Optimising the diagnostic accuracy of First post-contrAst SubtracTed breast MRI (FAST MRI) through interpretation-training: a multicentre e-learning study, mapping the learning curve of NHS Breast Screening Programme (NHSBSP) mammogram readers using an enriched dataset.

BACKGROUND: Abbreviated breast MRI (FAST MRI) is being introduced into clinical practice to screen women with mammographically dense breasts or with a personal history of breast cancer. This study aimed to optimise diagnostic accuracy through the adaptation of interpretation-training. METHODS: A FAST MRI interpretation-training programme (short presentations and guided hands-on workstation teaching) was adapted to provide additional training during the assessment task (interpretation of an enriched dataset of 125 FAST MRI scans) by giving readers feedback about the true outcome of each scan immediately after each scan was interpreted (formative assessment). Reader interaction with the FAST MRI scans used developed software (RiViewer) that recorded reader opinions and reading times for each scan. The training programme was additionally adapted for remote e-learning delivery. STUDY DESIGN: Prospective, blinded interpretation of an enriched dataset by multiple readers. RESULTS: 43 mammogram readers completed the training, 22 who interpreted breast MRI in their clinical role (Group 1) and 21 who did not (Group 2). Overall sensitivity was 83% (95%CI 81-84%; 1994/2408), specificity 94% (95%CI 93-94%; 7806/8338), readers' agreement with the true outcome kappa = 0.75 (95%CI 0.74-0.77) and diagnostic odds ratio = 70.67 (95%CI 61.59-81.09). Group 1 readers showed similar sensitivity (84%) to Group 2 (82% p = 0.14), but slightly higher specificity (94% v. 93%, p = 0.001). Concordance with the ground truth increased significantly with the number of FAST MRI scans read through the formative assessment task (p = 0.002) but by differing amounts depending on whether or not a reader had previously attended FAST MRI training (interaction p = 0.02). Concordance with the ground truth was significantly associated with reading batch size (p = 0.02), tending to worsen when more than 50 scans were read per batch. Group 1 took a median of 56 seconds (range 8-47,466) to interpret each FAST MRI scan compared with 78 (14-22,830, p < 0.0001) for Group 2. CONCLUSIONS: Provision of immediate feedback to mammogram readers during the assessment test set reading task increased specificity for FAST MRI interpretation and achieved high diagnostic accuracy. Optimal reading-batch size for FAST MRI was 50 reads per batch. Trial registration (25/09/2019): ISRCTN16624917.

Abbreviated Breast MRI for Supplemental Screening in Patients With Dense Breasts: Comparison of Baseline Versus Subsequent-Round Examinations.

BACKGROUND. Abbreviated breast MRI (AB-MRI) achieves a higher cancer detection rate (CDR) than digital breast tomosynthesis when applied for baseline (i.e., first-round) supplemental screening of individuals with dense breasts. Limited literature has evaluated subsequent (i.e., sequential) AB-MRI screening rounds. OBJECTIVE. This study aimed to compare outcomes between baseline and subsequent rounds of screening AB-MRI in individuals with dense breasts who otherwise had an average risk for breast cancer. METHODS. This retrospective study included patients with dense breasts who otherwise had an average risk for breast cancer and underwent AB-MRI for supplemental screening between December 20, 2016, and May 10, 2023. The clinical interpretations and results of recommended biopsies for AB-MRI examinations were extracted from the EMR. Baseline and subsequent-round AB-MRI examinations were compared. RESULTS. The final sample included 2585 AB-MRI examinations (2007 baseline and 578 subsequent-round examinations) performed for supplemental screening of 2007 women (mean age, 57.1 years old) with dense breasts. Of 2007 baseline examinations, 1658 (82.6%) were assessed as BI-RADS category 1 or 2, 171 (8.5%) as BI-RADS category 3, and 178 (8.9%) as BI-RADS category 4 or 5. Of 578 subsequent-round examinations, 533 (92.2%) were assessed as BI-RADS category 1 or 2, 20 (3.5%) as BI-RADS category 3, and 25 (4.3%) as BI-RADS category 4 or 5 (p < .001). The abnormal interpretation rate (AIR) was 17.4% (349/2007) for baseline examinations versus 7.8% (45/578) for subsequent-round examinations (p < .001). For baseline examinations, PPV2 was 21.3% (38/178), PPV3 was 26.6% (38/143), and the CDR was 18.9 cancers per 1000 examinations (38/2007). For subsequent-round examinations, PPV2 was 28.0% (7/25) (p = .45), PPV3 was 29.2% (7/24) (p = .81), and the CDR was 12.1 cancers per 1000 examinations (7/578) (p = .37). All 45 cancers diagnosed by baseline or subsequent-round AB-MRI were stage 0 or 1. Seven cancers diagnosed by subsequent-round AB-MRI had a mean interval of 872 ± 373 (SD) days since prior AB-MRI and node-negative status at surgical axillary evaluation; six had an invasive component, all measuring 1.2 cm or less. CONCLUSION. Subsequent rounds of AB-MRI screening of individuals with dense breasts had lower AIR than baseline examinations while maintaining a high CDR. All cancers detected by subsequent-round examinations were early-stage node-negative cancers. CLINICAL IMPACT. The findings support sequential AB-MRI for supplemental screening in individuals with dense breasts. Further investigations are warranted to optimize the screening interval.

Implementation of ultrasound and fast magnetic resonance imaging pathway reduces computed tomography utilization in children with suspected appendicitis.

PURPOSE: Computed tomography (CT) is still used in the imaging diagnosis of acute appendicitis in children at many hospitals. We implemented an ultrasound (US) and fast magnetic resonance imaging (MRI) pathway for suspected appendicitis at our institution with the goal of reducing radiation exposure in children. METHODS: All children (< 18 years old) who underwent appendectomy between January 2011 and July 2021 were reviewed. Data were collected on all imaging studies performed. In December 2015, we initiated an imaging pathway for suspected acute appendicitis. US was the initial imaging study, and a rapid protocol MRI was performed if US was equivocal. Those could not tolerate MRI underwent CT. We evaluated the difference in percentage of patients who underwent CT before and after pathway initiation. RESULTS: 554 patients who underwent appendectomy did not have prior imaging studies on presentation to our hospital and were included in analysis. After initiating the pathway, the use of abdominal US increased from 87% (220 of 254) to 97% (291 of 300, p < 0.0001) and the use of MRI increased by 100% (0 MRIs pre-protocol, 90 of 300 patients post-protocol, p < 0.0001). CT utilization decreased significantly from 32% (82 of 254) to 2% (6 of 300, p < 0.0001). CONCLUSION: Embracing a new US and rapid MRI pathway to evaluate pediatric patients with suspected acute appendicitis resulted in significant reduction in CT utilization and therefore radiation exposure.

Evaluating the effectiveness of abbreviated breast MRI (abMRI) interpretation training for mammogram readers: a multi-centre study assessing diagnostic performance, using an enriched dataset.

BACKGROUND: Abbreviated breast MRI (abMRI) is being introduced in breast screening trials and clinical practice, particularly for women with dense breasts. Upscaling abMRI provision requires the workforce of mammogram readers to learn to effectively interpret abMRI. The purpose of this study was to examine the diagnostic accuracy of mammogram readers to interpret abMRI after a single day of standardised small-group training and to compare diagnostic performance of mammogram readers experienced in full-protocol breast MRI (fpMRI) interpretation (Group 1) with that of those without fpMRI interpretation experience (Group 2). METHODS: Mammogram readers were recruited from six NHS Breast Screening Programme sites. Small-group hands-on workstation training was provided, with subsequent prospective, independent, blinded interpretation of an enriched dataset with known outcome. A simplified form of abMRI (first post-contrast subtracted images (FAST MRI), displayed as maximum-intensity projection (MIP) and subtracted slice stack) was used. Per-breast and per-lesion diagnostic accuracy analysis was undertaken, with comparison across groups, and double-reading simulation of a consecutive screening subset. RESULTS: 37 readers (Group 1: 17, Group 2: 20) completed the reading task of 125 scans (250 breasts) (total = 9250 reads). Overall sensitivity was 86% (95% confidence interval (CI) 84-87%; 1776/2072) and specificity 86% (95%CI 85-86%; 6140/7178). Group 1 showed significantly higher sensitivity (843/952; 89%; 95%CI 86-91%) and higher specificity (2957/3298; 90%; 95%CI 89-91%) than Group 2 (sensitivity = 83%; 95%CI 81-85% (933/1120) p < 0.0001; specificity = 82%; 95%CI 81-83% (3183/3880) p < 0.0001). Inter-reader agreement was higher for Group 1 (kappa = 0.73; 95%CI 0.68-0.79) than for Group 2 (kappa = 0.51; 95%CI 0.45-0.56). Specificity improved for Group 2, from the first 55 cases (81%) to the remaining 70 (83%) (p = 0.02) but not for Group 1 (90-89% p = 0.44), whereas sensitivity remained consistent for both Group 1 (88-89%) and Group 2 (83-84%). CONCLUSIONS: Single-day abMRI interpretation training for mammogram readers achieved an overall diagnostic performance within benchmarks published for fpMRI but was insufficient for diagnostic accuracy of mammogram readers new to breast MRI to match that of experienced fpMRI readers. Novice MRI reader performance improved during the reading task, suggesting that additional training could further narrow this performance gap.

NeuroMix-A single-scan brain exam.

PURPOSE: Implement a fast, motion-robust pulse sequence that acquires T1 -weighted, T2 -weighted, T2* -weighted, T2 fluid-attenuated inversion recovery, and DWI data in one run with only one prescription and one prescan. METHODS: A software framework was developed that configures and runs several sequences in one main sequence. Based on that framework, the NeuroMix sequence was implemented, containing motion robust single-shot sequences using EPI and fast spin echo (FSE) readouts (without EPI distortions). Optional multi-shot sequences that provide better contrast, higher resolution, or isotropic resolution could also be run within the NeuroMix sequence. An optimized acquisition order was implemented that minimizes times where no data is acquired. RESULTS: NeuroMix is customizable and takes between 1:20 and 4 min for a full brain scan. A comparison with the predecessor EPIMix revealed significant improvements for T2 -weighted and T2 fluid-attenuated inversion recovery, while taking only 8 s longer for a similar configuration. The optional contrasts were less motion robust but offered a significant increase in quality, detail, and contrast. Initial clinical scans on 1 pediatric and 1 adult patient showed encouraging image quality. CONCLUSION: The single-shot FSE readouts for T2 -weighted and T2 fluid-attenuated inversion recovery and the optional multishot FSE and 3D-EPI contrasts significantly increased diagnostic value compared with EPIMix, allowing NeuroMix to be considered as a standalone brain MRI application.

Fast T2-Weighted Imaging With Deep Learning-Based Reconstruction: Evaluation of Image Quality and Diagnostic Performance in Patients Undergoing Radical Prostatectomy.

BACKGROUND: Deep learning-based reconstruction (DLR) can potentially improve image quality by reduction of noise, thereby enabling fast acquisition of magnetic resonance imaging (MRI). However, a systematic evaluation of image quality and diagnostic performance of MRI using short acquisition time with DLR has rarely been investigated in men with prostate cancer. PURPOSE: To assess the image quality and diagnostic performance of MRI using short acquisition time with DLR for the evaluation of extraprostatic extension (EPE). STUDY TYPE: Retrospective. POPULATION: One hundred and nine men. FIELD STRENGTH/SEQUENCE: 3 T; turbo spin echo T2-weighted images (T2WI), echo-planar diffusion-weighted, and spoiled gradient echo dynamic contrast-enhanced images. ASSESSMENT: To compare image quality, signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) and subjective analysis using Likert scales on three T2WIs (MRI using conventional acquisition time, MRI using short acquisition time [fast MRI], and fast MRI with DLR) were performed. The diagnostic performance for EPE was evaluated by three independent readers. STATISTICAL TESTS: SNR, CNR, and image quality scores across the three imaging protocols were compared using Friedman tests. The diagnostic performance for EPE was assessed using the area under receiver operating characteristic curves (AUCs). P < 0.05 was considered statistically significant. RESULTS: Fast MRI with DLR demonstrated significantly higher SNR (mean ± SD, 14.7 ± 6.8 vs. 8.8 ± 4.9) and CNR (mean ± SD, 6.5 ± 6.3 vs. 3.4 ± 3.6) values and higher image quality scores (median, 4.0 vs. 3.0 for three readers) than fast MRI. The AUCs for EPE were significantly higher with the use of DLR (0.86 vs. 0.75 for reader 2 and 0.82 vs. 0.73 for reader 3) compared with fast MRI, whereas differences were not significant for reader 1 (0.81 vs. 0.74; P = 0.09). DATA CONCLUSION: DLR may be useful in reducing the acquisition time of prostate MRI without compromising image quality or diagnostic performance. LEVEL OF EVIDENCE: 4 TECHNICAL EFFICACY: Stage 3.

Edge-enhanced dual discriminator generative adversarial network for fast MRI with parallel imaging using multi-view information.

In clinical medicine, magnetic resonance imaging (MRI) is one of the most important tools for diagnosis, triage, prognosis, and treatment planning. However, MRI suffers from an inherent slow data acquisition process because data is collected sequentially in k-space. In recent years, most MRI reconstruction methods proposed in the literature focus on holistic image reconstruction rather than enhancing the edge information. This work steps aside this general trend by elaborating on the enhancement of edge information. Specifically, we introduce a novel parallel imaging coupled dual discriminator generative adversarial network (PIDD-GAN) for fast multi-channel MRI reconstruction by incorporating multi-view information. The dual discriminator design aims to improve the edge information in MRI reconstruction. One discriminator is used for holistic image reconstruction, whereas the other one is responsible for enhancing edge information. An improved U-Net with local and global residual learning is proposed for the generator. Frequency channel attention blocks (FCA Blocks) are embedded in the generator for incorporating attention mechanisms. Content loss is introduced to train the generator for better reconstruction quality. We performed comprehensive experiments on Calgary-Campinas public brain MR dataset and compared our method with state-of-the-art MRI reconstruction methods. Ablation studies of residual learning were conducted on the MICCAI13 dataset to validate the proposed modules. Results show that our PIDD-GAN provides high-quality reconstructed MR images, with well-preserved edge information. The time of single-image reconstruction is below 5ms, which meets the demand of faster processing.

Visualization of Human Aortic Valve Dynamics Using Magnetic Resonance Imaging with Sub-Millisecond Temporal Resolution.

BACKGROUND: Visualization of aortic valve dynamics is important in diagnosing valvular diseases but is challenging to perform with magnetic resonance imaging (MRI) due to the limited temporal resolution. PURPOSE: To develop an MRI technique with sub-millisecond temporal resolution and demonstrate its application in visualizing rapid aortic valve opening and closing in human subjects in comparison with echocardiography and conventional MRI techniques. STUDY TYPE: Prospective. POPULATION: Twelve healthy subjects. FIELD STRENGTH/SEQUENCE: 3 T; gradient-echo-train-based sub-millisecond periodic event encoded imaging (get-SPEEDI) and balanced steady-state free precession (bSSFP). ASSESSMENT: Images were acquired using get-SPEEDI with a temporal resolution of 0.6 msec. get-SPEEDI was triggered by an electrocardiogram so that each echo in the gradient echo train corresponded to an image at a specific time point, providing a time-resolved characterization of aortic valve dynamics. For comparison, bSSFP was also employed with 12 msec and 24 msec temporal resolutions, respectively. The durations of the aortic valve rapid opening (Tro ), rapid closing (Trc ), and the maximal aortic valve area (AVA) normalized to height were measured with all three temporal resolutions. M-mode echocardiograms with a temporal resolution of 0.8 msec were obtained for further comparison. STATISTICAL TEST: Parameters were compared between the three sequences, together with the echocardiography results, with a Mann-Whitney U test. RESULTS: Significantly shorter Tro (mean ± SD: 27.5 ± 6.7 msec) and Trc (43.8 ± 11.6 msec) and larger maximal AVA/height (2.01 ± 0.29 cm2 /m) were measured with get-SPEEDI compared to either bSSFP sequence (Tro of 56.3 ± 18.8 and 63.8 ± 20.2 msec; Trc of 68.2 ± 16.6 and 72.8 ± 18.2 msec; maximal AVA/height of 1.63 ± 0.28 and 1.65 ± 0.32 cm2 /m for 12 msec and 24 msec temporal resolutions, respectively, P < 0.05). In addition, the get-SPEEDI results were more consistent with those measured using echocardiography, especially for Tro (29.0 ± 4.1 msec, P = 0.79) and Trc (41.6 ± 4.3 msec, P = 0.16). DATA CONCLUSION: get-SPEEDI allows for visualization of human aortic valve dynamics and provided values closer to those measured using echocardiography than the bSSFP sequences. LEVEL OF EVIDENCE: 1 TECHNICAL EFFICACY STAGE: 1.

Rapid Musculoskeletal MRI in 2021: Value and Optimized Use of Widely Accessible Techniques.

OBJECTIVE. The purpose of this article is to provide a practice-focused review of accelerating musculoskeletal MRI with the use of widely accessible techniques and to assess the effects of such acceleration on the value of musculoskeletal MRI. CONCLUSION. Echo-train compaction with fast radiofrequency pulses, high gradient performance modes, and high receiver bandwidth, as well as basic phase undersampling techniques, affords at least twofold acceleration of musculoskeletal MRI examinations while retaining image quality, comprehensiveness, and diagnostic performance. Optimized efficiency is a cornerstone for adding value to musculoskeletal MRI.

Success of Nonsedated Neuroradiologic MRI in Children 1-7 Years Old.

BACKGROUND. MRI use and the need for monitored anesthesia care (MAC) in children have increased. However, MAC is associated with examination delays, increased cost, and safety concerns. OBJECTIVE. The purpose of this study was to evaluate the success rate of nonsedated neuroradiologic MRI studies in children 1-7 years old and to investigate factors associated with success. METHODS. We retrospectively reviewed data from our institutional nonsedated MRI program. Inclusion criteria were outpatient nonsedated MRI referral, age 1-7 years old, and neuroradiologic indication. Exclusion criteria were MRI examinations for ventricular checks and contrast material use. Success was determined by reviewing the clinical MRI report. We recorded patient age and sex, type of MRI examination (brain, spine, craniospinal, head and neck, and brain with MRA), protocol length, presence of child life specialist, video goggle use, and MRI appointment time (routine daytime appointment or evening appointment). We used descriptive statistics to summarize patient demographics and clinical data and logistic regression models to evaluate predictors of success in the entire sample. Subset analyses were performed for children from 1 to < 3 years old and 3 to 7 years old. RESULTS. We analyzed 217 patients who underwent nonsedated MRI examinations (median age, 5.1 years). Overall success rate was 82.0% (n = 178). The success rates were 81.4% (n = 127) for brain, 90.3% (n = 28) for spine, 71.4% (n = 10) for craniospinal, 66.7% (n = 6) for head and neck, and 100% (n = 7) for brain with MRA. Age was significantly associated with success (odds ratio [OR], 1.33; p = .009). In children 1 to < 3 years old, none of the factors analyzed were significant predictors of success (all, p > .48). In children 3-7 years old, protocol duration (OR, 0.96; 95% CI, 0.93-0.99; p = .02) and video goggle use (OR, 6.38; 95% CI, 2.16-18.84; p = .001) were significantly associated with success. CONCLUSION. A multidisciplinary approach with age-appropriate resources enables a high success rate for nonsedated neuroradiologic MRI in children 1-7 years old. CLINICAL IMPACT. Using age as the primary criterion to determine the need for MAC may lead to overuse of these services. Dissemination of information regarding nonsedated MRI practice could reduce the rate of sedated MRI in young children.

Concurrent electrophysiological and hemodynamic measurements of evoked neural oscillations in human visual cortex using sparsely interleaved fast fMRI and EEG.

Electroencephalography (EEG) concurrently collected with functional magnetic resonance imaging (fMRI) is heavily distorted by the repetitive gradient coil switching during the fMRI acquisition. The performance of the typical template-based gradient artifact suppression method can be suboptimal because the artifact changes over time. Gradient artifact residuals also impede the subsequent suppression of ballistocardiography artifacts. Here we propose recording continuous EEG with temporally sparse fast fMRI (fast fMRI-EEG) to minimize the EEG artifacts caused by MRI gradient coil switching without significantly compromising the field-of-view and spatiotemporal resolution of fMRI. Using simultaneous multi-slice inverse imaging to achieve whole-brain fMRI with isotropic 5-mm resolution in 0.1 â€‹s, and performing these acquisitions once every 2 â€‹s, we have 95% of the duty cycle available to record EEG with substantially less gradient artifact. We found that the standard deviation of EEG signals over the entire acquisition period in fast fMRI-EEG was reduced to 54% of that in conventional concurrent echo-planar imaging (EPI) and EEG recordings (EPI-EEG) across participants. When measuring 15-Hz steady-state visual evoked potentials (SSVEPs), the baseline-normalized oscillatory neural response in fast fMRI-EEG was 2.5-fold of that in EPI-EEG. The functional MRI responses associated with the SSVEP delineated by EPI and fast fMRI were similar in the spatial distribution, the elicited waveform, and detection power. Sparsely interleaved fast fMRI-EEG provides high-quality EEG without substantially compromising the quality of fMRI in evoked response measurements, and has the potential utility for applications where the onset of the target stimulus cannot be precisely determined, such as epilepsy.

How Fast Can We Go: Abbreviated Prostate MR Protocols.

PURPOSE OF REVIEW: Multiparametric MRI (mpMRI), composed of T2WI, DWI, and DCE sequences, is effective in identifying prostate cancer (PCa), but length and cost preclude its application as a PCa screening tool. Here we review abbreviated MRI protocols that shorten or omit conventional mpMRI components to reduce scan time and expense without forgoing diagnostic accuracy. RECENT FINDINGS: The DCE sequence, which plays a limited diagnostic role in PI-RADS, is eliminated in variations of the biparametric MRI (bpMRI). T2WI, the lengthiest sequence, is truncated by only acquiring the axial plane or utilizing 3D acquisition with subsequent 2D reconstruction. DW-EPISMS further accelerates DWI acquisition. The fastest protocol described to date consists of just DW-EPISMS and axial-only 2D T2WI and runs less than 5 min. Abbreviated protocols can mitigate scan expense and increase scan access, allowing prostate MRI to become an efficient PCa screening tool.

Rapid brain MRI protocols reduce head computerized tomography use in the pediatric emergency department.

BACKGROUND: Rapid magnetic resonance imaging (MRI) protocols may be effective in the emergency department (ED) to evaluate nontraumatic neurologic complaints. We evaluate neuroimaging (rapid MRI [rMRI]), head computerized tomography [HCT], and full MRI) use following widespread implementation of rMRI protocols in a pediatric emergency department (ED). METHODS: We conducted a retrospective study in a tertiary care pediatric ED of encounters with neuroimaging during two 9-month periods: one prior to (control period) and one after generalized availability of 4 rMRI protocols (rMRI period). The primary outcome was differences in neuroimaging rates between the two periods. Secondary outcomes included ED process measures, unsuccessful imaging, and undetected pathology, with full MRI within 14 days as the reference standard. RESULTS: There were 1052 encounters with neuroimaging during the control and 1308 during the rMRI periods. Differences in neuroimaging between periods were 27.7% for rMRI (95% CI, 24.4, 31.0), - 21.5% for HCT (95% CI, - 25.5, - 17.5), and - 6.2% for full MRI (95% CI, - 9.3, - 3.1%.) Time to imaging (182 [IQR 138-255] versus 86 [IQR 52-137] minutes) as well as ED length of stay (396 [IQR 304-484] versus 257 [IQR 196-334] minutes) was longer for rMRI versus HCT (p < 0.01). Between the control and rMRI periods, there were differences in types of neuroimaging performed for patients with altered mental status, headache, seizure, shunt dysfunction, stroke, syncope, trauma, vomiting, infection, and other neurologic complaints (p < 0.05). rMRI studies were unsuccessful in 3.6% of studies versus 0.0% of HCTs (p < 0.01). The 22 unsuccessful rMRI studies were unsuccessful due to artifacts from dental hardware (n = 2) and patient motion (n = 20). None of the rMRI studies with full MRI follow-up imaging had undetected pathology; the false negative rate for the HCT exams was as high as 25%. CONCLUSIONS: After routine ED use of 4 rMRI protocols, there was a more than 20% decrease in HCT use without missed diagnoses. Time to neuroimaging and length of stay were longer for rMRI than HCT, with higher rates of unsuccessful imaging. Despite these limitations, rMRI may be an alternative to HCT for nontraumatic complaints in the ED.

Artificial Intelligence in Musculoskeletal Imaging: Current Status and Future Directions.

OBJECTIVE. The objective of this article is to show how artificial intelligence (AI) has impacted different components of the imaging value chain thus far as well as to describe its potential future uses. CONCLUSION. The use of AI has the potential to greatly enhance every component of the imaging value chain. From assessing the appropriateness of imaging orders to helping predict patients at risk for fracture, AI can increase the value that musculoskeletal imagers provide to their patients and to referring clinicians by improving image quality, patient centricity, imaging efficiency, and diagnostic accuracy.

Potential role of abbreviated MRI for breast cancer screening in an academic medical center.

The purpose is to determine whether an abbreviated MRI protocol (ABMR) is ready to be used for breast cancer screening in an academic practice setting. Two hundred and fifty nine breast MRIs from 1/1/2012 to 6/30/2012 were retrospectively reviewed using ABMR (MIP, Pre-contrastT1, single dynamic post-contrastT1, and subtraction). Five breast radiologists (4-28 year-expr) participated in this reader study performed in two phases: Phase1 - radiologist's privy to clinical history but not to comparison imaging. Phase2 - radiologists provided comparison imaging. For phase1, studies were reviewed using three steps: (a) MIP only (positive/negative/intermediate); (b) ABMR (recall/no recall) and (c) With T2 (for changes in recommendations). Radiologist also recorded total time for interpretation. In Phase2 the MRIs coded as "recall" were re-reviewed with available comparison studies, noting changes in final recommendation. The abnormal interpretation rates (AIRs) were calculated for phase1 and phase2 results with comparison to the original full protocol. Of the 259 patients (avg. age-52 years; range 26-78), there were seven cancers (three invasive, three DCIS and one breast lymphoma). Acquisition time for ABMR was 3 minutes, ABMR + T2-8 minutes, and original full protocol 16 minutes. Average MIP was positive or indeterminate in 86% (6/7) and negative in 14% (1/7) cancers. The average AIR for MIP only was 20.8% (sens-77.1%; spec-80.8%. The AIR w/o comparisons was 25.6% (sens-91.4%; spec- 76.2%); however the average AIR decreased in phase 2 with comparisons to 13.7% (sens-91.4%; spec-88.5%). The AIR of the original full protocol read was 16.2% (sens-100%; spec-85.7%). Addition of T2 changed assessment in only 3% (1.2%-6.5%). Avg. read time for ABMR including T2 was 2.5 minutes (1.6-4.0 minutes). ABMR is reliable for breast cancer screening, with acceptable interpretation time and acceptable AIR.

Deep Learning-based Image Enhancement Techniques for Fast MRI in Neuroimaging.

Despite its superior soft tissue contrast and non-invasive nature, MRI requires long scan times due to its intrinsic signal acquisition principles, a main drawback which technological advancements in MRI have been focused on. In particular, scan time reduction is a natural requirement in neuroimaging due to detailed structures requiring high resolution imaging and often volumetric (3D) acquisitions, and numerous studies have recently attempted to harness deep learning (DL) technology in enabling scan time reduction and image quality improvement. Various DL-based image reconstruction products allow for additional scan time reduction on top of existing accelerated acquisition methods without compromising the image quality.

MLMFNet: A multi-level modality fusion network for multi-modal accelerated MRI reconstruction.

Magnetic resonance imaging produces detailed anatomical and physiological images of the human body that can be used in the clinical diagnosis and treatment of diseases. However, MRI suffers its comparatively longer acquisition time than other imaging methods and is thus vulnerable to motion artifacts, which ultimately lead to likely failed or even wrong diagnosis. In order to perform faster reconstruction, deep learning-based methods along with traditional strategies such as parallel imaging and compressed sensing come into play in recent years in this field. Meanwhile, in order to better analyze the diseases, it is also often necessary to acquire images in the same region of interest under different modalities, which yield images with different contrast levels. However, most of these aforementioned methods tend to use single-modal images for reconstruction, neglecting the correlation and redundancy information embedded in MR images acquired with different modalities. While there are works on multi-modal reconstruction, the information is yet to be efficiently explored. In this paper, we propose an end-to-end neural network called MLMFNet, which helps the reconstruction of the target modality by using information from the auxiliary modality across feature channels and layers. Specifically, this is highlighted by three components: (I) An encoder based on UNet with a single-stream strategy that fuses auxiliary and target modalities; (II) a decoder that tends to multi-level features from all layers of the encoder, and (III) a channel attention module. Quantitative and qualitative analyses are performed on a public brain dataset and knee brain dataset, which show that the proposed method achieves satisfying results in MRI reconstruction within the multi-modal context, and also demonstrate its effectiveness and potential to be used in clinical practice.

Fast MRI Reconstruction Using Deep Learning-based Compressed Sensing: A Systematic Review.

Magnetic resonance imaging (MRI) has revolutionized medical imaging, providing a non-invasive and highly detailed look into the human body. However, the long acquisition times of MRI present challenges, causing patient discomfort, motion artifacts, and limiting real-time applications. To address these challenges, researchers are exploring various techniques to reduce acquisition time and improve the overall efficiency of MRI. One such technique is compressed sensing (CS), which reduces data acquisition by leveraging image sparsity in transformed spaces. In recent years, deep learning (DL) has been integrated with CS-MRI, leading to a new framework that has seen remarkable growth. DL-based CS-MRI approaches are proving to be highly effective in accelerating MR imaging without compromising image quality. This review comprehensively examines DL-based CS-MRI techniques, focusing on their role in increasing MR imaging speed. We provide a detailed analysis of each category of DL-based CS-MRI including end-to-end, unroll optimization, self-supervised, and federated learning. Our systematic review highlights significant contributions and underscores the exciting potential of DL in CS-MRI. Additionally, our systematic review efficiently summarizes key results and trends in DL-based CS-MRI including quantitative metrics, the dataset used, acceleration factors, and the progress of and research interest in DL techniques over time. Finally, we discuss potential future directions and the importance of DL-based CS-MRI in the advancement of medical imaging. To facilitate further research in this area, we provide a GitHub repository that includes up-to-date DL-based CS-MRI publications and publicly available datasets - https://github.com/mosaf/Awesome-DL-based-CS-MRI.

Dual-domain faster Fourier convolution based network for MR image reconstruction.

Deep learning methods for fast MRI have shown promise in reconstructing high-quality images from undersampled multi-coil k-space data, leading to reduced scan duration. However, existing methods encounter challenges related to limited receptive fields in dual-domain (k-space and image domains) reconstruction networks, rigid data consistency operations, and suboptimal refinement structures, which collectively restrict overall reconstruction performance. This study introduces a comprehensive framework that addresses these challenges and enhances MR image reconstruction quality. Firstly, we propose Faster Inverse Fourier Convolution (FasterIFC), a frequency domain convolutional operator that significantly expands the receptive field of k-space domain reconstruction networks. Expanding the information extraction range to the entire frequency spectrum according to the spectral convolution theorem in Fourier theory enables the network to easily utilize richer redundant long-range information from adjacent, symmetrical, and diagonal locations of multi-coil k-space data. Secondly, we introduce a novel softer Data Consistency (softerDC) layer, which achieves an enhanced balance between data consistency and smoothness. This layer facilitates the implementation of diverse data consistency strategies across distinct frequency positions, addressing the inflexibility observed in current methods. Finally, we present the Dual-Domain Faster Fourier Convolution Based Network (D2F2), which features a centrosymmetric dual-domain parallel structure based on FasterIFC. This architecture optimally leverages dual-domain data characteristics while substantially expanding the receptive field in both domains. Coupled with the softerDC layer, D2F2 demonstrates superior performance on the NYU fastMRI dataset at multiple acceleration factors, surpassing state-of-the-art methods in both quantitative and qualitative evaluations.

Five-minute knee MRI: An AI-based super resolution reconstruction approach for compressed sensing. A validation study on healthy volunteers.

PURPOSE: To investigate the potential of combining Compressed Sensing (CS) and a newly developed AI-based super resolution reconstruction prototype consisting of a series of convolutional neural networks (CNN) for a complete five-minute 2D knee MRI protocol. METHODS: In this prospective study, 20 volunteers were examined using a 3T-MRI-scanner (Ingenia Elition X, Philips). Similar to clinical practice, the protocol consists of a fat-saturated 2D-proton-density-sequence in coronal, sagittal and transversal orientation as well as a sagittal T1-weighted sequence. The sequences were acquired with two different resolutions (standard and low resolution) and the raw data reconstructed with two different reconstruction algorithms: a conventional Compressed SENSE (CS) and a new CNN-based algorithm for denoising and subsequently to interpolate and therewith increase the sharpness of the image (CS-SuperRes). Subjective image quality was evaluated by two blinded radiologists reviewing 8 criteria on a 5-point Likert scale and signal-to-noise ratio calculated as an objective parameter. RESULTS: The protocol reconstructed with CS-SuperRes received higher ratings than the time-equivalent CS reconstructions, statistically significant especially for low resolution acquisitions (e.g., overall image impression: 4.3 ±â€¯0.4 vs. 3.4 ±â€¯0.4, p < 0.05). CS-SuperRes reconstructions for the low resolution acquisition were comparable to traditional CS reconstructions with standard resolution for all parameters, achieving a scan time reduction from 11:01 min to 4:46 min (57 %) for the complete protocol (e.g. overall image impression: 4.3 ±â€¯0.4 vs. 4.0 ±â€¯0.5, p < 0.05). CONCLUSION: The newly-developed AI-based reconstruction algorithm CS-SuperRes allows to reduce scan time by 57% while maintaining unchanged image quality compared to the conventional CS reconstruction.