Computed tomography (CT) and magnetic resonance imaging (MRI) can quantify muscle mass and quality. However, it is still unclear if CT and MRI derived measurements can be used interchangeable. In ...this prospective study, fifty consecutive participants of a cancer screening program underwent same day low-dose chest CT and MRI. Cross-sectional areas (CSA) of the paraspinal skeletal muscles were obtained. CT and MRI muscle fat infiltration (MFI) were assessed by mean radiodensity in Hounsfield units (HU) and proton density fat fraction (MRI
), respectively. CSA and MFI were highly correlated between CT and MRI (CSA: r = 0.93, P < 0.001; MFI: r = - 0.90, P < 0.001). Mean CSA was higher in CT compared to MRI (46.6cm
versus 43.0cm
; P = 0.05) without significance. Based on MRI
, a linear regression model was established to directly estimate skeletal muscle fat content from CT. Bland-Altman plots showed a difference between measurements of - 0.5 cm
to 7.6 cm
and - 4.2% to 2.4% regarding measurements of CSA and MFI, respectively. In conclusion, the provided results indicate interchangeability of CT and MRI derived imaging biomarkers of skeletal muscle quantity and quality. Comparable to MRI
, skeletal muscle fat content can be quantified from CT, which might have an impact of analyses in larger cohort studies, particularly in sarcopenia patients.
Although CT and MRI are standard procedures in cirrhosis diagnosis, differentiation of etiology based on imaging is not established. This proof-of-concept study explores the potential of deep ...learning (DL) to support imaging-based differentiation of the etiology of liver cirrhosis. This retrospective, monocentric study included 465 patients with confirmed diagnosis of (a) alcoholic (n = 221) and (b) other-than-alcoholic (n = 244) cirrhosis. Standard T2-weighted single-slice images at the caudate lobe level were randomly split for training with fivefold cross-validation (85%) and testing (15%), balanced for (a) and (b). After automated upstream liver segmentation, two different ImageNet pre-trained convolutional neural network (CNN) architectures (ResNet50, DenseNet121) were evaluated for classification of alcohol-related versus non-alcohol-related cirrhosis. The highest classification performance on test data was observed for ResNet50 with unfrozen pre-trained parameters, yielding an area under the receiver operating characteristic curve of 0.82 (95% confidence interval (CI) 0.71-0.91) and an accuracy of 0.75 (95% CI 0.64-0.85). An ensemble of both models did not lead to significant improvement in classification performance. This proof-of-principle study shows that deep-learning classifiers have the potential to aid in discriminating liver cirrhosis etiology based on standard MRI.
Objectives
To investigate the diagnostic performance of deep transfer learning (DTL) to detect liver cirrhosis from clinical MRI.
Methods
The dataset for this retrospective analysis consisted of 713 ...(343 female) patients who underwent liver MRI between 2017 and 2019. In total, 553 of these subjects had a confirmed diagnosis of liver cirrhosis, while the remainder had no history of liver disease. T2-weighted MRI slices at the level of the caudate lobe were manually exported for DTL analysis. Data were randomly split into training, validation, and test sets (70%/15%/15%). A ResNet50 convolutional neural network (CNN) pre-trained on the ImageNet archive was used for cirrhosis detection with and without upstream liver segmentation. Classification performance for detection of liver cirrhosis was compared to two radiologists with different levels of experience (4
th
-year resident, board-certified radiologist). Segmentation was performed using a U-Net architecture built on a pre-trained ResNet34 encoder. Differences in classification accuracy were assessed by the
χ
2
-test.
Results
Dice coefficients for automatic segmentation were above 0.98 for both validation and test data. The classification accuracy of liver cirrhosis on validation (vACC) and test (tACC) data for the DTL pipeline with upstream liver segmentation (vACC = 0.99, tACC = 0.96) was significantly higher compared to the resident (vACC = 0.88,
p
< 0.01; tACC = 0.91,
p
= 0.01) and to the board-certified radiologist (vACC = 0.96,
p
< 0.01; tACC = 0.90,
p
< 0.01).
Conclusion
This proof-of-principle study demonstrates the potential of DTL for detecting cirrhosis based on standard T2-weighted MRI. The presented method for image-based diagnosis of liver cirrhosis demonstrated expert-level classification accuracy.
Key Points
• A pipeline consisting of two convolutional neural networks (CNNs) pre-trained on an extensive natural image database (ImageNet archive) enables detection of liver cirrhosis on standard T2-weighted MRI.
• High classification accuracy can be achieved even without altering the pre-trained parameters of the convolutional neural networks.
• Other abdominal structures apart from the liver were relevant for detection when the network was trained on unsegmented images.
Objectives
To develop a pipeline for automated body composition analysis and skeletal muscle assessment with integrated quality control for large-scale application in opportunistic imaging.
Methods
...First, a convolutional neural network for extraction of a single slice at the L3/L4 lumbar level was developed on CT scans of 240 patients applying the nnU-Net framework. Second, a 2D competitive dense fully convolutional U-Net for segmentation of visceral and subcutaneous adipose tissue (VAT, SAT), skeletal muscle (SM), and subsequent determination of fatty muscle fraction (FMF) was developed on single CT slices of 1143 patients. For both steps, automated quality control was integrated by a logistic regression model classifying the presence of L3/L4 and a linear regression model predicting the segmentation quality in terms of Dice score. To evaluate the performance of the entire pipeline end-to-end, body composition metrics, and FMF were compared to manual analyses including 364 patients from two centers.
Results
Excellent results were observed for slice extraction (
z
-deviation = 2.46 ± 6.20 mm) and segmentation (Dice score for SM = 0.95 ± 0.04, VAT = 0.98 ± 0.02, SAT = 0.97 ± 0.04) on the dual-center test set excluding cases with artifacts due to metallic implants. No data were excluded for end-to-end performance analyses. With a restrictive setting of the integrated segmentation quality control, 39 of 364 patients were excluded containing 8 cases with metallic implants. This setting ensured a high agreement between manual and fully automated analyses with mean relative area deviations of ΔSM = 3.3 ± 4.1%, ΔVAT = 3.0 ± 4.7%, ΔSAT = 2.7 ± 4.3%, and ΔFMF = 4.3 ± 4.4%.
Conclusions
This study presents an end-to-end automated deep learning pipeline for large-scale opportunistic assessment of body composition metrics and sarcopenia biomarkers in clinical routine.
Key Points
• Body composition metrics and skeletal muscle quality can be opportunistically determined from routine abdominal CT scans.
• A pipeline consisting of two convolutional neural networks allows an end-to-end automated analysis.
• Machine-learning-based quality control ensures high agreement between manual and automatic analysis.
Purpose
To study the clinical potential of a deep learning neural network (convolutional neural networks CNN) as a supportive tool for detection of intracranial aneurysms from 3D time-of-flight ...magnetic resonance angiography (TOF-MRA) by comparing the diagnostic performance to that of human readers.
Methods
In this retrospective study a pipeline for detection of intracranial aneurysms from clinical TOF-MRA was established based on the framework DeepMedic. Datasets of 85 consecutive patients served as ground truth and were used to train and evaluate the model. The ground truth without annotation was presented to two blinded human readers with different levels of experience in diagnostic neuroradiology (reader 1: 2 years, reader 2: 12 years). Diagnostic performance of human readers and the CNN was studied and compared using the χ
2
-test and Fishers’ exact test.
Results
Ground truth consisted of 115 aneurysms with a mean diameter of 7 mm (range: 2–37 mm). Aneurysms were categorized as small (S; <3 mm;
N
= 13), medium (M; 3–7 mm;
N
= 57), and large (L; >7 mm;
N
= 45) based on the diameter. No statistically significant differences in terms of overall sensitivity (OS) were observed between the CNN and both of the human readers (reader 1 vs. CNN,
P
= 0.141; reader 2 vs. CNN,
P
= 0.231). The OS of both human readers was improved by combination of each readers’ individual detections with the detections of the CNN (reader 1: 98% vs. 95%,
P
= 0.280; reader 2: 97% vs. 94%,
P
= 0.333).
Conclusion
A CNN is able to detect intracranial aneurysms from clinical TOF-MRA data with a sensitivity comparable to that of expert radiologists and may have the potential to improve detection rates of incidental findings in a clinical setting.
Background
Chemical shift‐encoding based water‐fat MRI is an emerging method to noninvasively assess proton density fat fraction (PDFF), a promising quantitative imaging biomarker for estimating ...tissue fat concentration. However, in vivo validation of PDFF is still lacking for bone marrow applications.
Purpose
To determine the accuracy and precision of MRI‐determined vertebral bone marrow PDFF among different readers and across different field strengths and imager manufacturers.
Study Type
Repeatability/reproducibility.
Subjects
Twenty‐four adult volunteers underwent lumbar spine MRI with one 1.5T and two different 3.0T MR scanners from two vendors on the same day.
Field Strength/Sequence
1.5T and 3.0T/3D spoiled‐gradient echo multipoint Dixon sequences.
Assessment
Two independent readers measured intravertebral PDFF for the three most central slices of the L1–5 vertebral bodies. Single‐voxel MR spectroscopy (MRS)‐determined PDFF served as the reference standard for PDFF estimation.
Statistical Tests
Accuracy and bias were assessed by Pearson correlation, linear regression analysis, and Bland–Altman plots. Repeatability and reproducibility were evaluated by Wilcoxon signed rank test, Friedman test, and coefficients of variation. Intraclass correlation coefficients were used to validate intra‐ and interreader as well as intraimager agreements.
Results
MRI‐based PDFF estimates of lumbar bone marrow were highly correlated (r2 = 0.899) and accurate (mean bias, –0.6%) against the MRS‐determined PDFF reference standard. PDFF showed high linearity (r2 = 0.972–0.978) and small mean bias (0.6–1.5%) with 95% limits of agreement within ±3.4% across field strengths, imaging platforms, and readers. Repeatability and reproducibility of PDFF were high, with the mean overall coefficient of variation being 0.86% and 2.77%, respectively. The overall intraclass correlation coefficient was 0.986 as a measure for an excellent interreader agreement.
Data Conclusion
MRI‐based quantification of vertebral bone marrow PDFF is highly accurate, repeatable, and reproducible among readers, field strengths, and MRI platforms, indicating its robustness as a quantitative imaging biomarker for multicentric studies.
Level of Evidence: 3
Technical Efficacy: Stage 2
J. Magn. Reson. Imaging 2019;50:1762–1772.
Impaired skeletal muscle quality is a major risk factor for adverse outcomes in acute respiratory failure. However, conventional methods for skeletal muscle assessment are inapplicable in the ...critical care setting. This study aimed to determine the prognostic value of computed tomography (CT) fatty muscle fraction (FMF) as a biomarker of muscle quality in patients undergoing extracorporeal membrane oxygenation (ECMO). To calculate FMF, paraspinal skeletal muscle area was obtained from clinical CT and separated into areas of fatty and lean muscle based on densitometric thresholds. The cohort was binarized according to median FMF. Patients with high FMF displayed significantly increased 1-year mortality (72.7% versus 55.8%, P = 0.036) on Kaplan-Meier analysis. A multivariable logistic regression model was built to test the impact of FMF on outcome. FMF was identified as a significant predictor of 1-year mortality (hazard ratio per percent FMF, 1.017 95% confidence interval, 1.002-1.033; P = 0.031), independent of anthropometric characteristics, Charlson Comorbidity Index, Simplified Acute Physiology Score, Respiratory Extracorporeal Membrane Oxygenation Survival Prediction Score, and duration of ECMO support. To conclude, FMF predicted 1-year mortality independently of established clinical prognosticators in ECMO patients and may have the potential to become a new muscle quality imaging biomarker, which is available from clinical CT.
Purpose
The use of pharmacomechanical thrombectomy in patients with symptomatic iliofemoral deep venous thrombosis (DVT) not responsive to conservative treatment is under-investigated until now. This ...prompted us to review and analyze our results (technical/clinical outcome, complications) and compare them to the current literature.
Materials and Methods
Between 2013 and 2019, 19 patients (14 women and 5 men; mean age: 41.2 years, SD: 18.2) with iliofemoral DVT and excessive pain not responsive to conservative treatment were treated with pharmacomechanical thrombectomy. Patients were followed up for 12 months. Demographics, technical success and clinical outcome data (pain score/Villalta score) were collected.
Results
Thrombectomy ± thrombolysis was successful in all cases (
n
= 19). No major complications were observed. Eight out of nineteen cases developed hematoma at the sheath insertion site not requiring further treatment. Seven out of nineteen cases required additional continuous lysis for complete iliofemoral clot solution. All patients received balloon angioplasty to treat post-thrombotic strictures. In 16/19 cases, stents were placed to preserve iliofemoral caliber and maintain unrestricted iliac venous outflow. Three patients (16%) required re-intervention due to re-thrombosis or in-stent stenosis after 4, 14 days and 4 months, respectively. Symptoms could be improved temporarily or indefinitely in 19 out of 19 patients. 1 year following thrombectomy mean pain score was reduced by 87%, mean Villalta score was 2.6 (SD: 4), and iliofemoral veins were patent in 15/17 patients.
Conclusion
In symptomatic patients with iliofemoral DVT, refractory to conservative treatment, catheter-directed thrombectomy enables rapid and long-lasting pain relief. High patency rates can be achieved in patients receiving PTA and venous stenting post-thrombectomy.
Medication-related osteonecrosis of the jaw (MRONJ) is a well known side-effect of anti-resorptive drugs. Changes in bone density might potentially constitute the development of ONJ. This study aimed ...to investigate, to which degree bisphosphonates (bp) and denosumab (db) induce changes in bone density that can be determined from routine diagnostic CT.
CT scans of 101 patients were investigated. MRONJ was present in 61 patients (
= 26: db-treated;
= 35 bp-treated). 40 patients were included as a reference group. Bone density was measured at two distinct locations in the mandible (M1: anterior of the mental foramen; M2: retromolar), each on the contralateral side to the necrosis.
The bone density values measured at both locations were found to be significantly higher in the bp-group compared to the db-group (
= 0.027) and to the reference-group (
= 0.016). Almost no difference (
= 0.84) in bone density value was found between the db- and reference-groups.Investigating the effect of duration of treatment, none of the measured values showed significant differences in both locations of db- and bp-group.
The findings from this study suggest that that bisphosphonates change the microarchitecture of the alveolar bone by being embedded in the mandible, which may subsequently lead to a bp-specific corticalization, and a decrease in vascularization of the lower jaw. This process may be distinctive for bp-treatment and seems to induce the congestion of cancellous bone rather rapidly after the first administrations. This effect could not be determined in denosumab-treated patients.