Assessing deep learning reconstruction for faster prostate MRI: visual vs. diagnostic performance metrics.

OBJECTIVE: Deep learning (DL) MRI reconstruction enables fast scan acquisition with good visual quality, but the diagnostic impact is often not assessed because of large reader study requirements. This study used existing diagnostic DL to assess the diagnostic quality of reconstructed images. MATERIALS AND METHODS: A retrospective multisite study of 1535 patients assessed biparametric prostate MRI between 2016 and 2020. Likely clinically significant prostate cancer (csPCa) lesions (PI-RADS ≥ 4) were delineated by expert radiologists. T2-weighted scans were retrospectively undersampled, simulating accelerated protocols. DL reconstruction (DLRecon) and diagnostic DL detection (DLDetect) were developed. The effect on the partial area under (pAUC), the Free-Response Operating Characteristic (FROC) curve, and the structural similarity (SSIM) were compared as metrics for diagnostic and visual quality, respectively. DLDetect was validated with a reader concordance analysis. Statistical analysis included Wilcoxon, permutation, and Cohen's kappa tests for visual quality, diagnostic performance, and reader concordance. RESULTS: DLRecon improved visual quality at 4- and 8-fold (R4, R8) subsampling rates, with SSIM (range: -1 to 1) improved to 0.78 ± 0.02 (p < 0.001) and 0.67 ± 0.03 (p < 0.001) from 0.68 ± 0.03 and 0.51 ± 0.03, respectively. However, diagnostic performance at R4 showed a pAUC FROC of 1.33 (CI 1.28-1.39) for DL and 1.29 (CI 1.23-1.35) for naive reconstructions, both significantly lower than fully sampled pAUC of 1.58 (DL: p = 0.024, naïve: p = 0.02). Similar trends were noted for R8. CONCLUSION: DL reconstruction produces visually appealing images but may reduce diagnostic accuracy. Incorporating diagnostic AI into the assessment framework offers a clinically relevant metric essential for adopting reconstruction models into clinical practice. CLINICAL RELEVANCE STATEMENT: In clinical settings, caution is warranted when using DL reconstruction for MRI scans. While it recovered visual quality, it failed to match the prostate cancer detection rates observed in scans not subjected to acceleration and DL reconstruction.

Using deep learning to optimize the prostate MRI protocol by assessing the diagnostic efficacy of MRI sequences.

PURPOSE: To explore diagnostic deep learning for optimizing the prostate MRI protocol by assessing the diagnostic efficacy of MRI sequences. METHOD: This retrospective study included 840 patients with a biparametric prostate MRI scan. The MRI protocol included a T2-weighted image, three DWI sequences (b50, b400, and b800 s/mm2), a calculated ADC map, and a calculated b1400 sequence. Two accelerated MRI protocols were simulated, using only two acquired b-values to calculate the ADC and b1400. Deep learning models were trained to detect prostate cancer lesions on accelerated and full protocols. The diagnostic performances of the protocols were compared on the patient-level with the area under the receiver operating characteristic (AUROC), using DeLong's test, and on the lesion-level with the partial area under the free response operating characteristic (pAUFROC), using a permutation test. Validation of the results was performed among expert radiologists. RESULTS: No significant differences in diagnostic performance were found between the accelerated protocols and the full bpMRI baseline. Omitting b800 reduced 53% DWI scan time, with a performance difference of + 0.01 AUROC (p = 0.20) and -0.03 pAUFROC (p = 0.45). Omitting b400 reduced 32% DWI scan time, with a performance difference of -0.01 AUROC (p = 0.65) and + 0.01 pAUFROC (p = 0.73). Multiple expert radiologists underlined the findings. CONCLUSIONS: This study shows that deep learning can assess the diagnostic efficacy of MRI sequences by comparing prostate MRI protocols on diagnostic accuracy. Omitting either the b400 or the b800 DWI sequence can optimize the prostate MRI protocol by reducing scan time without compromising diagnostic quality.

Artificial intelligence in medicine: mitigating risks and maximizing benefits via quality assurance, quality control, and acceptance testing.

The adoption of artificial intelligence (AI) tools in medicine poses challenges to existing clinical workflows. This commentary discusses the necessity of context-specific quality assurance (QA), emphasizing the need for robust QA measures with quality control (QC) procedures that encompass (1) acceptance testing (AT) before clinical use, (2) continuous QC monitoring, and (3) adequate user training. The discussion also covers essential components of AT and QA, illustrated with real-world examples. We also highlight what we see as the shared responsibility of manufacturers or vendors, regulators, healthcare systems, medical physicists, and clinicians to enact appropriate testing and oversight to ensure a safe and equitable transformation of medicine through AI.

Automated Quantitative Analysis of CT Perfusion to Classify Vascular Phenotypes of Pancreatic Ductal Adenocarcinoma.

CT perfusion (CTP) analysis is difficult to implement in clinical practice. Therefore, we investigated a novel semi-automated CTP AI biomarker and applied it to identify vascular phenotypes of pancreatic ductal adenocarcinoma (PDAC) and evaluate their association with overall survival (OS). METHODS: From January 2018 to November 2022, 107 PDAC patients were prospectively included, who needed to undergo CTP and a diagnostic contrast-enhanced CT (CECT). We developed a semi-automated CTP AI biomarker, through a process that involved deformable image registration, a deep learning segmentation model of tumor and pancreas parenchyma volume, and a trilinear non-parametric CTP curve model to extract the enhancement slope and peak enhancement in segmented tumors and pancreas. The biomarker was validated in terms of its use to predict vascular phenotypes and their association with OS. A receiver operating characteristic (ROC) analysis with five-fold cross-validation was performed. OS was assessed with Kaplan-Meier curves. Differences between phenotypes were tested using the Mann-Whitney U test. RESULTS: The final analysis included 92 patients, in whom 20 tumors (21%) were visually isovascular. The AI biomarker effectively discriminated tumor types, and isovascular tumors showed higher enhancement slopes (2.9 Hounsfield unit HU/s vs. 2.0 HU/s, p < 0.001) and peak enhancement (70 HU vs. 47 HU, p < 0.001); the AUC was 0.86. The AI biomarker's vascular phenotype significantly differed in OS (p < 0.01). CONCLUSIONS: The AI biomarker offers a promising tool for robust CTP analysis. In PDAC, it can distinguish vascular phenotypes with significant OS prognostication.

The Effect of Image Resampling on the Performance of Radiomics-Based Artificial Intelligence in Multicenter Prostate MRI.

BACKGROUND: Single center MRI radiomics models are sensitive to data heterogeneity, limiting the diagnostic capabilities of current prostate cancer (PCa) radiomics models. PURPOSE: To study the impact of image resampling on the diagnostic performance of radiomics in a multicenter prostate MRI setting. STUDY TYPE: Retrospective. POPULATION: Nine hundred thirty patients (nine centers, two vendors) with 737 eligible PCa lesions, randomly split into training (70%, N = 500), validation (10%, N = 89), and a held-out test set (20%, N = 148). FIELD STRENGTH/SEQUENCE: 1.5T and 3T scanners/T2-weighted imaging (T2W), diffusion-weighted imaging (DWI), and apparent diffusion coefficient maps. ASSESSMENT: A total of 48 normalized radiomics datasets were created using various resampling methods, including different target resolutions (T2W: 0.35, 0.5, and 0.8 mm; DWI: 1.37, 2, and 2.5 mm), dimensionalities (2D/3D) and interpolation techniques (nearest neighbor, linear, Bspline and Blackman windowed-sinc). Each of the datasets was used to train a radiomics model to detect clinically relevant PCa (International Society of Urological Pathology grade ≥ 2). Baseline models were constructed using 2D and 3D datasets without image resampling. The resampling configurations with highest validation performance were evaluated in the test dataset and compared to the baseline models. STATISTICAL TESTS: Area under the curve (AUC), DeLong test. The significance level used was 0.05. RESULTS: The best 2D resampling model (T2W: Bspline and 0.5 mm resolution, DWI: nearest neighbor and 2 mm resolution) significantly outperformed the 2D baseline (AUC: 0.77 vs. 0.64). The best 3D resampling model (T2W: linear and 0.8 mm resolution, DWI: nearest neighbor and 2.5 mm resolution) significantly outperformed the 3D baseline (AUC: 0.79 vs. 0.67). DATA CONCLUSION: Image resampling has a significant effect on the performance of multicenter radiomics artificial intelligence in prostate MRI. The recommended 2D resampling configuration is isotropic resampling with T2W at 0.5 mm (Bspline interpolation) and DWI at 2 mm (nearest neighbor interpolation). For the 3D radiomics, this work recommends isotropic resampling with T2W at 0.8 mm (linear interpolation) and DWI at 2.5 mm (nearest neighbor interpolation). EVIDENCE LEVEL: 3 TECHNICAL EFFICACY: Stage 2.

Complexities of deep learning-based undersampled MR image reconstruction.

Artificial intelligence has opened a new path of innovation in magnetic resonance (MR) image reconstruction of undersampled k-space acquisitions. This review offers readers an analysis of the current deep learning-based MR image reconstruction methods. The literature in this field shows exponential growth, both in volume and complexity, as the capabilities of machine learning in solving inverse problems such as image reconstruction are explored. We review the latest developments, aiming to assist researchers and radiologists who are developing new methods or seeking to provide valuable feedback. We shed light on key concepts by exploring the technical intricacies of MR image reconstruction, highlighting the importance of raw datasets and the difficulty of evaluating diagnostic value using standard metrics.Relevance statement Increasingly complex algorithms output reconstructed images that are difficult to assess for robustness and diagnostic quality, necessitating high-quality datasets and collaboration with radiologists.Key points• Deep learning-based image reconstruction algorithms are increasing both in complexity and performance.• The evaluation of reconstructed images may mistake perceived image quality for diagnostic value.• Collaboration with radiologists is crucial for advancing deep learning technology.

Semisupervised Learning with Report-guided Pseudo Labels for Deep Learning-based Prostate Cancer Detection Using Biparametric MRI.

Purpose: To evaluate a novel method of semisupervised learning (SSL) guided by automated sparse information from diagnostic reports to leverage additional data for deep learning-based malignancy detection in patients with clinically significant prostate cancer. Materials and Methods: This retrospective study included 7756 prostate MRI examinations (6380 patients) performed between January 2014 and December 2020 for model development. An SSL method, report-guided SSL (RG-SSL), was developed for detection of clinically significant prostate cancer using biparametric MRI. RG-SSL, supervised learning (SL), and state-of-the-art SSL methods were trained using 100, 300, 1000, or 3050 manually annotated examinations. Performance on detection of clinically significant prostate cancer by RG-SSL, SL, and SSL was compared on 300 unseen examinations from an external center with a histopathologically confirmed reference standard. Performance was evaluated using receiver operating characteristic (ROC) and free-response ROC analysis. P values for performance differences were generated with a permutation test. Results: At 100 manually annotated examinations, mean examination-based diagnostic area under the ROC curve (AUC) values for RG-SSL, SL, and the best SSL were 0.86 ± 0.01 (SD), 0.78 ± 0.03, and 0.81 ± 0.02, respectively. Lesion-based detection partial AUCs were 0.62 ± 0.02, 0.44 ± 0.04, and 0.48 ± 0.09, respectively. Examination-based performance of SL with 3050 examinations was matched by RG-SSL with 169 manually annotated examinations, thus requiring 14 times fewer annotations. Lesion-based performance was matched with 431 manually annotated examinations, requiring six times fewer annotations. Conclusion: RG-SSL outperformed SSL in clinically significant prostate cancer detection and achieved performance similar to SL even at very low annotation budgets.Keywords: Annotation Efficiency, Computer-aided Detection and Diagnosis, MRI, Prostate Cancer, Semisupervised Deep Learning Supplemental material is available for this article. Published under a CC BY 4.0 license.

Prototypical few-shot segmentation for cross-institution male pelvic structures with spatial registration.

The prowess that makes few-shot learning desirable in medical image analysis is the efficient use of the support image data, which are labelled to classify or segment new classes, a task that otherwise requires substantially more training images and expert annotations. This work describes a fully 3D prototypical few-shot segmentation algorithm, such that the trained networks can be effectively adapted to clinically interesting structures that are absent in training, using only a few labelled images from a different institute. First, to compensate for the widely recognised spatial variability between institutions in episodic adaptation of novel classes, a novel spatial registration mechanism is integrated into prototypical learning, consisting of a segmentation head and an spatial alignment module. Second, to assist the training with observed imperfect alignment, support mask conditioning module is proposed to further utilise the annotation available from the support images. Extensive experiments are presented in an application of segmenting eight anatomical structures important for interventional planning, using a data set of 589 pelvic T2-weighted MR images, acquired at seven institutes. The results demonstrate the efficacy in each of the 3D formulation, the spatial registration, and the support mask conditioning, all of which made positive contributions independently or collectively. Compared with the previously proposed 2D alternatives, the few-shot segmentation performance was improved with statistical significance, regardless whether the support data come from the same or different institutes.

Prediction Variability to Identify Reduced AI Performance in Cancer Diagnosis at MRI and CT.

Background A priori identification of patients at risk of artificial intelligence (AI) failure in diagnosing cancer would contribute to the safer clinical integration of diagnostic algorithms. Purpose To evaluate AI prediction variability as an uncertainty quantification (UQ) metric for identifying cases at risk of AI failure in diagnosing cancer at MRI and CT across different cancer types, data sets, and algorithms. Materials and Methods Multicenter data sets and publicly available AI algorithms from three previous studies that evaluated detection of pancreatic cancer on contrast-enhanced CT images, detection of prostate cancer on MRI scans, and prediction of pulmonary nodule malignancy on low-dose CT images were analyzed retrospectively. Each task's algorithm was extended to generate an uncertainty score based on ensemble prediction variability. AI accuracy percentage and partial area under the receiver operating characteristic curve (pAUC) were compared between certain and uncertain patient groups in a range of percentile thresholds (10%-90%) for the uncertainty score using permutation tests for statistical significance. The pulmonary nodule malignancy prediction algorithm was compared with 11 clinical readers for the certain group (CG) and uncertain group (UG). Results In total, 18 022 images were used for training and 838 images were used for testing. AI diagnostic accuracy was higher for the cases in the CG across all tasks (P < .001). At an 80% threshold of certain predictions, accuracy in the CG was 21%-29% higher than in the UG and 4%-6% higher than in the overall test data sets. The lesion-level pAUC in the CG was 0.25-0.39 higher than in the UG and 0.05-0.08 higher than in the overall test data sets (P < .001). For pulmonary nodule malignancy prediction, accuracy of AI was on par with clinicians for cases in the CG (AI results vs clinician results, 80% [95% CI: 76, 85] vs 78% [95% CI: 70, 87]; P = .07) but worse for cases in the UG (AI results vs clinician results, 50% [95% CI: 37, 64] vs 68% [95% CI: 60, 76]; P < .001). Conclusion An AI-prediction UQ metric consistently identified reduced performance of AI in cancer diagnosis. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Babyn in this issue.

Radiomics based automated quality assessment for T2W prostate MR images.

PURPOSE: The guidelines for prostate cancer recommend the use of MRI in the prostate cancer pathway. Due to the variability in prostate MR image quality, the reliability of this technique in the detection of prostate cancer is highly variable in clinical practice. This leads to the need for an objective and automated assessment of image quality to ensure an adequate acquisition and hereby to improve the reliability of MRI. The aim of this study is to investigate the feasibility of Blind/referenceless image spatial quality evaluator (Brisque) and radiomics in automated image quality assessment of T2-weighted (T2W) images. METHOD: Anonymized axial T2W images from 140 patients were scored for quality using a five-point Likert scale (low, suboptimal, acceptable, good, very good quality) in consensus by two readers. Images were dichotomized into clinically acceptable (very good, good and acceptable quality images) and clinically unacceptable (low and suboptimal quality images) in order to train and verify the model. Radiomics and Brisque features were extracted from a central cuboid volume including the prostate. A reduced feature set was used to fit a Linear Discriminant Analysis (LDA) model to predict image quality. Two hundred times repeated 5-fold cross-validation was used to train the model and test performance by assessing the classification accuracy, the discrimination accuracy as receiver operating curve - area under curve (ROC-AUC), and by generating confusion matrices. RESULTS: Thirty-four images were classified as clinically unacceptable and 106 were classified as clinically acceptable. The accuracy of the independent test set (mean ± standard deviation) was 85.4 ± 5.5%. The ROC-AUC was 0.856 (0.851 - 0.861) (mean; 95% confidence interval). CONCLUSIONS: Radiomics AI can automatically detect a significant portion of T2W images of suboptimal image quality. This can help improve image quality at the time of acquisition, thus reducing repeat scans and improving diagnostic accuracy.

Quantifiable Measures of Abdominal Wall Motion for Quality Assessment of Cine-MRI Slices in Detection of Abdominal Adhesions.

Abdominal adhesions present a diagnostic challenge, and classic imaging modalities can miss their presence. Cine-MRI, which records visceral sliding during patient-controlled breathing, has proven useful in detecting and mapping adhesions. However, patient movements can affect the accuracy of these images, despite there being no standardized algorithm for defining sufficiently high-quality images. This study aims to develop a biomarker for patient movements and determine which patient-related factors influence movement during cine-MRI. Included patients underwent cine-MRI to detect adhesions for chronic abdominal complaints, data were collected from electronic patient files and radiologic reports. Ninety slices of cine-MRI were assessed for quality, using a five-point scale to quantify amplitude, frequency, and slope, from which an image-processing algorithm was developed. The biomarkers closely correlated with qualitative assessments, with an amplitude of 6.5 mm used to distinguish between sufficient and insufficient-quality slices. In multivariable analysis, the amplitude of movement was influenced by age, sex, length, and the presence of a stoma. Unfortunately, no factor was changeable. Strategies for mitigating their impact may be challenging. This study highlights the utility of the developed biomarker in evaluating image quality and providing useful feedback for clinicians. Future studies could improve diagnostic quality by implementing automated quality criteria during cine-MRI.

Inter- and Intra-Observer Variability and the Effect of Experience in Cine-MRI for Adhesion Detection.

Cine-MRI for adhesion detection is a promising novel modality that can help the large group of patients developing pain after abdominal surgery. Few studies into its diagnostic accuracy are available, and none address observer variability. This retrospective study explores the inter- and intra-observer variability, diagnostic accuracy, and the effect of experience. A total of 15 observers with a variety of experience reviewed 61 sagittal cine-MRI slices, placing box annotations with a confidence score at locations suspect for adhesions. Five observers reviewed the slices again one year later. Inter- and intra-observer variability are quantified using Fleiss' (inter) and Cohen's (intra) κ and percentage agreement. Diagnostic accuracy is quantified with receiver operating characteristic (ROC) analysis based on a consensus standard. Inter-observer Fleiss' κ values range from 0.04 to 0.34, showing poor to fair agreement. High general and cine-MRI experience led to significantly (p < 0.001) better agreement among observers. The intra-observer results show Cohen's κ values between 0.37 and 0.53 for all observers, except one with a low κ of -0.11. Group AUC scores lie between 0.66 and 0.72, with individual observers reaching 0.78. This study confirms that cine-MRI can diagnose adhesions, with respect to a radiologist consensus panel and shows that experience improves reading cine-MRI. Observers without specific experience adapt to this modality quickly after a short online tutorial. Observer agreement is fair at best and area under the receiver operating characteristic curve (AUC) scores leave room for improvement. Consistently interpreting this novel modality needs further research, for instance, by developing reporting guidelines or artificial intelligence-based methods.

Automated deep-learning system in the assessment of MRI-visible prostate cancer: comparison of advanced zoomed diffusion-weighted imaging and conventional technique.

BACKGROUND: Deep-learning-based computer-aided diagnosis (DL-CAD) systems using MRI for prostate cancer (PCa) detection have demonstrated good performance. Nevertheless, DL-CAD systems are vulnerable to high heterogeneities in DWI, which can interfere with DL-CAD assessments and impair performance. This study aims to compare PCa detection of DL-CAD between zoomed-field-of-view echo-planar DWI (z-DWI) and full-field-of-view DWI (f-DWI) and find the risk factors affecting DL-CAD diagnostic efficiency. METHODS: This retrospective study enrolled 354 consecutive participants who underwent MRI including T2WI, f-DWI, and z-DWI because of clinically suspected PCa. A DL-CAD was used to compare the performance of f-DWI and z-DWI both on a patient level and lesion level. We used the area under the curve (AUC) of receiver operating characteristics analysis and alternative free-response receiver operating characteristics analysis to compare the performances of DL-CAD using f- DWI and z-DWI. The risk factors affecting the DL-CAD were analyzed using logistic regression analyses. P values less than 0.05 were considered statistically significant. RESULTS: DL-CAD with z-DWI had a significantly better overall accuracy than that with f-DWI both on patient level and lesion level (AUCpatient: 0.89 vs. 0.86; AUClesion: 0.86 vs. 0.76; P < .001). The contrast-to-noise ratio (CNR) of lesions in DWI was an independent risk factor of false positives (odds ratio [OR] = 1.12; P < .001). Rectal susceptibility artifacts, lesion diameter, and apparent diffusion coefficients (ADC) were independent risk factors of both false positives (ORrectal susceptibility artifact = 5.46; ORdiameter, = 1.12; ORADC = 0.998; all P < .001) and false negatives (ORrectal susceptibility artifact = 3.31; ORdiameter = 0.82; ORADC = 1.007; all P ≤ .03) of DL-CAD. CONCLUSIONS: Z-DWI has potential to improve the detection performance of a prostate MRI based DL-CAD. TRIAL REGISTRATION: ChiCTR, NO. ChiCTR2100041834 . Registered 7 January 2021.

AAPM task group report 273: Recommendations on best practices for AI and machine learning for computer-aided diagnosis in medical imaging.

Rapid advances in artificial intelligence (AI) and machine learning, and specifically in deep learning (DL) techniques, have enabled broad application of these methods in health care. The promise of the DL approach has spurred further interest in computer-aided diagnosis (CAD) development and applications using both "traditional" machine learning methods and newer DL-based methods. We use the term CAD-AI to refer to this expanded clinical decision support environment that uses traditional and DL-based AI methods. Numerous studies have been published to date on the development of machine learning tools for computer-aided, or AI-assisted, clinical tasks. However, most of these machine learning models are not ready for clinical deployment. It is of paramount importance to ensure that a clinical decision support tool undergoes proper training and rigorous validation of its generalizability and robustness before adoption for patient care in the clinic. To address these important issues, the American Association of Physicists in Medicine (AAPM) Computer-Aided Image Analysis Subcommittee (CADSC) is charged, in part, to develop recommendations on practices and standards for the development and performance assessment of computer-aided decision support systems. The committee has previously published two opinion papers on the evaluation of CAD systems and issues associated with user training and quality assurance of these systems in the clinic. With machine learning techniques continuing to evolve and CAD applications expanding to new stages of the patient care process, the current task group report considers the broader issues common to the development of most, if not all, CAD-AI applications and their translation from the bench to the clinic. The goal is to bring attention to the proper training and validation of machine learning algorithms that may improve their generalizability and reliability and accelerate the adoption of CAD-AI systems for clinical decision support.

A concurrent, deep learning-based computer-aided detection system for prostate multiparametric MRI: a performance study involving experienced and less-experienced radiologists.

OBJECTIVES: To evaluate the effect of a deep learning-based computer-aided diagnosis (DL-CAD) system on experienced and less-experienced radiologists in reading prostate mpMRI. METHODS: In this retrospective, multi-reader multi-case study, a consecutive set of 184 patients examined between 01/2018 and 08/2019 were enrolled. Ground truth was combined targeted and 12-core systematic transrectal ultrasound-guided biopsy. Four radiologists, two experienced and two less-experienced, evaluated each case twice, once without (DL-CAD-) and once assisted by DL-CAD (DL-CAD+). ROC analysis, sensitivities, specificities, PPV and NPV were calculated to compare the diagnostic accuracy for the diagnosis of prostate cancer (PCa) between the two groups (DL-CAD- vs. DL-CAD+). Spearman's correlation coefficients were evaluated to assess the relationship between PI-RADS category and Gleason score (GS). Also, the median reading times were compared for the two reading groups. RESULTS: In total, 172 patients were included in the final analysis. With DL-CAD assistance, the overall AUC of the less-experienced radiologists increased significantly from 0.66 to 0.80 (p = 0.001; cutoff ISUP GG ≥ 1) and from 0.68 to 0.80 (p = 0.002; cutoff ISUP GG ≥ 2). Experienced radiologists showed an AUC increase from 0.81 to 0.86 (p = 0.146; cutoff ISUP GG ≥ 1) and from 0.81 to 0.84 (p = 0.433; cutoff ISUP GG ≥ 2). Furthermore, the correlation between PI-RADS category and GS improved significantly in the DL-CAD + group (0.45 vs. 0.57; p = 0.03), while the median reading time was reduced from 157 to 150 s (p = 0.023). CONCLUSIONS: DL-CAD assistance increased the mean detection performance, with the most significant benefit for the less-experienced radiologist; with the help of DL-CAD less-experienced radiologists reached performances comparable to that of experienced radiologists. KEY POINTS: • DL-CAD used as a concurrent reading aid helps radiologists to distinguish between benign and cancerous lesions in prostate MRI. • With the help of DL-CAD, less-experienced radiologists may achieve detection performances comparable to that of experienced radiologists. • DL-CAD assistance increases the correlation between PI-RADS category and cancer grade.

Artificial intelligence for prostate MRI: open datasets, available applications, and grand challenges.

Artificial intelligence (AI) for prostate magnetic resonance imaging (MRI) is starting to play a clinical role for prostate cancer (PCa) patients. AI-assisted reading is feasible, allowing workflow reduction. A total of 3,369 multi-vendor prostate MRI cases are available in open datasets, acquired from 2003 to 2021 in Europe or USA at 3 T (n = 3,018; 89.6%) or 1.5 T (n = 296; 8.8%), 346 cases scanned with endorectal coil (10.3%), 3,023 (89.7%) with phased-array surface coils; 412 collected for anatomical segmentation tasks, 3,096 for PCa detection/classification; for 2,240 cases lesions delineation is available and 56 cases have matching histopathologic images; for 2,620 cases the PSA level is provided; the total size of all open datasets amounts to approximately 253 GB. Of note, quality of annotations provided per dataset highly differ and attention must be paid when using these datasets (e.g., data overlap). Seven grand challenges and commercial applications from eleven vendors are here considered. Few small studies provided prospective validation. More work is needed, in particular validation on large-scale multi-institutional, well-curated public datasets to test general applicability. Moreover, AI needs to be explored for clinical stages other than detection/characterization (e.g., follow-up, prognosis, interventions, and focal treatment).

Setting the Research Agenda for Clinical Artificial Intelligence in Pancreatic Adenocarcinoma Imaging.

Pancreatic ductal adenocarcinoma (PDAC), estimated to become the second leading cause of cancer deaths in western societies by 2030, was flagged as a neglected cancer by the European Commission and the United States Congress. Due to lack of investment in research and development, combined with a complex and aggressive tumour biology, PDAC overall survival has not significantly improved the past decades. Cross-sectional imaging and histopathology play a crucial role throughout the patient pathway. However, current clinical guidelines for diagnostic workup, patient stratification, treatment response assessment, and follow-up are non-uniform and lack evidence-based consensus. Artificial Intelligence (AI) can leverage multimodal data to improve patient outcomes, but PDAC AI research is too scattered and lacking in quality to be incorporated into clinical workflows. This review describes the patient pathway and derives touchpoints for image-based AI research in collaboration with a multi-disciplinary, multi-institutional expert panel. The literature exploring AI to address these touchpoints is thoroughly retrieved and analysed to identify the existing trends and knowledge gaps. The results show absence of multi-institutional, well-curated datasets, an essential building block for robust AI applications. Furthermore, most research is unimodal, does not use state-of-the-art AI techniques, and lacks reliable ground truth. Based on this, the future research agenda for clinically relevant, image-driven AI in PDAC is proposed.

The Medical Segmentation Decathlon.

International challenges have become the de facto standard for comparative assessment of image analysis algorithms. Although segmentation is the most widely investigated medical image processing task, the various challenges have been organized to focus only on specific clinical tasks. We organized the Medical Segmentation Decathlon (MSD)-a biomedical image analysis challenge, in which algorithms compete in a multitude of both tasks and modalities to investigate the hypothesis that a method capable of performing well on multiple tasks will generalize well to a previously unseen task and potentially outperform a custom-designed solution. MSD results confirmed this hypothesis, moreover, MSD winner continued generalizing well to a wide range of other clinical problems for the next two years. Three main conclusions can be drawn from this study: (1) state-of-the-art image segmentation algorithms generalize well when retrained on unseen tasks; (2) consistent algorithmic performance across multiple tasks is a strong surrogate of algorithmic generalizability; (3) the training of accurate AI segmentation models is now commoditized to scientists that are not versed in AI model training.

A deep learning masked segmentation alternative to manual segmentation in biparametric MRI prostate cancer radiomics.

OBJECTIVES: To determine the value of a deep learning masked (DLM) auto-fixed volume of interest (VOI) segmentation method as an alternative to manual segmentation for radiomics-based diagnosis of clinically significant (CS) prostate cancer (PCa) on biparametric magnetic resonance imaging (bpMRI). MATERIALS AND METHODS: This study included a retrospective multi-center dataset of 524 PCa lesions (of which 204 are CS PCa) on bpMRI. All lesions were both semi-automatically segmented with a DLM auto-fixed VOI method (averaging < 10 s per lesion) and manually segmented by an expert uroradiologist (averaging 5 min per lesion). The DLM auto-fixed VOI method uses a spherical VOI (with its center at the location of the lowest apparent diffusion coefficient of the prostate lesion as indicated with a single mouse click) from which non-prostate voxels are removed using a deep learning-based prostate segmentation algorithm. Thirteen different DLM auto-fixed VOI diameters (ranging from 6 to 30 mm) were explored. Extracted radiomics data were split into training and test sets (4:1 ratio). Performance was assessed with receiver operating characteristic (ROC) analysis. RESULTS: In the test set, the area under the ROC curve (AUCs) of the DLM auto-fixed VOI method with a VOI diameter of 18 mm (0.76 [95% CI: 0.66-0.85]) was significantly higher (p = 0.0198) than that of the manual segmentation method (0.62 [95% CI: 0.52-0.73]). CONCLUSIONS: A DLM auto-fixed VOI segmentation can provide a potentially more accurate radiomics diagnosis of CS PCa than expert manual segmentation while also reducing expert time investment by more than 97%. KEY POINTS: • Compared to traditional expert-based segmentation, a deep learning mask (DLM) auto-fixed VOI placement is more accurate at detecting CS PCa. • Compared to traditional expert-based segmentation, a DLM auto-fixed VOI placement is faster and can result in a 97% time reduction. • Applying deep learning to an auto-fixed VOI radiomics approach can be valuable.