities. The discriminatory power of DVHs for grade ≥2 toxicity endpoints was assessed using the area under the receiver operating characteristic curve (AUC). RESULTS: Subregion DVHs were significantly different between patients with and without toxicities for late rectal bleeding (V44-V74), acute urinary incontinence (V68-V72), late dysuria (V56-V68), and late retention (V14-V64). The dose to the rectal subregion and the whole rectum were equally predictive of rectal bleeding (V68; AUC = 0.61). The doses to 3 out of the 5 urethro-vesical subregions were found to be more predictive than the dose to the whole bladder: in the urethra for acute incontinence (V71 AUC = 0.69 vs V71 AUC = 0.66), in the posterior part of the bladder for late dysuria (V65 AUC = 0.66 vs V68 AUC = 0.59), and late retention (V39 AUC = 0.74 vs no significant AUC). CONCLUSIONS: Three subregions located in the urethra and the bladder were successfully validated as more predictive of urinary toxicity than the whole bladder for urinary incontinence, retention, and dysuria. Sparing the posterior part of the bladder in particular in treatment planning may reduce the risk of late urinary retention.

Local dose analysis to predict acute and late urinary toxicities after prostate cancer radiotherapy: Assessment of cohort and method effects.

PURPOSE: To perform bladder dose-surface map (DSM) analysis for (1) identifying symptom-related sub-surfaces (Ssurf) and evaluating their prediction capability of urinary toxicity, (2) comparing DSM with dose-volume map (DVM) (method effect), and (3) assessing the reproducibility of DSM (cohort effect). METHODS AND MATERIALS: Urinary toxicities were prospectively analyzed for 254 prostate cancer patients treated with IMRT/IGRT at 78/80 Gy. DSMs were generated by unfolding bladder surfaces in a 2D plane. Pixel-by-pixel analysis was performed to identify symptom-related Ssurf. Likewise, voxel-by-voxel DVM analysis was performed to identify sub-volumes (Svol). The prediction capability of Ssurf and Svol DVHs was assessed by logistic/Cox regression using the area under the ROC curve (AUC). The Ssurf localization and prediction capability were compared to (1) the Svol obtained by DVM analysis in the same cohort and (2) the Ssurf obtained from other DSM studies. RESULTS: Three Ssurf were identified in the bladder: posterior for acute retention (AUC = 0.64), posterior-superior for late retention (AUC = 0.68), and inferior-anterior-lateral for late dysuria (AUC = 0.73). Five Svol were identified: one in the urethra for acute incontinence and four in the posterior bladder part for acute and late retention, late dysuria, and hematuria. The overlap between Ssurf and Svol was moderate for acute retention, good for late retention, and bad for late dysuria, and AUCs ranged from 0.62 to 0.81. The prediction capabilities of Ssurf and Svol models were not significantly different. Among five symptoms comparable between cohorts, common Ssurf was found only for late dysuria, with a good spatial agreement. CONCLUSION: Spatial agreement between methods is relatively good although DVM identified more sub-regions. Reproducibility of identified Ssurf between cohorts is low.

Comparison of Deep Learning-Based and Patch-Based Methods for Pseudo-CT Generation in MRI-Based Prostate Dose Planning.

PURPOSE: Deep learning methods (DLMs) have recently been proposed to generate pseudo-CT (pCT) for magnetic resonance imaging (MRI) based dose planning. This study aims to evaluate and compare DLMs (U-Net and generative adversarial network [GAN]) using various loss functions (L2, single-scale perceptual loss [PL], multiscale PL, weighted multiscale PL) and a patch-based method (PBM). METHODS AND MATERIALS: Thirty-nine patients received a volumetric modulated arc therapy for prostate cancer (78 Gy). T2-weighted MRIs were acquired in addition to planning CTs. The pCTs were generated from the MRIs using 7 configurations: 4 GANs (L2, single-scale PL, multiscale PL, weighted multiscale PL), 2 U-Net (L2 and single-scale PL), and the PBM. The imaging endpoints were mean absolute error and mean error, in Hounsfield units, between the reference CT (CTref) and the pCT. Dose uncertainties were quantified as mean absolute differences between the dose volume histograms (DVHs) calculated from the CTref and pCT obtained by each method. Three-dimensional gamma indexes were analyzed. RESULTS: Considering the image uncertainties in the whole pelvis, GAN L2 and U-Net L2 showed the lowest mean absolute error (≤34.4 Hounsfield units). The mean errors were not different than 0 (P ≤ .05). The PBM provided the highest uncertainties. Very few DVH points differed when comparing GAN L2 or U-Net L2 DVHs and CTref DVHs (P ≤ .05). Their dose uncertainties were ≤0.6% for the prostate planning target Volume V95%, ≤0.5% for the rectum V70Gy, and ≤0.1% for the bladder V50Gy. The PBM, U-Net PL, and GAN PL presented the highest systematic dose uncertainties. The gamma pass rates were >99% for all DLMs. The mean calculation time to generate 1 pCT was 15 s for the DLMs and 62 min for the PBM. CONCLUSIONS: Generating pCT for MRI dose planning with DLMs and PBM provided low-dose uncertainties. In particular, the GAN L2 and U-Net L2 provided the lowest dose uncertainties together with a low computation time.

Voxel-Based Analysis for Identification of Urethrovesical Subregions Predicting Urinary Toxicity After Prostate Cancer Radiation Therapy.

PURPOSE: To apply a voxel-based analysis to identify urethrovesical symptom-related subregions (SRSs) associated with acute and late urinary toxicity in prostate cancer radiation therapy. METHODS AND MATERIALS: Two hundred seventy-two patients with prostate cancer treated with intensity-modulated radiation therapy/image-guided radiation therapy were analyzed prospectively. Each patient's computed tomography imaging was spatially normalized to a common coordinate system via nonrigid registration. The obtained deformation fields were used to map the dose of each patient to the common coordinate system. A voxel-based statistical analysis was applied to generate 3-dimensional dose-volume maps for different urinary symptoms, allowing the identification of corresponding SRSs with statistically significant dose differences between patients with or without toxicity. Each SRS was propagated back to each individual's native space, and dose-volume histograms (DVHs) for the SRSs and the whole bladder were computed. Logistic and Cox regression were used to estimate the SRS's prediction capability compared with the whole bladder. RESULTS: A local dose-effect relationship was found in the bladder and the urethra. SRSs were identified for 5 symptoms: acute incontinence in the urethra, acute retention in the bladder trigone, late retention and dysuria in the posterior part of the bladder, and late hematuria in the superior part of the bladder, with significant dose differences between patients with and without toxicity, ranging from 1.2 to 9.3 Gy. The doses to the SRSs were significantly predictive of toxicity, with maximum areas under the receiver operating characteristic curve of 0.73 for acute incontinence, 0.62 for acute retention, 0.70 for late retention, 0.81 for late dysuria, and 0.67 for late hematuria. The bladder DVH was predictive only for late retention, dysuria, and hematuria (area under the curve, 0.65-0.72). CONCLUSIONS: The dose delivered to the urethra and the posterior and superior parts of the bladder was predictive of acute incontinence and retention and of late retention, dysuria, and hematuria. The dose to the whole bladder was moderately predictive.

Multi-atlas-based segmentation of prostatic urethra from planning CT imaging to quantify dose distribution in prostate cancer radiotherapy.

BACKGROUND AND PURPOSE: Segmentation of intra-prostatic urethra for dose assessment from planning CT may help explaining urinary toxicity in prostate cancer radiotherapy. This work sought to: i) propose an automatic method for urethra segmentation in CT, ii) compare it with previously proposed surrogate models and iii) quantify the dose received by the urethra in patients treated with IMRT. MATERIALS AND METHODS: A weighted multi-atlas-based urethra segmentation method was devised from a training data set of 55 CT scans of patients receiving brachytherapy with visible urinary catheters. Leave-one-out cross validation was performed to quantify the error between the urethra segmentation and the catheter ground truth with two scores: the centerlines distance (CLD) and the percentage of centerline within a certain distance from the catheter (PWR). The segmentation method was then applied to a second test data set of 95 prostate cancer patients having received 78Gy IMRT to quantify dose to the urethra. RESULTS: Mean CLD was 3.25±1.2mm for the whole urethra and 3.7±1.7mm, 2.52±1.5mm, and 3.01±1.7mm for the top, middle, and bottom thirds, respectively. In average, 53% of the segmented centerlines were within a radius<3.5mm from the centerline ground truth and 83% in a radius<5mm. The proposed method outperformed existing surrogate models. In IMRT, urethra DVH was significantly higher than prostate DVH from V74Gy to V79Gy. CONCLUSION: A multi-atlas-based segmentation method was proposed enabling assessment of the dose within the prostatic urethra.