Deep Learning Prostate MRI Segmentation Accuracy and Robustness: A Systematic Review.

"Just Accepted" papers have undergone full peer review and have been accepted for publication in Radiology: Artificial Intelligence. This article will undergo copyediting, layout, and proof review before it is published in its final version. Please note that during production of the final copyedited article, errors may be discovered which could affect the content. Purpose To investigate the accuracy and robustness of prostate segmentation using deep learning across various training data sizes, MRI vendors, prostate zones, and testing methods relative to fellowship-trained diagnostic radiologists. Materials and methods In this systematic review, EMBASE, PubMed, Scopus and Web of Science databases were queried for English-language articles using keywords and related terms for prostate MRI segmentation and deep learning algorithms dated to July 31, 2022. A total of 691 articles from the search query were collected, and subsequently filtered to 48 based on predefined inclusion and exclusion criteria. Multiple characteristics were extracted from selected studies, such as deep learning algorithm performance, MRI vendor, and training dataset features. The primary outcome was comparison of mean Dice similarity coefficient (DSC) for prostate segmentation for deep learning algorithms versus diagnostic radiologists. Results Forty-eight studies were included. The vast majority of published deep learning algorithms for whole prostate gland segmentation (39/42 or 93%) had a DSC at or above expert level (DSC ≥ 0.86). The mean DSC was 0.79 ± 0.06 for peripheral zone, 0.87 ± 0.05 for transition zone, and 0.90 ± 0.04 for whole prostate gland segmentation. For selected studies using one major MRI vendor, the mean DSCs of each were as follows: GE (3/48 studies) 0.92 ± 0.03, Philips (4/48 studies) 0.92 ± 0.02, and Siemens (6/48 studies) 0.91 ± 0.03. Conclusion Deep learning algorithms for prostate MRI segmentation demonstrated comparable accuracy to expert radiologists despite varying parameters, therefore future research should shift toward evaluating segmentation robustness and patient outcomes across diverse clinical settings. ©RSNA, 2024.

Integrated Automatic Examination Assignment Reduces Radiologist Interruptions: A 2-Year Cohort Study of 232,022 Examinations.

Radiology departments face challenges in delivering timely and accurate imaging reports, especially in high-volume, subspecialized settings. In this retrospective cohort study at a tertiary cancer center, we assessed the efficacy of an Automatic Assignment System (AAS) in improving radiology workflow efficiency by analyzing 232,022 CT examinations over a 12-month period post-implementation and compared it to a historical control period. The AAS was integrated with the hospital-wide scheduling system and set up to automatically prioritize and distribute unreported CT examinations to available radiologists based on upcoming patient appointments, coupled with an email notification system. Following this AAS implementation, despite a 9% rise in CT volume, coupled with a concurrent 8% increase in the number of available radiologists, the mean daily urgent radiology report requests (URR) significantly decreased by 60% (25 ± 12 to 10 ± 5, t = -17.6, p < 0.001), and URR during peak days (95th quantile) was reduced by 52.2% from 46 to 22 requests. Additionally, the mean turnaround time (TAT) for reporting was significantly reduced by 440 min for patients without immediate appointments and by 86 min for those with same-day appointments. Lastly, patient waiting time sampled in one of the outpatient clinics was not negatively affected. These results demonstrate that AAS can substantially decrease workflow interruptions and improve reporting efficiency.

First in-human evaluation of [1-13C]pyruvate in D2O for hyperpolarized MRI of the brain: A safety and feasibility study.

PURPOSE: To investigate the safety and value of hyperpolarized (HP) MRI of [1-13C]pyruvate in healthy volunteers using deuterium oxide (D2O) as a solvent. METHODS: Healthy volunteers (n = 5), were injected with HP [1-13C]pyruvate dissolved in D2O and imaged with a metabolite-specific 3D dual-echo dynamic EPI sequence at 3T at one site (Site 1). Volunteers were monitored following the procedure to assess safety. Image characteristics, including SNR, were compared to data acquired in a separate cohort using water as a solvent (n = 5) at another site (Site 2). The apparent spin-lattice relaxation time (T1) of [1-13C]pyruvate was determined both in vitro and in vivo from a mono-exponential fit to the image intensity at each time point of our dynamic data. RESULTS: All volunteers completed the study safely and reported no adverse effects. The use of D2O increased the T1 of [1-13C]pyruvate from 66.5 ± 1.6 s to 92.1 ± 5.1 s in vitro, which resulted in an increase in signal by a factor of 1.46 ± 0.03 at the time of injection (90 s after dissolution). The use of D2O also increased the apparent relaxation time of [1-13C]pyruvate by a factor of 1.4 ± 0.2 in vivo. After adjusting for inter-site SNR differences, the use of D2O was shown to increase image SNR by a factor of 2.6 ± 0.2 in humans. CONCLUSIONS: HP [1-13C]pyruvate in D2O is safe for human imaging and provides an increase in T1 and SNR that may improve image quality.

Bridging Communication Gaps Between Radiologists, Referring Physicians, and Patients Through Standardized Structured Cancer Imaging Reporting: The Experience with Female Pelvic MRI Assessment Using O-RADS and a Simulated Cohort Patient Group.

RATIONALE AND OBJECTIVES: This study aimed to evaluate whether implementing structured reporting based on Ovarian-Adnexal Reporting and Data System (O-RADS) magnetic resonance imaging (MRI) in women with sonographically indeterminate adnexal masses improves communication between radiologists, referrers, and patients/caregivers and enhances diagnostic performance for determining adnexal malignancy. MATERIALS AND METHODS: We retrospectively analyzed prospectively issued MRI reports in 2019-2022 performed for characterizing adnexal masses before and after implementing O-RADS MRI; 56 patients/caregivers and nine gynecologic oncologists ("referrers") were surveyed about report interpretability/clarity/satisfaction; responses for pre- and post-implementation reports were compared using Fisher's exact and Chi-squared tests. Diagnostic performance was assessed using receiver operating characteristic curves. RESULTS: A total of 123 reports from before and 119 reports from after O-RADS MRI implementation were included. Survey response rates were 35.7% (20/56) for patients/caregivers and 66.7% (6/9) for referrers. For patients/caregivers, O-RADS MRI reports were clearer (p < 0.001) and more satisfactory (p < 0.001) than unstructured reports, but interpretability did not differ significantly (p = 0.14), as 28.0% (28/100) of postimplementation and 38.0% (38/100) of preimplementation reports were considered difficult to interpret. For referrers, O-RADS MRI reports were clearer, more satisfactory, and easier to interpret (p < 0.001); only 1.3% (1/77) were considered difficult to interpret. For differentiating benign from malignant adnexal lesions, O-RADS MRI showed area under the curve of 0.92 (95% confidence interval [CI], 0.85-0.99), sensitivity of 0.81 (95% CI, 0.58-0.95), and specificity of 0.91 (95% CI, 0.83-0.96). Diagnostic performance of reports before implementation could not be calculated due to many different phrases used to describe the likelihood of malignancy. CONCLUSION: Implementing standardized structured reporting using O-RADS MRI for characterizing adnexal masses improved clarity and satisfaction for patients/caregivers and referrers. Interpretability improved for referrers but remained limited for patients/caregivers.

Deep learning for [18F]fluorodeoxyglucose-PET-CT classification in patients with lymphoma: a dual-centre retrospective analysis.

BACKGROUND: The rising global cancer burden has led to an increasing demand for imaging tests such as [18F]fluorodeoxyglucose ([18F]FDG)-PET-CT. To aid imaging specialists in dealing with high scan volumes, we aimed to train a deep learning artificial intelligence algorithm to classify [18F]FDG-PET-CT scans of patients with lymphoma with or without hypermetabolic tumour sites. METHODS: In this retrospective analysis we collected 16 583 [18F]FDG-PET-CTs of 5072 patients with lymphoma who had undergone PET-CT before or after treatment at the Memorial Sloa Kettering Cancer Center, New York, NY, USA. Using maximum intensity projection (MIP), three dimensional (3D) PET, and 3D CT data, our ResNet34-based deep learning model (Lymphoma Artificial Reader System [LARS]) for [18F]FDG-PET-CT binary classification (Deauville 1-3 vs 4-5), was trained on 80% of the dataset, and tested on 20% of this dataset. For external testing, 1000 [18F]FDG-PET-CTs were obtained from a second centre (Medical University of Vienna, Vienna, Austria). Seven model variants were evaluated, including MIP-based LARS-avg (optimised for accuracy) and LARS-max (optimised for sensitivity), and 3D PET-CT-based LARS-ptct. Following expert curation, areas under the curve (AUCs), accuracies, sensitivities, and specificities were calculated. FINDINGS: In the internal test cohort (3325 PET-CTs, 1012 patients), LARS-avg achieved an AUC of 0·949 (95% CI 0·942-0·956), accuracy of 0·890 (0·879-0·901), sensitivity of 0·868 (0·851-0·885), and specificity of 0·913 (0·899-0·925); LARS-max achieved an AUC of 0·949 (0·942-0·956), accuracy of 0·868 (0·858-0·879), sensitivity of 0·909 (0·896-0·924), and specificity of 0·826 (0·808-0·843); and LARS-ptct achieved an AUC of 0·939 (0·930-0·948), accuracy of 0·875 (0·864-0·887), sensitivity of 0·836 (0·817-0·855), and specificity of 0·915 (0·901-0·927). In the external test cohort (1000 PET-CTs, 503 patients), LARS-avg achieved an AUC of 0·953 (0·938-0·966), accuracy of 0·907 (0·888-0·925), sensitivity of 0·874 (0·843-0·904), and specificity of 0·949 (0·921-0·960); LARS-max achieved an AUC of 0·952 (0·937-0·965), accuracy of 0·898 (0·878-0·916), sensitivity of 0·899 (0·871-0·926), and specificity of 0·897 (0·871-0·922); and LARS-ptct achieved an AUC of 0·932 (0·915-0·948), accuracy of 0·870 (0·850-0·891), sensitivity of 0·827 (0·793-0·863), and specificity of 0·913 (0·889-0·937). INTERPRETATION: Deep learning accurately distinguishes between [18F]FDG-PET-CT scans of lymphoma patients with and without hypermetabolic tumour sites. Deep learning might therefore be potentially useful to rule out the presence of metabolically active disease in such patients, or serve as a second reader or decision support tool. FUNDING: National Institutes of Health-National Cancer Institute Cancer Center Support Grant.

A preoperative nomogram incorporating CT to predict the probability of ovarian clear cell carcinoma.

OBJECTIVES: To evaluate clinical, laboratory, and radiological variables from preoperative contrast-enhanced computed tomography (CECT) for their ability to distinguish ovarian clear cell carcinoma (OCCC) from non-OCCC and to develop a nomogram to preoperatively predict the probability of OCCC. METHODS: This IRB-approved, retrospective study included consecutive patients who underwent surgery for an ovarian tumor from 1/1/2000 to 12/31/2016 and CECT of the abdomen and pelvis ≤90 days before primary debulking surgery. Using a standardized form, two experienced oncologic radiologists independently analyzed imaging features and provided a subjective 5-point impression of the probability of the histological diagnosis. Nomogram models incorporating clinical, laboratory, and radiological features were created to predict histological diagnosis of OCCC over non-OCCC. RESULTS: The final analysis included 533 patients with surgically confirmed OCCC (n = 61) and non-OCCC (n = 472); history of endometriosis was more often found in patients with OCCC (20% versus 3.6%; p < 0.001), while CA-125 was significantly higher in patients with non-OCCC (351 ng/mL versus 70 ng/mL; p < 0.001). A nomogram model incorporating clinical (age, history of endometriosis and adenomyosis), laboratory (CA-125) and imaging findings (peritoneal implant distribution, morphology, laterality, and diameter of ovarian lesion and of the largest solid component) had an AUC of 0.9 (95% CI: 0.847, 0.949), which was comparable to the AUCs of the experienced radiologists' subjective impressions [0.8 (95% CI: 0.822, 0.891) and 0.9 (95% CI: 0.865, 0.936)]. CONCLUSIONS: A presurgical nomogram model incorporating readily accessible clinical, laboratory, and CECT variables was a powerful predictor of OCCC, a subtype often requiring a distinctive treatment approach.

Radiology 2040.

A translation of this article in Spanish is available in the supplement. Una traducción de este artículo en español está disponible en el suplemento.

Fat-containing adnexal masses on MRI: solid tissue volume and fat distribution as a guide for O-RADS Score assignment.

PURPOSE: To explore ways to improve O-RADS MRI scoring for fat-containing adnexal masses, by investigating methods for quantifying solid tissue volume and fat distribution and evaluating their associations with malignancy. METHODS: This retrospective, single-center study included patients with fat-containing adnexal masses on MRI during 2008-2021. Two radiologists independently reviewed overall size (Sizeoverall), size of any solid tissue (Sizeanysolid), size of solid tissue that was not Rokitansky nodule (Sizenon-Rokitansky), and fat distribution. Wilcoxon test, Fisher-exact test, and ROC curve analysis were performed. Reference standard was pathology or follow-up > 24 months. RESULTS: 188 women (median age 35 years) with 163 benign and 25 malignant lesions were included. Sizeoverall (R1, 9.9 cm vs 5.9 cm; R2, 12.4 cm vs 6.0 cm), Sizeanysolid (R1, 5.1 cm vs 1.2 cm; R2, 3.2 cm vs 0.0 cm), Sizenon-Rokitansky (R1, 5.1 cm vs 0.0 cm; R2, 3.1 cm vs 0.0 cm), and fat distribution differed significantly between malignant and benign lesions (p < 0.01). Area under ROC curve was greatest using Sizenon-Rokitansky (R1, 0.83; R2, 0.86) vs Sizeoverall (R1, 0.78; R2, 0.81) or Sizeanysolid (R1, 0.79; R2, 0.81), though differences were non-significant (p = 0.48-0.93). Cutoffs for Sizenon-Rokitansky (R1, ≥ 1.2 cm; R2, ≥ 1.0 cm) yielded sensitivity and specificity of 0.72 and 0.93 (R1) and 0.76 and 0.95 (R2). Among immature teratomas, 85.7% displayed scattered fat. CONCLUSION: Overall size, size of (any or non-Rokitansky-nodule) solid tissue, and fat distribution differed between benign and malignant fat-containing adnexal masses. Incorporating these would constitute simple and practical approaches to refining O-RADS MRI scoring.

Programmatic Implementation of a Custom Subspecialized Oncologic Imaging Workflow Manager at a Tertiary Cancer Center.

PURPOSE: To evaluate whether a custom programmatic workflow manager reduces reporting turnaround times (TATs) from a body oncologic imaging workflow at a tertiary cancer center. METHODS: A custom software program was developed and implemented in the programming language R. Other aspects of the workflow were left unchanged. TATs were measured over a 12-month period (June-May). The same prior 12-month period served as a historical control. Median TATs of magnetic resonance imaging (MRI) and computed tomography (CT) examinations were compared with a Wilcoxon test. A chi-square test was used to compare the numbers of examinations reported within 24 hours and after 72 hours as well as the proportions of examinations assigned according to individual radiologist preferences. RESULTS: For all MRI and CT examinations (124,507 in 2019/2020 and 138,601 in 2020/2021), the median TAT decreased from 4 (interquartile range: 1-22 hours) to 3 hours (1-17 hours). Reports completed within 24 hours increased from 78% (124,127) to 89% (138,601). For MRI, TAT decreased from 22 (5-49 hours) to 8 hours (2-21 hours), and reports completed within 24 hours increased from 55% (14,211) to 80% (23,744). For CT, TAT decreased from 3 (1-19 hours) to 2 hours (1-13 hours), and reports completed within 24 hours increased from 84% (82,342) to 92% (99,922). Delayed reports (with a TAT > 72 hours) decreased from 17.0% (4,176) to 2.2% (649) for MRI and from 2.5% (2,500) to 0.7% (745) for CT. All differences were statistically significant (P < .001). CONCLUSION: The custom workflow management software program significantly decreased MRI and CT report TATs.

MRI-Detectability of Clinically Significant Prostate Cancer Relates to Oncologic Outcomes After Prostatectomy.

INTRODUCTION/BACKGROUND: Magnetic resonance imaging (MRI) misses a proportion of "clinically significant" prostate cancers (csPC) as defined by histopathology criteria. The aim of this study was to analyze whether long-term oncologic outcomes differ between MRI-detectable and MRI-occult csPC. PATIENTS AND METHODS: Retrospective analysis of 1449 patients with pre-prostatectomy MRI and csPC on prostatectomy specimens (ie, Grade group ≥2 or extraprostatic spread) between 2001-2006. T2-weighted MRIs were classified according to the Prostate Imaging Reporting and Data System into MRI-occult (categories 1, 2), MRI-equivocal (category 3), and MRI-detectable (categories 4, 5). Cumulative incidence of biochemical recurrence (BCR), metastatic disease, and cancer-specific mortality, estimated with competing risk models. The median follow-up in survivors was 11.0 years (IQR: 8.9-13.1). RESULTS: In 188 (13%) cases, csPC was MRI-occult, 435 (30%) MRIs were equivocal, and 826 (57%) csPC were MRI-detectable. The 15-year cumulative incidence [95% CI] of BCR was 8.3% [2.2, 19.5] for MRI-occult cases, 17.4% [11.1, 24.8] for MRI-equivocal cases, and 43.3% [38.7, 47.8] for MRI-detectable cases (P < .001). The cumulative incidences of metastases were 0.61% [0.06, 3.1], 3.5% [1.5, 6.9], and 19.6% [15.4, 24.2] for MRI-occult, MRI-equivocal, and MRI-detectable cases, respectively (P < .001). There were no deaths from prostate cancer observed in patients with MRI-occult csPC, compared to an estimated 1.9% [0.54, 4.9], and 7.1 % [4.5, 10.6] for patients with MRI-equivocal and MRI-detectable cancer, respectively (P < .001). CONCLUSION: Oncologic outcomes after prostatectomy for csPC differ between MRI-occult and MRI-detectable lesions. Judging the clinical significance of a negative prostate MRI based on histopathologic surrogates alone might be misleading. MICROABSTRACT: Among 1449 patients with pre-prostatectomy MRI and clinically significant prostate cancer on prostatectomy histopathology, MRI-occult cancers (n = 188, 13%) were less likely to recur biochemically (8% vs. 43%, P < .001), metastasize (0.6% vs. 20%, P < .001), or lead to prostate cancer mortality (0% vs. 7%, P < .001) than MRI-detectable cancers (n = 826, 57%). MRI-occult cancers constitute a prognostically distinct subgroup among higher-grade prostate cancers.

The role of MRI in prostate cancer: current and future directions.

There has been an increasing role of magnetic resonance imaging (MRI) in the management of prostate cancer. MRI already plays an essential role in the detection and staging, with the introduction of functional MRI sequences. Recent advancements in radiomics and artificial intelligence are being tested to potentially improve detection, assessment of aggressiveness, and provide usefulness as a prognostic marker. MRI can improve pretreatment risk stratification and therefore selection of and follow-up of patients for active surveillance. MRI can also assist in guiding targeted biopsy, treatment planning and follow-up after treatment to assess local recurrence. MRI has gained importance in the evaluation of metastatic disease with emerging technology including whole-body MRI and integrated positron emission tomography/MRI, allowing for not only better detection but also quantification. The main goal of this article is to review the most recent advances on MRI in prostate cancer and provide insights into its potential clinical roles from the radiologist's perspective. In each of the sections, specific roles of MRI tailored to each clinical setting are discussed along with its strengths and weakness including already established material related to MRI and the introduction of recent advancements on MRI.

Automatic Forecasting of Radiology Examination Volume Trends for Optimal Resource Planning and Allocation.

The aim of the study was to evaluate the performance of the Prophet forecasting procedure, part of the Facebook open-source Artificial Intelligence portfolio, for forecasting variations in radiological examination volumes. Daily CT and MRI examination volumes from our institution were extracted from the radiology information system (RIS) database. Data from January 1, 2015, to December 31, 2019, was used for training the Prophet algorithm, and data from January 2020 was used for validation. Algorithm performance was then evaluated prospectively in February and August 2020. Total error and mean error per day were evaluated, and computational time was logged using different Markov chain Monte Carlo (MCMC) samples. Data from 610,570 examinations were used for training; the majority were CTs (82.3%). During retrospective testing, prediction error was reduced from 19 to < 1 per day in CT (total 589 to 17) and from 5 to < 1 per day (total 144 to 27) in MRI by fine-tuning the Prophet procedure. Prospective prediction error in February was 11 per day in CT (9934 predicted, 9667 actual) and 1 per day in MRI (2484 predicted, 2457 actual) and was significantly better than manual weekly predictions (p = 0.001). Inference with MCMC added no substantial improvements while vastly increasing computational time. Prophet accurately models weekly, seasonal, and overall trends paving the way for optimal resource allocation for radiology exam acquisition and interpretation.

Prognostic Utility of MRI Features in Intradiverticular Bladder Tumor.

BACKGROUND: Intradiverticular bladder tumors (IDBT) are rare but clinically important, as they are difficult to assess endoscopically due to limited anatomic access and risk of perforation. MRI may be helpful in assessing IDBT and providing relevant staging and prognostic information. PURPOSE: To assess MRI findings of IDBT and their relationship with overall survival. METHODS: This retrospective study included 31 consecutive patients with IDBT undergoing MRI from 2008 to 2018 identified through electronic medical records and PACS database search. Two radiologists independently assessed the following MRI features: size (>3 vs ≤3 cm), diverticular neck involvement, Vesical Imaging-Reporting and Data System (VI-RADS) score (>3 vs ≤3), perivesical fat infiltration, additional tumors and suspicious pelvic lymph nodes. Overall survival was estimated using Kaplan-Meier analysis; and the relationship with clinicopathological and MRI features was determined using the Cox proportional-hazards regression model. Inter-reader agreement was assessed using intraclass correlation coefficients (ICC) and Cohen's kappa (K). RESULTS: Median follow-up was 1044 days (interquartile range, 474-1952 days). Twenty-six (83.9%) patients underwent surgical treatment with or without neoadjuvant chemotherapy. On MRI, greater tumor size (>3 cm), diverticular neck involvement, perivesical extension, and suspicious lymph nodes were associated with lower overall survival (HR = 3.6-8.1 and 4.3-6.3 for the 2 radiologists, p ≤ 0.03). Other clinicopathological or MRI findings were not associated with survival (p = 0.27-0.65). Inter-reader agreement was excellent for tumor size (ICC = 0.991; 95% CI 0.982-0.996), fair for VI-RADS (K = 0.52, 95% CI, 0.22-0.82), and moderate for others (K = 0.61-0.79). CONCLUSION: In patients with IDBT, several MRI features were significantly associated with overall survival. Utilizing all available clinicopathological and imaging information may improve estimation of prognosis.

Correlation Between Imaging-Based Intermediate Endpoints and Overall Survival in Men With Metastatic Castration-Resistant Prostate Cancer: Analysis of 28 Randomized Trials Using the Prostate Cancer Clinical Trials Working Group (PCWG2) Criteria in 16,511 Patients.

INTRODUCTION/BACKGROUND: Radiographic progression-free survival (rPFS) based on Prostate Cancer Working Group 2 (PCWG2) has been increasingly used as a meaningful imaging-based intermediate endpoint (IBIE) for overall survival (OS) in patients with metastatic castration-resistant prostate cancer (mCRPC). In randomized phase III trials, rPFS showed good correlation with OS at the individual trial level. We aimed to assess the correlation between the hazard ratios (HR) of IBIE and OS among PCWG2-based randomized trials. MATERIALS AND METHODS: PubMed and EMBASE databases were systematically searched for randomized trials evaluating systemic treatments on mCRPC using PCWG2 up to April 15, 2020. Hazard ratios for OS and IBIEs were extracted and their correlation was assessed using weighted linear regression. Subgroup analyses were performed according to various clinical settings: prior chemotherapy, drug category, type of IBIE (rPFS vs. composite IBIE, latter defined as progression by imaging and one or a combination of PSA, pain, skeletal-related events, and performance status), and publication year. RESULTS: Twenty-eight phase II-III randomized trials (16,511 patients) were included. Correlation between OS and IBIE was good (R2 = 0.57, 95% confidence interval [CI], 0.35-0.78). Trials using rPFS showed substantially higher correlation than those using a composite IBIE (R2 = 0.58, 95% CI, 0.32-0.82 vs. 0.00, 95% CI, -0.01 to 0.01). Correlations between OS and IBIE in other subgroups were at least moderate in nearly all subgroups (R2 = 0.32-0.91). CONCLUSION: IBIEs in the era of PCWG2 correlate well with OS in randomized trials for systemic drugs in patients with mCRPC. PCWG2-based rPFS should be used instead of a composite IBIE that includes PSA and other clinical variables.

Local Extent of Prostate Cancer at MRI versus Prostatectomy Histopathology: Associations with Long-term Oncologic Outcomes.

Background It is unknown how the imperfect accuracy of MRI for local staging of prostate cancer relates to oncologic outcomes. Purpose To analyze how staging discordances between MRI and histopathologic evaluation relate to recurrence and survival after radical prostatectomy. Materials and Methods Health Insurance Portability and Accountability Act-compliant retrospective analysis of preprostatectomy T2-weighted prostate MRI (January 2001 to December 2006). Extraprostatic extension and seminal vesicle invasion were assessed by using five-point Likert scales; scores of 4 or higher were classified as positive. Biochemical recurrence (BCR), metastases, and prostate cancer-specific mortality rates were estimated with Kaplan-Meier and Cox models. Results A total of 2160 patients (median age, 60 years; interquartile range, 55-64 years) were evaluated. Among patients with histopathologic extraprostatic (pT3) disease (683 of 2160; 32%), those with organ-confined disease at MRI (384 of 683; 56%) experienced better outcomes than those with concordant extraprostatic disease at MRI and pathologic analysis: 15-year risk for BCR, 30% (95% CI: 22, 40) versus 68% (95% CI: 60, 75); risk for metastases, 14% (95% CI: 8.4, 24) versus 32% (95% CI: 26, 39); risk for prostate cancer-specific mortality, 3% (95% CI: 1, 6) versus 15% (95% CI: 9.5, 23) (P < .001 for all comparisons). Among patients with histopathologic organ-confined disease (pT2) (1477 of 2160; 68%), those with extraprostatic disease at MRI (102 of 1477; 7%) were at higher risk for BCR (27% [95% CI: 19, 37] vs 10% [95% CI: 8, 14]; P < .001), metastases (19% [95% CI: 6, 48] vs 3% [95% CI: 1, 6]; P < .001), and prostate cancer-specific mortality (2% [95% CI: 1, 9] vs 1% [95% CI: 0, 5]; P = .009) than those with concordant organ-confined disease at MRI and pathologic analysis. At multivariable analyses, tumor extent at MRI (hazard ratio range, 4.1-5.2) and histopathologic evaluation (hazard ratio range, 3.6-6.7) was associated with the risk for BCR, metastases, and prostate cancer-specific mortality (P < .001 for all analyses). Conclusion The local extent of prostate cancer at MRI is associated with oncologic outcomes after prostatectomy, independent of pathologic tumor stage. This might inform a strategy on how to integrate MRI into a clinical staging algorithm. © RSNA, 2022 Online supplemental material is available for this article. See also the editorial by Gottlieb in this issue.

Impact of 18F-Fluorodeoxyglucose positron emission tomography on management of cancer of unknown primary: systematic review and meta-analysis.

BACKGROUND: Cancers of unknown primary (CUP) have traditionally been treated empirically, with a dismal prognosis. Compared with standard diagnostic tests, including CT and MRI, imaging with 18F-fluorodeoxyglucose (FDG) PET or PET/CT has shown the capacity to better identify the primary tumour site and detect additional sites of metastasis. However, its clinical impact is not well established. We performed a systematic review and meta-analysis of prior studies to assess the impact of FDG-PET or PET/CT on the management of patients with CUP. MATERIALS AND METHODS: Pubmed and EMBASE databases were searched up to 4th February 2021. Studies that reported the proportion of patients with CUP who experienced a management change after FDG-PET or PET/ computed tomography (CT) were included and the proportions were pooled using the random-effects model. Study quality was assessed using QUADAS-2. Subgroup analysis was conducted to explore heterogeneity. RESULTS: Thirty-eight studies (involving 2795 patients) were included. The pooled proportion of patients with management changes was 35% (95% confidence interval 31%-40%). There was substantial heterogeneity among the studies (Q-test, p < 0.01; I2 = 82%). The specific reason for management change was more commonly detection of the primary site (22% [95% CI 18-28%]) than detection of additional metastatic sites (14% [95% CI 10-19%]). The pooled proportions of patients with management changes were similar among numerous subgroups (range, 32.8%-38.2%). CONCLUSION: FDG-PET or PET/CT had a meaningful impact on the management of patients with CUP. Approximately, a third of patients had their management changed because of FDG-PET or PET/CT results, and this finding was consistent across numerous subgroups.

Emergency room imaging in pediatric patients with cancer: analysis of the spectrum and frequency of imaging modalities and findings in a tertiary cancer center and their relationship with survival.

BACKGROUND: To assess the spectrum and frequency of modalities used for emergency room (ER) imaging and their findings in pediatric cancer patients and assess their relationship with survival. METHODS: Consecutive pediatric cancer patients that underwent imaging during an ER visit at our tertiary cancer center over a 5-year period were retrospectively analyzed. Imaging findings were considered positive when they were relevant to the ER presenting complaint. Imaging positivity was correlated with inpatient admission. Overall survival (OS) was assessed with Kaplan-Meier curves and uni- and multi-variate Cox proportional hazards model was used to identify significant factors associated with OS. RESULTS: Two hundred sixty-one patients (135 males and 126 females; median age 11 years [interquartile range 5-16 years] with 348 visits and a total of 406 imaging studies were included. Common chief complaints were related to the chest (100 [28.7 %]) and fever (99 [28.4 %]). ER imaging was positive in 207 visits (59.5 %), commonly revealing increased metastases (50 [14.4 %]), pneumonia (47 [13.5 %]), and other lung problems (12 [2.9 %]). Positive ER imaging was associated with inpatient admission (69.3 % [133/192] vs. 40.4 % [63/156], p < 0.01). Multivariate survival analysis showed that positive ER imaging (hazard ratio [HR] = 2.35 [95% CI 1.44-3.83, p < 0.01), admission (HR = 1.86 [95% CI 1.17-3.00], p < 0.01), number of ER visits (HR = 3.08 [95% CI 1.62-5.83], p < 0.01 for ≥ 3 visits) were associated with poorer survival. CONCLUSIONS: Imaging was able to delineate the cause for ER visits in children with cancer in over half of the cases. Positive ER imaging was associated with admission and worse survival.

Patterns of Metastatic Disease in Patients with Cancer Derived from Natural Language Processing of Structured CT Radiology Reports over a 10-year Period.

Background Patterns of metastasis in cancer are increasingly relevant to prognostication and treatment planning but have historically been documented by means of autopsy series. Purpose To show the feasibility of using natural language processing (NLP) to gather accurate data from radiology reports for assessing spatial and temporal patterns of metastatic spread in a large patient cohort. Materials and Methods In this retrospective longitudinal study, consecutive patients who underwent CT from July 2009 to April 2019 and whose CT reports followed a departmental structured template were included. Three radiologists manually curated a sample of 2219 reports for the presence or absence of metastases across 13 organs; these manually curated reports were used to develop three NLP models with an 80%-20% split for training and test sets. A separate random sample of 448 manually curated reports was used for validation. Model performance was measured by accuracy, precision, and recall for each organ. The best-performing NLP model was used to generate a final database of metastatic disease across all patients. For each cancer type, statistical descriptive reports were provided by analyzing the frequencies of metastatic disease at the report and patient levels. Results In 91 665 patients (mean age ± standard deviation, 61 years ± 15; 46 939 women), 387 359 reports were labeled. The best-performing NLP model achieved accuracies from 90% to 99% across all organs. Metastases were most frequently reported in abdominopelvic (23.6% of all reports) and thoracic (17.6%) nodes, followed by lungs (14.7%), liver (13.7%), and bones (9.9%). Metastatic disease tropism is distinct among common cancers, with the most common first site being bones in prostate and breast cancers and liver among pancreatic and colorectal cancers. Conclusion Natural language processing may be applied to cancer patients' CT reports to generate a large database of metastatic phenotypes. Such a database could be combined with genomic studies and used to explore prognostic imaging phenotypes with relevance to treatment planning. © RSNA, 2021 Online supplemental material is available for this article.