Standard T2-weighted MRI Research Articles

Detection of liver cirrhosis in standard T2-weighted MRI using deep transfer learning

ObjectivesTo investigate the diagnostic performance of deep transfer learning (DTL) to detect liver cirrhosis from clinical MRI.MethodsThe 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 (4th-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 χ2-test.ResultsDice 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).ConclusionThis 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.

T1 Mapping Quantifies Spinal Cord Compression in Patients With Various Degrees of Cervical Spinal Canal Stenosis.

Age-related degeneration of the cervical spinal column is the most common cause of spinal cord lesions. T1 mapping has been shown to indicate the grade and site of spinal cord compression in low grade spinal canal stenosis (SCS). Aim of our study was to further investigate the diagnostic potential of a novel T1 mapping method at 0.75 mm resolution and 4 s acquisition time in 31 patients with various grades of degenerative cervical SCS. T1 mapping was performed in axial sections of the stenosis as well as above and below. Included subjects received standard T2-weighted MRI of the cervical spine (including SCS-grading 0-III), electrophysiological, and clinical examination. We found that patients with cervical SCS showed a significant difference in T1 relaxation times within the stenosis (727 ± 66 ms, mean ± standard deviation) in comparison to non-stenotic segments above (854 ± 104 ms, p < 0.001) and below (893 ± 137 ms, p < 0.001). There was no difference in mean T1 in non-stenotic segments in patients (p = 0.232) or between segments in controls (p = 0.272). Mean difference of the T1 relaxation times was significantly higher in grade III stenosis (234 ± 45) vs. in grade II stenosis (176 ± 45, p = 0.037) vs. in grade I stenosis (90 ± 87 ms, p = 0.010). A higher difference in T1 relaxation time was associated with a central efferent conduction deficit. In conclusion, T1 mapping may be useful as a tool for SCS quantification in all grades of SCS, including high-grade stenosis with myelopathy signal in conventional T2-weighted imaging.

TSPO-PET and diffusion-weighted MRI for imaging a mouse model of infiltrative human glioma.

Glioblastoma (GBM) is the most devastating brain tumor. Despite the use of multimodal treatments, most patients relapse, often due to the highly invasive nature of gliomas. However, the detection of glioma infiltration remains challenging. The aim of this study was to assess advanced PET and MRI techniques for visualizing biological activity and infiltration of the tumor. Using multimodality imaging, we investigated [18F]DPA-714, a radiotracer targeting the 18 kDa translocator protein (TSPO), [18F]FET PET, non-Gaussian diffusion MRI (apparent diffusion coefficient, kurtosis), and the S-index, a composite diffusion metric, to detect tumor infiltration in a human invasive glioma model. In vivo imaging findings were confirmed by autoradiography and immunofluorescence. Increased tumor-to-contralateral [18F]DPA-714 uptake ratios (1.49 ± 0.11) were found starting 7 weeks after glioma cell implantation. TSPO-PET allowed visualization of glioma infiltration into the contralateral hemisphere 2 weeks earlier compared with the clinically relevant biomarker for biological glioma activity [18F]FET. Diffusion-weighted imaging (DWI), in particular kurtosis, was more sensitive than standard T2-weighted MRI to detect differences between the glioma-bearing and the contralateral hemisphere at 5 weeks. Immunofluorescence data reflect in vivo findings. Interestingly, labeling for tumoral and stromal TSPO indicates a predominant expression of TSPO by tumor cells. These results suggest that advanced PET and MRI methods, such as [18F]DPA-714 and DWI, may be superior to standard imaging methods to visualize glioma growth and infiltration at an early stage.

Quantification of spinal cord compression using T1 mapping in patients with cervical spinal canal stenosis - Preliminary experience.

BackgroundDegenerative changes of the cervical spinal column are the most common cause of spinal cord lesions in the elderly. Conventional clinical, electrophysiological and radiological diagnostics of spinal cord compression are often inconsistent.Materials and methodsThe feasibility and diagnostic potential of a novel T1 mapping method at 0.5 mm resolution and 4 s acquisition time was evaluated in 14 patients with degenerative cervical spinal canal stenosis (SCS) and 6 healthy controls. T1 mapping was performed in axial sections of the stenosis as well as above and below. All subjects received standard T2-weighted MRI of the cervical spine (including SCS-grading 0-III), electrophysiological and clinical examinations.ResultsPatients revealed significantly decreased T1 relaxation times of the compressed spinal cord within the SCS (912 ± 53 ms, mean ± standard deviation) in comparison to unaffected segments above (1027 ± 39 ms, p < .001) and below (1056 ± 93 ms, p < .001). There was no difference in mean T1 in unaffected segments in patients (p = .712) or between segments in controls (p = .443). Moreover, T1 values were significantly lower in grade II (881 ± 46 ms, p = .005) than in grade I SCS (954 ± 29 ms). Patients with central conduction deficit tended to have lower T1 values within the SCS than patients without (909 ± 50 ms vs 968 ± 7 ms, p = .069).ConclusionRapid high-resolution T1 mapping is a robust MRI method for quantifying spinal cord compression in patients with cervical SCS. It promises additional diagnostic insights and warrants more extended patient studies.

SU‐E‐P‐33: Critical Role of T2‐Weighted Imaging Combined with Diffusion‐Weighted Imaging of MRI in Diagnosis of Loco‐Regional Recurrent Esophageal Cancer After Radical Surgery

Purpose:We perform this study to investigate the diagnostic efficacy of T2‐weighted MRI (T2WI) and diffusion‐weighted MRI (DWI) in confirming local relapses of esophageal cancer in patients highly suspected of recurrence after eradicating surgery.Methods:Forty‐two postoperative esophageal cancer patients with clinical suspicions of cancer recurrence underwent 3.0T MRI applying axial, coronal, sagittal T2WI and axial DWI sequences. Two experienced radiologists (R1 and R2) both used two methods (T2WI, T2WI+DWI) to observe the images, and graded the patients ranging from 1 to 5 to represent severity of the disease based on visual signal intensity (patients equal to or more than grade 3 was confirmed as recurrent disease)Results:27/42patients were verified of recurrent disease by pathologic findings and/or imaging findings during follow‐up. The sensitivity, specificity and accuracy of R1 applying T2WI+DWI are 96%, 87% and 93% versus 81%, 80% and 77% on T2WI, these figures by R2 were 96%, 93% and 95% versus 89%, 93% and 90%. The receiver operating curve (ROC) analyses suggest that both of the two readers can obtain better accuracy when adding DWI to T2WI compared with T2WI alone. Kappa test between R1 and R2 indicates excellent inter‐observer agreement on T2WI+DWI.Conclusion:Standard T2WI in combination DWI can achieve better accuracy than T2WI alone in diagnosing local recurrence of esophageal cancer, and improve consistency between different readers.