To evaluate the impact of a commercially available deep learning-based reconstruction (DLR) algorithm with varying combinations of DLR noise reduction settings and imaging parameters on quantitative and qualitative image quality, PI-RADS classification and examination time in prostate T2-weighted (T2WI) and diffusion-weighted (DWI) imaging. Forty patients were included. Standard-of-care (SoC) prostate MRI sequences including T2WI and DWI were reconstructed without and with different DLR de-noising levels (low, medium, high). In addition, faster T2WI(Fast) and DWI(Fast) sequences, and a higher resolution T2WI(HR) sequence were evaluated. Quantitative analysis included signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR) and apparent diffusion coefficient (ADC) values. Two radiologists performed qualitative analysis, independently evaluating imaging datasets using 5-point scoring scales for image quality and artifacts. PI-RADS category assignment was also performed by the more experienced radiologist. All DLR levels resulted in significantly higher SNR and CNR compared to the DLR(off) acquisitions. DLR allowed the acquisition time to be reduced by 33% for T2WI(Fast) and 49% for DWI(Fast) compared to SoC, without affecting image quality, whilst T2WI(HR) with DLR allowed for a 73% increase in spatial resolution in the phase encode direction compared to SoC. The inter-reader agreement for image quality and artifact scores was substantial for all subjective measurements on T2WI and DWI. The T2WI(Fast) protocol with DLR(medium) and DWI(Fast) with DLR(low) received the highest qualitative quality score. DLR can reduce T2WI and DWI acquisition time and increase SNR and CNR without compromising image quality or altering PI-RADS classification.

Automated semantic segmentation of multiple knee joint tissues is desirable to allow faster and more reliable analysis of large datasets and to enable further downstream processing e.g. automated diagnosis. In this work, we evaluate the use of conditional Generative Adversarial Networks (cGANs) as a robust and potentially improved method for semantic segmentation compared to other extensively used convolutional neural network, such as the U-Net. As cGANs have not yet been widely explored for semantic medical image segmentation, we analysed the effect of training with different objective functions and discriminator receptive field sizes on the segmentation performance of the cGAN. Additionally, we evaluated the possibility of using transfer learning to improve the segmentation accuracy. The networks were trained on i) the SKI10 dataset which comes from the MICCAI grand challenge "Segmentation of Knee Images 2010″, ii) the OAI ZIB dataset containing femoral and tibial bone and cartilage segmentations of the Osteoarthritis Initiative cohort and iii) a small locally acquired dataset (Advanced MRI of Osteoarthritis (AMROA) study) consisting of 3D fat-saturated spoiled gradient recalled-echo knee MRIs with manual segmentations of the femoral, tibial and patellar bone and cartilage, as well as the cruciate ligaments and selected peri-articular muscles. The Sørensen-Dice Similarity Coefficient (DSC), volumetric overlap error (VOE) and average surface distance (ASD) were calculated for segmentation performance evaluation. DSC ≥ 0.95 were achieved for all segmented bone structures, DSC ≥ 0.83 for cartilage and muscle tissues and DSC of ≈0.66 were achieved for cruciate ligament segmentations with both cGAN and U-Net on the in-house AMROA dataset. Reducing the receptive field size of the cGAN discriminator network improved the networks segmentation performance and resulted in segmentation accuracies equivalent to those of the U-Net. Pretraining not only increased segmentation accuracy of a few knee joint tissues of the fine-tuned dataset, but also increased the network's capacity to preserve segmentation capabilities for the pretrained dataset. cGAN machine learning can generate automated semantic maps of multiple tissues within the knee joint which could increase the accuracy and efficiency for evaluating joint health.

Background: Determining the compositional response of articular cartilage to dynamic joint‐loading using MRI may be a more sensitive assessment of cartilage status than conventional static imaging. However, distinguishing the effects of joint‐loading vs. inherent measurement variability remains difficult, as the repeatability of these quantitative methods is often not assessed or reported.Purpose: To assess exercise‐induced changes in femoral, tibial, and patellar articular cartilage composition and compare these against measurement repeatability.Study Type: Prospective observational study.Population: Phantom and 19 healthy participants.Field Strength/Sequence: 3T; 3D fat‐saturated spoiled gradient recalled‐echo; T_1ρ‐ and T₂‐prepared pseudosteady‐state 3D fast spin echo.Assessment: The intrasessional repeatability of T_1ρ and T₂ relaxation mapping, with and without knee repositioning between two successive measurements, was determined in 10 knees. T_1ρ and T₂ relaxation mapping of nine knees was performed before and at multiple timepoints after a 5‐minute repeated, joint‐loading stepping activity. 3D surface models were created from patellar, femoral, and tibial articular cartilage.Statistical Tests: Repeatability was assessed using root‐mean‐squared‐CV (RMS‐CV). Using Bland–Altman analysis, thresholds defined as the smallest detectable difference (SDD) were determined from the repeatability data with knee repositioning.Results: Without knee repositioning, both surface‐averaged T_1ρ and T₂ were very repeatable on all cartilage surfaces, with RMS‐CV SDD) average exercise‐induced in T_1ρ and T₂ of femoral (–8.0% and –5.3%), lateral tibial (–6.9% and –5.9%), medial tibial (+5.8% and +2.9%), and patellar (–7.9% and +2.8%) cartilage were observed.Data Conclusion: Joint‐loading with a stepping activity resulted in T_1ρ and T₂ changes above background measurement error.Evidence Level: 2Technical Efficacy Stage: 1 J. MAGN. RESON. IMAGING 2020;52:1753–1764.

Under the guidance of high-resolution Magnetic Resonance Imaging (MRI), robotic devices offer a great advantage for prostate intervention. This paper presents an MR-safe robot, where a needle is attached to the needle guide to obtain prostate biopsies during surgeries. The robot is powered by three actuators, two of them are customized to function as a work plane that allows the needle to move horizontally and vertically, and the third actuator controls the rotation of the work plane, allowing the needle to be inserted into the prostate from different directions. All the actuators are pneumatically actuated to allow them to work in a Magnetic Resonance (MR) environment. The kinematics and mechanism of the robot are analyzed. A user interface developed using LabView is created to calculate the target position and generate a control signal for the valves. In the open-air test, the needle can reach the target with an accuracy of 1.3 mm. The signal-to-noise ratio (SNR) variation was measured below 5% under a 3T MR scanner.

Prostate cancer (PCa) is currently the second most prevalent cancer among men. Accurate diagnosis of PCa can provide effective treatment for patients and reduce mortality. Previous works have merely focused on either lesion detection or lesion classification of PCa from magnetic resonance imaging (MRI). In this work we focus on a critical, yet underexplored task of the PCa clinical workflow: distinguishing cases with cancer presence (pathologically confirmed PCa patients) from conditions with no suspicious PCa findings (no cancer presence). To this end, we conduct large-scale experiments for this task for the first time by adopting and processing the multi-centric ProstateNET Imaging Archive which contains more than 6 million image representations of PCa from more than 11,000 PCa cases, representing the largest collection of PCa MR images. Bi-parametric MR (bpMRI) images of 4504 patients alongside their clinical variables are used for training, while the architectures are evaluated on two hold-out test sets of 975 retrospective and 435 prospective patients. Our proposed multi-encoder-cross-attention-fusion architecture achieved a promising area under the receiver operating characteristic curve (AUC) of 0.91. This demonstrates our method’s capability of fusing complex bi-parametric imaging modalities and enhancing model robustness, paving the way towards the clinical adoption of deep learning models for accurately determining the presence of PCa across patient populations.