Image quality and metal artifact reduction in total hip arthroplasty CT: deep learning-based algorithm versus virtual monoenergetic imaging and orthopedic metal artifact reduction.

BACKGROUND: To compare image quality, metal artifacts, and diagnostic confidence of conventional computed tomography (CT) images of unilateral total hip arthroplasty patients (THA) with deep learning-based metal artifact reduction (DL-MAR) to conventional CT and 130-keV monoenergetic images with and without orthopedic metal artifact reduction (O-MAR). METHODS: Conventional CT and 130-keV monoenergetic images with and without O-MAR and DL-MAR images of 28 unilateral THA patients were reconstructed. Image quality, metal artifacts, and diagnostic confidence in bone, pelvic organs, and soft tissue adjacent to the prosthesis were jointly scored by two experienced musculoskeletal radiologists. Contrast-to-noise ratios (CNR) between bladder and fat and muscle and fat were measured. Wilcoxon signed-rank tests with Holm-Bonferroni correction were used. RESULTS: Significantly higher image quality, higher diagnostic confidence, and less severe metal artifacts were observed on DL-MAR and images with O-MAR compared to images without O-MAR (p < 0.001 for all comparisons). Higher image quality, higher diagnostic confidence for bone and soft tissue adjacent to the prosthesis, and less severe metal artifacts were observed on DL-MAR when compared to conventional images and 130-keV monoenergetic images with O-MAR (p ≤ 0.014). CNRs were higher for DL-MAR and images with O-MAR compared to images without O-MAR (p < 0.001). Higher CNRs were observed on DL-MAR images compared to conventional images and 130-keV monoenergetic images with O-MAR (p ≤ 0.010). CONCLUSIONS: DL-MAR showed higher image quality, diagnostic confidence, and superior metal artifact reduction compared to conventional CT images and 130-keV monoenergetic images with and without O-MAR in unilateral THA patients. RELEVANCE STATEMENT: DL-MAR resulted into improved image quality, stronger reduction of metal artifacts, and improved diagnostic confidence compared to conventional and virtual monoenergetic images with and without metal artifact reduction, bringing DL-based metal artifact reduction closer to clinical application. KEY POINTS: • Metal artifacts introduced by total hip arthroplasty hamper radiologic assessment on CT. • A deep-learning algorithm (DL-MAR) was compared to dual-layer CT images with O-MAR. • DL-MAR showed best image quality and diagnostic confidence. • Highest contrast-to-noise ratios were observed on the DL-MAR images.

External validation of the RSNA 2020 pulmonary embolism detection challenge winning deep learning algorithm.

PURPOSE: To evaluate the diagnostic performance and generalizability of the winning DL algorithm of the RSNA 2020 PE detection challenge to a local population using CTPA data from two hospitals. MATERIALS AND METHODS: Consecutive CTPA images from patients referred for suspected PE were retrospectively analysed. The winning RSNA 2020 DL algorithm was retrained on the RSNA-STR Pulmonary Embolism CT (RSPECT) dataset. The algorithm was tested in hospital A on multidetector CT (MDCT) images of 238 patients and in hospital B on spectral detector CT (SDCT) and virtual monochromatic images (VMI) of 114 patients. The output of the DL algorithm was compared with a reference standard, which included a consensus reading by at least two experienced cardiothoracic radiologists for both hospitals. Areas under the receiver operating characteristic curve (AUCs) were calculated. Sensitivity and specificity were determined using the maximum Youden index. RESULTS: According to the reference standard, PE was present in 73 patients (30.7%) in hospital A and 33 patients (29.0%) in hospital B. For the DL algorithm the AUC was 0.96 (95% CI 0.92-0.98) in hospital A, 0.89 (95% CI 0.81-0.94) for conventional reconstruction in hospital B and 0.87 (95% CI 0.80-0.93) for VMI. CONCLUSION: The RSNA 2020 pulmonary embolism detection on CTPA challenge winning DL algorithm, retrained on the RSPECT dataset, showed high diagnostic accuracy on MDCT images. A somewhat lower performance was observed on SDCT images, which suggest additional training on novel CT technology may improve generalizability of this DL algorithm.

Advances in metal artifact reduction in CT images: A review of traditional and novel metal artifact reduction techniques.

Metal artifacts degrade CT image quality, hampering clinical assessment. Numerous metal artifact reduction methods are available to improve the image quality of CT images with metal implants. In this review, an overview of traditional methods is provided including the modification of acquisition and reconstruction parameters, projection-based metal artifact reduction techniques (MAR), dual energy CT (DECT) and the combination of these techniques. Furthermore, the additional value and challenges of novel metal artifact reduction techniques that have been introduced over the past years are discussed such as photon counting CT (PCCT) and deep learning based metal artifact reduction techniques.

Is AI the way forward for reducing metal artifacts in CT? Development of a generic deep learning-based method and initial evaluation in patients with sacroiliac joint implants.

PURPOSE: To develop a deep learning-based metal artifact reduction technique (dl-MAR) and quantitatively compare metal artifacts on dl-MAR-corrected CT-images, orthopedic metal artifact reduction (O-MAR)-corrected CT-images and uncorrected CT-images after sacroiliac (SI) joint fusion. METHODS: dl-MAR was trained on CT-images with simulated metal artifacts. Pre-surgery CT-images and uncorrected, O-MAR-corrected and dl-MAR-corrected post-surgery CT-images of twenty-five patients undergoing SI joint fusion were retrospectively obtained. Image registration was applied to align pre-surgery with post-surgery CT-images within each patient, allowing placement of regions of interest (ROIs) on the same anatomical locations. Six ROIs were placed on the metal implant and the contralateral side in bone lateral of the SI joint, the gluteus medius muscle and the iliacus muscle. Metal artifacts were quantified as the difference in Hounsfield units (HU) between pre- and post-surgery CT-values within the ROIs on the uncorrected, O-MAR-corrected and dl-MAR-corrected images. Noise was quantified as standard deviation in HU within the ROIs. Metal artifacts and noise in the post-surgery CT-images were compared using linear multilevel regression models. RESULTS: Metal artifacts were significantly reduced by O-MAR and dl-MAR in bone (p < 0.001), contralateral bone (O-MAR: p = 0.009; dl-MAR: p < 0.001), gluteus medius (p < 0.001), contralateral gluteus medius (p < 0.001), iliacus (p < 0.001) and contralateral iliacus (O-MAR: p = 0.024; dl-MAR: p < 0.001) compared to uncorrected images. Images corrected with dl-MAR resulted in stronger artifact reduction than images corrected with O-MAR in contralateral bone (p < 0.001), gluteus medius (p = 0.006), contralateral gluteus medius (p < 0.001), iliacus (p = 0.017), and contralateral iliacus (p < 0.001). Noise was reduced by O-MAR in bone (p = 0.009) and gluteus medius (p < 0.001) while noise was reduced by dl-MAR in all ROIs (p < 0.001) in comparison to uncorrected images. CONCLUSION: dl-MAR showed superior metal artifact reduction compared to O-MAR in CT-images with SI joint fusion implants.

Diffusion-weighted MRI with deep learning for visualizing treatment results of MR-guided HIFU ablation of uterine fibroids.

OBJECTIVES: No method is available to determine the non-perfused volume (NPV) repeatedly during magnetic resonance-guided high-intensity focused ultrasound (MR-HIFU) ablations of uterine fibroids, as repeated acquisition of contrast-enhanced T1-weighted (CE-T1w) scans is inhibited by safety concerns. The objective of this study was to develop and test a deep learning-based method for translation of diffusion-weighted imaging (DWI) into synthetic CE-T1w scans, for monitoring MR-HIFU treatment progression. METHODS: The algorithm was retrospectively trained and validated on data from 33 and 20 patients respectively who underwent an MR-HIFU treatment of uterine fibroids between June 2017 and January 2019. Postablation synthetic CE-T1w images were generated by a deep learning network trained on paired DWI and reference CE-T1w scans acquired during the treatment procedure. Quantitative analysis included calculation of the Dice coefficient of NPVs delineated on synthetic and reference CE-T1w scans. Four MR-HIFU radiologists assessed the outcome of MR-HIFU treatments and NPV ratio based on the synthetic and reference CE-T1w scans. RESULTS: Dice coefficient of NPVs was 71% (± 22%). The mean difference in NPV ratio was 1.4% (± 22%) and not statistically significant (p = 0.79). Absolute agreement of the radiologists on technical treatment success on synthetic and reference CE-T1w scans was 83%. NPV ratio estimations on synthetic and reference CE-T1w scans were not significantly different (p = 0.27). CONCLUSIONS: Deep learning-based synthetic CE-T1w scans derived from intraprocedural DWI allow gadolinium-free visualization of the predicted NPV, and can potentially be used for repeated gadolinium-free monitoring of treatment progression during MR-HIFU therapy for uterine fibroids. KEY POINTS: • Synthetic CE-T1w scans can be derived from diffusion-weighted imaging using deep learning. • Synthetic CE-T1w scans may be used for visualization of the NPV without using a contrast agent directly after MR-HIFU ablations of uterine fibroids.

Bone visualization of the cervical spine with deep learning-based synthetic CT compared to conventional CT: A single-center noninferiority study on image quality.

PURPOSE: To investigate whether the image quality of a specific deep learning-based synthetic CT (sCT) of the cervical spine is noninferior to conventional CT. METHOD: Paired MRI and CT data were collected from 25 consecutive participants (≥ 50 years) with cervical radiculopathy. The MRI exam included a T1-weighted multiple gradient echo sequence for sCT reconstruction. For qualitative image assessment, four structures at two vertebral levels were evaluated on sCT and compared with CT by three assessors using a four-point scale (range 1-4). The noninferiority margin was set at 0.5 point on this scale. Additionally, acceptable image quality was defined as a score of 3-4 in ≥ 80% of the scans. Quantitative assessment included geometrical analysis and voxelwise comparisons. RESULTS: Qualitative image assessment showed that sCT was noninferior to CT for overall bone image quality, artifacts, imaging of intervertebral joints and neural foramina at levels C3-C4 and C6-C7, and cortical delineation at C6-C7. Noninferiority was weak to absent for cortical delineation at level C3-C4 and trabecular bone at both levels. Acceptable image quality was achieved for all structures in sCT and CT, except for trabecular bone in sCT and level C6-C7 in CT. Geometrical analysis of the sCT showed good to excellent agreement with CT. Voxelwise comparisons showed a mean absolute error of 80.05 (±6.12) HU, dice similarity coefficient (cortical bone) of 0.84 (±0.04) and structural similarity index of 0.86 (±0.02). CONCLUSIONS: This deep learning-based sCT was noninferior to conventional CT for the general visualization of bony structures of the cervical spine, artifacts, and most detailed structure assessments.

Improved neonatal brain MRI segmentation by interpolation of motion corrupted slices.

BACKGROUND AND PURPOSE: To apply and evaluate an intensity-based interpolation technique, enabling segmentation of motion-affected neonatal brain MRI. METHODS: Moderate-late preterm infants were enrolled in a prospective cohort study (Brain Imaging in Moderate-late Preterm infants "BIMP-study") between August 2017 and November 2019. T2-weighted MRI was performed around term equivalent age on a 3T MRI. Scans without motion (n = 27 [24%], control group) and with moderate-severe motion (n = 33 [29%]) were included. Motion-affected slices were re-estimated using intensity-based shape-preserving cubic spline interpolation, and automatically segmented in eight structures. Quality of interpolation and segmentation was visually assessed for errors after interpolation. Reliability was tested using interpolated control group scans (18/54 axial slices). Structural similarity index (SSIM) was used to compare T2-weighted scans, and Sørensen-Dice was used to compare segmentation before and after interpolation. Finally, volumes of brain structures of the control group were used assessing sensitivity (absolute mean fraction difference) and bias (confidence interval of mean difference). RESULTS: Visually, segmentation of 25 scans (22%) with motion artifacts improved with interpolation, while segmentation of eight scans (7%) with adjacent motion-affected slices did not improve. Average SSIM was .895 and Sørensen-Dice coefficients ranged between .87 and .97. Absolute mean fraction difference was ≤0.17 for less than or equal to five interpolated slices. Confidence intervals revealed a small bias for cortical gray matter (0.14-3.07 cm3 ), cerebrospinal fluid (0.39-1.65 cm3 ), deep gray matter (0.74-1.01 cm3 ), and brainstem volumes (0.07-0.28 cm3 ) and a negative bias in white matter volumes (-4.47 to -1.65 cm3 ). CONCLUSION: According to qualitative and quantitative assessment, intensity-based interpolation reduced the percentage of discarded scans from 29% to 7%.

Quantitative analysis of metal artifact reduction in total hip arthroplasty using virtual monochromatic imaging and orthopedic metal artifact reduction, a phantom study.

OBJECTIVE: To quantify metal artifact reduction using 130 keV virtual monochromatic imaging (VMI) with and without orthopedic metal artifact reduction (O-MAR) in total hip arthroplasty. METHODS: Conventional polychromatic images and 130 keV VMI of a phantom with pellets representing bone with unilateral or bilateral prostheses were reconstructed with and without O-MAR on a dual-layer CT. Pellets were categorized as unaffected, mildly affected and severely affected. RESULTS: When 130 keV VMI with O-MAR was compared to conventional imaging with O-MAR, a relative metal artifact reduction in CT values, contrast-to-noise (CNR), signal-to-noise (SNR) and noise in mildly affected pellets (67%, 74%, 48%, 68%, respectively; p < 0.05) was observed but no significant relative metal artifact reduction in severely affected pellets. Comparison between 130 keV VMI without O-MAR and conventional imaging with O-MAR showed relative metal artifact reduction in CT values, CNR, SNR and noise in mildly affected pellets (92%, 72%, 38%, 51%, respectively; p < 0.05) but negative relative metal artifact reduction in CT values and noise in severely affected pellets (- 331% and -223%, respectively; p < 0.05), indicating aggravation of metal artifacts. CONCLUSION: Overall, VMI of 130 keV with O-MAR provided the strongest metal artifact reduction.

Mild brain lesions do not affect brain volumes in moderate-late preterm infants.

PURPOSE: It is unknown whether frequently occurring mild brain lesions affect brain volumes in moderate (MP2; 32+0-33+6 weeks' gestation) and late (LP3; 34+0-35+6 weeks' gestation) preterm infants. Therefore, we aimed to investigate the effect of mild brain lesions on brain volumes in moderate-late preterm (MLPT4) infants and to compare brain volumes between MP and LP infants. METHODS: From August 2017 to November 2019, eligible MLPT infants born at Isala Women and Children's Hospital were enrolled in a prospective cohort study (Brain Imaging in Moderate-late Preterm infants 'BIMP-study'). MRI was performed around term equivalent age (TEA5). MRI scans were assessed for (mild) brain lesions. T2-weighted images were used for automatic segmentation of eight brain structures. Linear regression analysis was performed to compare absolute and relative brain volumes between infants with and without mild brain lesions and between MP and LP infants. RESULTS: 36 MP and 68 LP infants were included. In infants with mild brain lesions, intracranial volume (B = 27.4 cm3, p = 0.02), cerebrospinal fluid (B = 8.78 cm3, p = 0.01) and cerebellar volumes (B = 1.70 cm3, p = 0.03) were significantly larger compared to infants without mild brain lesions. After correction for weight and postmenstrual age at MRI, these volumes were no longer significantly different. LP infants had larger brain volumes than MP infants, but differences were not significant. Relative brain volumes showed no significant differences in both analyses. CONCLUSION: Neither having mild brain lesions, nor being born moderate prematurely affected brain volumes at TEA in MLPT infants.

Enhancing perioperative landmark detection during sacroiliac joint fusion in patients suffering from low back pain.

Over the past decade, minimally invasive sacroiliac joint (SIJ) fusion has become an effective treatment for patients suffering from low back pain (LBP) originating from the SIJ. Perioperative C-arm fluoroscopy-assisted surgical navigation during SIJ fusion remains challenging due to the lack of 3D spatial information. This study developed and assessed a 3D CT/2D fluoroscopy integration approach based on digitally reconstructed radiographs (DRRs) obtained from pre-operative CT scans. Development of this approach proved feasible and landmarks were successfully translated, in retrospect, to perioperatively acquired fluoroscopies. Further expansion of and research into the proposed approach to increase perioperative navigation is indicated and additional validation should be performed.

CT radiation dose reduction in patients with total hip arthroplasties using model-based iterative reconstruction and orthopaedic metal artefact reduction.

OBJECTIVE: To evaluate the impact of radiation dose reduction on image quality in patients with metal-on-metal total hip arthroplasties (THAs) using model-based iterative reconstruction (MBIR) combined with orthopaedic metal artefact reduction (O-MAR). MATERIALS AND METHODS: Patients with metal-on-metal THAs received a pelvic CT with a full (FD) and a reduced radiation dose (RD) with -20%, -40%, -57%, or -80% CT radiation dose respectively, when assigned to group 1, 2, 3, or 4 respectively. FD acquisitions were reconstructed with iterative reconstruction, iDose4. RD acquisitions were additionally reconstructed with iterative model-based reconstruction (IMR) levels 1-3 with different levels of noise suppression. CT numbers, noise and contrast-to-noise ratios were measured in muscle, fat and bladder. Subjective image quality was evaluated on seven aspects including artefacts, osseous structures, prosthetic components and soft tissues. RESULTS: Seventy-six patients were randomly assigned to one of the four groups. While reducing radiation dose by 20%, 40%, 57%, or 80% in combination with IMR, CT numbers remained constant. Compared with iDose4, the noise decreased (p < 0.001) and contrast-to-noise ratios increased (p < 0.001) with IMR. O-MAR improved CT number accuracy in the bladder and reduced noise in the bladder, muscle and fat (p < 0.01). Subjective image quality was rated lower on RD IMR images than FD iDose4 images on all seven aspects (p < 0.05) and was not related to the applied radiation dose reduction. CONCLUSION: In RD IMR with O-MAR images, CT numbers remained constant, noise decreased and contrast-to-noise ratios between muscle and fat increased compared with FD iDose4 with O-MAR images in patients with metal-on-metal THAs. Subjective image quality reduced, regardless of the degree of radiation dose reduction.

Comparison of maximal Rubidium-82 activities for myocardial blood flow quantification between digital and conventional PET systems.

BACKGROUND: PET-based myocardial blood flow (MBF) quantification can be inaccurate when using high tracer activities. Our aim was to derive the maximal Rubidium-82 activity for MBF assessment using a new digital PET system and compare the results with conventional analog systems. METHODS: 1.8 GBq Rubidium-82 was injected into the cardiac insert of an anthropomorphic torso phantom. Data were acquired for 10 min using an Ingenuity TF (Philips Healthcare), Discovery 690 (D690, GE Healthcare), and digital PET prototype system (Philips Healthcare). The dynamic ranges, defined as the maximal measured activity in the reconstructed images deviating < 10% from the true present activity, were determined in all scans. RESULTS: The dynamic ranges were 312 MBq for Ingenuity TF, 650 MBq for D690, and 654 MBq for digital PET prototype. CONCLUSIONS: The maximal Rb-82 activity for MBF assessment using digital PET prototype is higher than that for its analog counterpart (Ingenuity TF), but seems comparable to the D690.

Technical note: how to determine the FDG activity for tumour PET imaging that satisfies European guidelines.

BACKGROUND: For tumour imaging with PET, the literature proposes to administer a patient-specific FDG activity that depends quadratically on a patient's body weight. However, a practical approach on how to implement such a protocol in clinical practice is currently lacking. We aimed to provide a practical method to determine a FDG activity formula for whole-body PET examinations that satisfies both the EANM guidelines and this quadratic relation. RESULTS: We have developed a methodology that results in a formula describing the patient-specific FDG activity to administer. A PET study using the NEMA NU-2001 image quality phantom forms the basis of our method. This phantom needs to be filled with 2.0 and 20.0 kBq FDG/mL in the background and spheres, respectively. After a PET acquisition of 10 min, a reconstruction has to be performed that results in sphere recovery coefficients (RCs) that are within the specifications as defined by the EANM Research Ltd (EARL). By performing reconstructions based on shorter scan durations, the minimal scan time per bed position (T min) needs to be extracted using an image coefficient of variation (COV) of 15 %. At T min, the RCs should be within EARL specifications as well. Finally, the FDG activity (in MBq) to administer can be described by [Formula: see text] with c a constant that is typically 0.0533 (MBq/kg(2)), w the patient's body weight (in kg), and t the scan time per bed position that is chosen in a clinical setting (in seconds). We successfully demonstrated this methodology using a state-of-the-art PET/CT scanner. CONCLUSIONS: We provide a practical method that results in a formula describing the FDG activity to administer to individual patients for whole-body PET examinations, taking into account both the EANM guidelines and a quadratic relation between FDG activity and patient's body weight. This formula is generally applicable to any PET system, using a specified image reconstruction and scan time per bed position.

Computed Tomography Imaging of a Hip Prosthesis Using Iterative Model-Based Reconstruction and Orthopaedic Metal Artefact Reduction: A Quantitative Analysis.

OBJECTIVES: To quantify the combined use of iterative model-based reconstruction (IMR) and orthopaedic metal artefact reduction (O-MAR) in reducing metal artefacts and improving image quality in a total hip arthroplasty phantom. METHODS: Scans acquired at several dose levels and kVps were reconstructed with filtered back-projection (FBP), iterative reconstruction (iDose) and IMR, with and without O-MAR. Computed tomography (CT) numbers, noise levels, signal-to-noise-ratios and contrast-to-noise-ratios were analysed. RESULTS: Iterative model-based reconstruction results in overall improved image quality compared to iDose and FBP (P < 0.001). Orthopaedic metal artefact reduction is most effective in reducing severe metal artefacts improving CT number accuracy by 50%, 60%, and 63% (P < 0.05) and reducing noise by 1%, 62%, and 85% (P < 0.001) whereas improving signal-to-noise-ratios by 27%, 47%, and 46% (P < 0.001) and contrast-to-noise-ratios by 16%, 25%, and 19% (P < 0.001) with FBP, iDose, and IMR, respectively. CONCLUSIONS: The combined use of IMR and O-MAR strongly improves overall image quality and strongly reduces metal artefacts in the CT imaging of a total hip arthroplasty phantom.

Influence of iterative image reconstruction on CT-based calcium score measurements.

Iterative reconstruction techniques for coronary CT angiography have been introduced as an alternative for traditional filter back projection (FBP) to reduce image noise, allowing improved image quality and a potential for dose reduction. However, the impact of iterative reconstruction on the coronary artery calcium score is not fully known. In 112 consecutive stable patients with suspected coronary artery disease, the coronary calcium scores were assessed. Comparisons were made between the Agatston, volume and mass scores obtained with traditional FBP, and by using adaptive statistical iterative reconstruction (ASIR). A significant reduction of the Agatston score, volume score and mass score was observed for ASIR when compared to FBP, with median differences of resp. 26, 5 mm(3) and 1 mg. Using the ASIR reconstruction, the number of patients with a calcium score of zero increased by 13 %. Iterative CT reconstruction significantly reduces the Agatston, volume and mass scores. Since the calcium score is used as a prognostic tool for coronary artery disease, caution must be taken when using iterative reconstruction.