Clinical utility of quantitative imaging

Future of Team-based Basic and Translational Science in Radiation Oncology

To further optimize radiotherapy, a more personalized treatment towards individual patient's risk profiles, dissecting both patient-specific tumor and normal tissue response to multimodality treatments is needed. Novel developments in radiobiology, using in vitro patient-specific complex tissue resembling 3D models and multiomics approaches at a spatial single-cell level, may provide unprecedented insight into the radiation responses of tumors and normal tissue. Here, we describe the necessary team effort, including all disciplines in radiation oncology, to integrate such data into clinical prediction models and link the relatively "big data" from the clinical practice, allowing accurate patient stratification for personalized treatment approaches.

Towards an EKG for SBO: A Neural Network for Detection and Characterization of Bowel Obstruction on CT

A neural network was developed to detect and characterize bowel obstruction, a common cause of acute abdominal pain. In this retrospective study, 202 CT scans of 165 patients with bowel obstruction from March to June 2022 were included and partitioned into training and test data sets. A multi-channel neural network was trained to segment the gastrointestinal tract, and to predict the diameter and the longitudinal position ("longitude") along the gastrointestinal tract using a novel embedding. Its performance was compared to manual segmentations using the Dice score, and to manual measurements of the diameter and longitude using intraclass correlation coefficients (ICC). ROC curves as well as sensitivity and specificity were calculated for diameters above a clinical threshold for obstruction, and for longitudes corresponding to small bowel. In the test data set, Dice score for segmentation of the gastrointestinal tract was 78 ± 8%. ICC between measured and predicted diameters was 0.72, indicating moderate agreement. ICC between measured and predicted longitude was 0.85, indicating good agreement. AUROC was 0.90 for detection of dilated bowel, and was 0.95 and 0.90 for differentiation of the proximal and distal gastrointestinal tract respectively. Overall sensitivity and specificity for dilated small bowel were 0.83 and 0.90. Since obstruction is diagnosed based on the diameter and longitude of the bowel, this neural network and embedding may enable detection and characterization of this important disease on CT.

Photon-Counting Computed Tomography for Microstructural Imaging of Bone and Joints

PURPOSE OF REVIEW

Recently, photon-counting computed tomography (PCCT) has been introduced in clinical research and diagnostics. This review describes the technological advances and provides an overview of recent applications with a focus on imaging of bone.

RECENT FINDINGS

PCCT is a full-body scanner with short scanning times that provides better spatial and spectral resolution than conventional energy-integrating-detector CT (EID-CT), along with an up to 50% reduced radiation dose. It can be used to quantify bone mineral density, to perform bone microstructural analyses and to assess cartilage quality with adequate precision and accuracy. Using a virtual monoenergetic image reconstruction, metal artefacts can be greatly reduced when imaging bone-implant interfaces. Current PCCT systems do not allow spectral imaging in ultra-high-resolution (UHR) mode. Given its improved resolution, reduced noise and spectral imaging capabilities PCCT has diagnostic capacities in both qualitative and quantitative imaging that outperform those of conventional CT. Clinical use in monitoring bone health has already been demonstrated. The full potential of PCCT systems will be unlocked when UHR spectral imaging becomes available.

Quantitative Cartilage T2 and T1rho Mapping: Is There a Clinical Role? From the AJR Special Series on Quantitative Imaging

Despite more than 20 years of development, the MRI-based cartilage compositional biomarkers T2 and T1rho have not been routinely applied in clinical practice. This review examines these measures' historical development and frames the challenges in the application of these quantitative imaging tools to the care of patients with cartilage injury and osteoarthritis using the hierarchical model of efficacy proposed by Fryback and Thornbury. T2 and T1rho have been validated for the evaluation of early compositional and structural changes in cartilage extracellular matrix. Yet, these biomarkers lack direct correlation with pain or function loss, lack standardization of methods for acquisition and analysis, and have a limited role in guiding therapeutic management given the absence of effective disease-modifying osteoarthritis drugs. These issues present significant challenges in the path to the biomarkers' future implementation in clinical care. Nonetheless, these MRI-based cartilage compositional biomarkers provide an essential tool for musculoskeletal research and can provide important information on the biophysical properties of cartilage that will continue to contribute to our understanding of cartilage injury and osteoarthritis pathogenesis.

DTI-MR fingerprinting for rapid high-resolution whole-brain T1 , T2 , proton density, ADC, and fractional anisotropy mapping

PURPOSE

This study aims to develop a high-efficiency and high-resolution 3D imaging approach for simultaneous mapping of multiple key tissue parameters for routine brain imaging, including T1 , T2 , proton density (PD), ADC, and fractional anisotropy (FA). The proposed method is intended for pushing routine clinical brain imaging from weighted imaging to quantitative imaging and can also be particularly useful for diffusion-relaxometry studies, which typically suffer from lengthy acquisition time.

METHODS

To address challenges associated with diffusion weighting, such as shot-to-shot phase variation and low SNR, we integrated several innovative data acquisition and reconstruction techniques. Specifically, we used M1-compensated diffusion gradients, cardiac gating, and navigators to mitigate phase variations caused by cardiac motion. We also introduced a data-driven pre-pulse gradient to cancel out eddy currents induced by diffusion gradients. Additionally, to enhance image quality within a limited acquisition time, we proposed a data-sharing joint reconstruction approach coupled with a corresponding sequence design.

RESULTS

The phantom and in vivo studies indicated that the T1 and T2 values measured by the proposed method are consistent with a conventional MR fingerprinting sequence and the diffusion results (including diffusivity, ADC, and FA) are consistent with the spin-echo EPI DWI sequence.

CONCLUSION

The proposed method can achieve whole-brain T1 , T2 , diffusivity, ADC, and FA maps at 1-mm isotropic resolution within 10 min, providing a powerful tool for investigating the microstructural properties of brain tissue, with potential applications in clinical and research settings.

Need for Objective Task-Based Evaluation of Image Segmentation Algorithms for Quantitative PET: A Study with ACRIN 6668/RTOG 0235 Multicenter Clinical Trial Data

Reliable performance of PET segmentation algorithms on clinically relevant tasks is required for their clinical translation. However, these algorithms are typically evaluated using figures of merit (FoMs) that are not explicitly designed to correlate with clinical task performance. Such FoMs include the Dice similarity coefficient (DSC), the Jaccard similarity coefficient (JSC), and the Hausdorff distance (HD). The objective of this study was to investigate whether evaluating PET segmentation algorithms using these task-agnostic FoMs yields interpretations consistent with evaluation on clinically relevant quantitative tasks. Methods: We conducted a retrospective study to assess the concordance in the evaluation of segmentation algorithms using the DSC, JSC, and HD and on the tasks of estimating the metabolic tumor volume (MTV) and total lesion glycolysis (TLG) of primary tumors from PET images of patients with non-small cell lung cancer. The PET images were collected from the American College of Radiology Imaging Network 6668/Radiation Therapy Oncology Group 0235 multicenter clinical trial data. The study was conducted in 2 contexts: (1) evaluating conventional segmentation algorithms, namely those based on thresholding (SUVmax40% and SUVmax50%), boundary detection (Snakes), and stochastic modeling (Markov random field-Gaussian mixture model); (2) evaluating the impact of network depth and loss function on the performance of a state-of-the-art U-net-based segmentation algorithm. Results: Evaluation of conventional segmentation algorithms based on the DSC, JSC, and HD showed that SUVmax40% significantly outperformed SUVmax50%. However, SUVmax40% yielded lower accuracy on the tasks of estimating MTV and TLG, with a 51% and 54% increase, respectively, in the ensemble normalized bias. Similarly, the Markov random field-Gaussian mixture model significantly outperformed Snakes on the basis of the task-agnostic FoMs but yielded a 24% increased bias in estimated MTV. For the U-net-based algorithm, our evaluation showed that although the network depth did not significantly alter the DSC, JSC, and HD values, a deeper network yielded substantially higher accuracy in the estimated MTV and TLG, with a decreased bias of 91% and 87%, respectively. Additionally, whereas there was no significant difference in the DSC, JSC, and HD values for different loss functions, up to a 73% and 58% difference in the bias of the estimated MTV and TLG, respectively, existed. Conclusion: Evaluation of PET segmentation algorithms using task-agnostic FoMs could yield findings discordant with evaluation on clinically relevant quantitative tasks. This study emphasizes the need for objective task-based evaluation of image segmentation algorithms for quantitative PET.

Accuracy of iodine quantification in dual energy CT: A phantom study across 3 different CT systems

INTRODUCTION

No study has rigorously compared the performances of iodine quantification on recent CT systems employing different emission-based technologies, depending on the manufacturers and models.

METHODS

A specific bespoke phantom was used for this study, with 12 known concentrations of iodinated contrast agent: 0.4, 0.5, 1.0, 2.0, 3.0, 4.0, 5.0, 10.0, 15.0, 20.0, 30.0 and 50.0 mg/mL. Three different dual-energy scanners were tested: one system using dual-source acquisition (CT#1) and two systems using Fast kilovolt-peak switching technology ± artificial intelligence (AI) reconstruction methods (CT#2 and #3) from two different manufacturers. For each system, helical scans were performed following recommended clinical protocols. Four acquisitions were performed per iodine concentration (mg/mL), and measurements were made on iodine-maps using ROIs. Mean measured values were compared to the known concentrations, and the absolute quantification error (AQE) and the relative percentage error (RPE) were used to compare the performances of each CT.

RESULTS

The accuracy of the obtained measurements varied depending on the studied model but not on the acquisition mode (dual-source vs kVp switch ± AI). The quantification was more precise at high concentrations. RPE values were below 10 % with CT#2 (kVp switch) and below 25 % with CT#1 (dual-source), but were significantly higher with CT#3 (kVp switch + AI), exceeding 50 % at low concentrations (<3 mg/mL).

CONCLUSIONS

With the help of a phantom, we identified variability in the results accuracy depending on the CT model, with sometimes significant deviation. Considering the performances of the different DECT technologies in iodine mapping, dual-source (CT#1) and kVp switch (CT#2) technologies appear more accurate than kVp switch technology combined with deep-learning-based reconstruction (CT#3) especially at low concentrations (<3 mg/mL).

IMPLICATIONS FOR PRACTICE

As the primary and daily user of medical imaging devices, the radiographer role is to be attentive to the performance of imaging systems, particularly when performing quantitative acquisitions like iodine-quantification. In CT quantitative imaging (iodine map), it's essential for radiographers to consider their CT systems as measuring tools, and to be aware of their accuracies and limits.

Smart capsule for monitoring inflammation profile throughout the gastrointestinal tract

Inflammatory bowel disease (IBD) has become alarmingly prevalent in the last two decades affecting 6.8 million people worldwide with a starkly high relapse rate of 40% within 1 year of remission. Existing visual endoscopy techniques rely on subjective assessment of images that are error-prone and insufficient indicators of early-stage IBD, rendering them unsuitable for frequent and quantitative monitoring of gastrointestinal health necessary for detecting regular relapses in IBD patients. To address these limitations, we have implemented a miniaturized smart capsule (2.2 cm × 11 mm) that allows monitoring reactive oxygen species (ROS) levels as a biomarker of inflammation for quantitative and frequent profiling of inflammatory lesions throughout the gastrointestinal tract. The capsule is composed of a pH and oxidation reduction potential (ORP) sensor to track the capsule's location and ROS levels throughout the gastrointestinal tract, respectively, and an optimized electronic interface for wireless sensing and data communication. The designed sensors provided a linear and stable performance within the physiologically relevant range of the GI tract (pH: 1-8 and ORP: -500 to +500 mV). Additionally, systematic design optimization of the wireless interface electronics offered an efficient sampling rate of 10 ms for long-running measurements up to 48 h for a complete evaluation of the entire gastrointestinal tract. As a proof-of-concept, the capsule the capsule's performance in detecting inflammation risks was validated by conducting tests on in vitro cell culture conditions, simulating healthy and inflamed gut-like environments. The capsule presented here achieves a new milestone in addressing the emerging need for smart ingestible electronics for better diagnosis and treatment of digestive diseases.

Tailored magnetic resonance fingerprinting

Neuroimaging of certain pathologies requires both multi-parametric qualitative and quantitative imaging. The role of the quantitative MRI (qMRI) is well accepted but suffers from long acquisition times leading to patient discomfort, especially in geriatric and pediatric patients. Previous studies show that synthetic MRI can be used in order to reduce the scan time and provide qMRI as well as multi-contrast data. However, this approach suffers from artifacts such as partial volume and flow. In order to increase the scan efficiency (the number of contrasts and quantitative maps acquired per unit time), we designed, simulated, and demonstrated rapid, simultaneous, multi-contrast qualitative (T₁ weighted, T₁ fluid attenuated inversion recovery (FLAIR), T₂ weighted, water, and fat), and quantitative imaging (T₁ and T₂ maps) through the approach of tailored MR fingerprinting (TMRF) to cover whole-brain in approximately four minutes. We performed TMRF on in vivo four healthy human brains and in vitro ISMRM/NIST phantom and compared with vendor supplied gold standard (GS) and MRF sequences. All scans were performed on a 3 T GE Premier system and images were reconstructed offline using MATLAB. The reconstructed qualitative images were then subjected to custom DL denoising and gradient anisotropic diffusion denoising. The quantitative tissue parametric maps were reconstructed using a dense neural network to gain computational speed compared to dictionary matching. The grey matter and white matter tissues in qualitative and quantitative data for the in vivo datasets were segmented semi-automatically. The SNR and mean contrasts were plotted and compared across all three methods. The GS images show better SNR in all four subjects compared to MRF and TMRF (GS > TMRF>MRF). The T₁ and T₂ values of MRF are relatively overestimated as compared to GS and TMRF. The scan efficiency for TMRF is 1.72 min^-1 which is higher compared to GS (0.32 min^-1) and MRF (0.90 min^-1).

In vivo quantitative MRI: T1 and T2 measurements of the human brain at 0.064 T

OBJECTIVE

To measure healthy brain [Formula: see text] and [Formula: see text] relaxation times at 0.064 T.

MATERIALS AND METHODS

[Formula: see text] and [Formula: see text] relaxation times were measured in vivo for 10 healthy volunteers using a 0.064 T magnetic resonance imaging (MRI) system and for 10 test samples on both the MRI and a separate 0.064 T nuclear magnetic resonance (NMR) system. In vivo [Formula: see text] and [Formula: see text] values are reported for white matter (WM), gray matter (GM), and cerebrospinal fluid (CSF) for automatic segmentation regions and manual regions of interest (ROIs).

RESULTS

[Formula: see text] sample measurements on the MRI system were within 10% of the NMR measurement for 9 samples, and one sample was within 11%. Eight [Formula: see text] sample MRI measurements were within 25% of the NMR measurement, and the two longest [Formula: see text] samples had more than 25% variation. Automatic segmentations generally resulted in larger [Formula: see text] and [Formula: see text] estimates than manual ROIs.

DISCUSSION

[Formula: see text] and [Formula: see text] times for brain tissue were measured at 0.064 T. Test samples demonstrated accuracy in WM and GM ranges of values but underestimated long [Formula: see text] in the CSF range. This work contributes to measuring quantitative MRI properties of the human body at a range of field strengths.

An untrained deep learning method for reconstructing dynamic MR images from accelerated model-based data

PURPOSE

To implement physics-based regularization as a stopping condition in tuning an untrained deep neural network for reconstructing MR images from accelerated data.

METHODS

The ConvDecoder (CD) neural network was trained with a physics-based regularization term incorporating the spoiled gradient echo equation that describes variable-flip angle data. Fully-sampled variable-flip angle k-space data were retrospectively accelerated by factors of R = {8, 12, 18, 36} and reconstructed with CD, CD with the proposed regularization (CD + r), locally low-rank (LR) reconstruction, and compressed sensing with L1-wavelet regularization (L1). Final images from CD + r training were evaluated at the "argmin" of the regularization loss; whereas the CD, LR, and L1 reconstructions were chosen optimally based on ground truth data. The performance measures used were the normalized RMS error, the concordance correlation coefficient, and the structural similarity index.

RESULTS

The CD + r reconstructions, chosen using the stopping condition, yielded structural similarity indexs that were similar to the CD (p = 0.47) and LR structural similarity indexs (p = 0.95) across R and that were significantly higher than the L1 structural similarity indexs (p = 0.04). The concordance correlation coefficient values for the CD + r T1 maps across all R and subjects were greater than those corresponding to the L1 (p = 0.15) and LR (p = 0.13) T1 maps, respectively. For R ≥ 12 (≤4.2 min scan time), L1 and LR T1 maps exhibit a loss of spatially refined details compared to CD + r.

CONCLUSION

The use of an untrained neural network together with a physics-based regularization loss shows promise as a measure for determining the optimal stopping point in training without relying on fully-sampled ground truth data.

The Additive Value of Radiomics Features Extracted from Baseline MR Images to the Barcelona Clinic Liver Cancer (BCLC) Staging System in Predicting Transplant-Free Survival in Patients with Hepatocellular Carcinoma: A Single-Center Retrospective Analysis

BACKGROUND

To study the additive value of radiomics features to the BCLC staging system in clustering HCC patients.

METHODS

A total of 266 patients with HCC were included in this retrospective study. All patients had undergone baseline MR imaging, and 95 radiomics features were extracted from 3D segmentations representative of lesions on the venous phase and apparent diffusion coefficient maps. A random forest algorithm was utilized to extract the most relevant features to transplant-free survival. The selected features were used alongside BCLC staging to construct Kaplan-Meier curves.

RESULTS

Out of 95 extracted features, the three most relevant features were incorporated into random forest classifiers. The Integrated Brier score of the prediction error curve was 0.135, 0.072, and 0.048 for the BCLC, radiomics, and combined models, respectively. The mean area under the receiver operating curve (ROC curve) over time for the three models was 81.1%, 77.3%, and 56.2% for the combined radiomics and BCLC models, respectively.

CONCLUSIONS

Radiomics features outperformed the BCLC staging system in determining prognosis in HCC patients. The addition of a radiomics classifier increased the classification capability of the BCLC model. Texture analysis features could be considered as possible biomarkers in predicting transplant-free survival in HCC patients.

MyCTC chip: microfluidic-based drug screen with patient-derived tumour cells from liquid biopsies

Cancer patients with advanced disease are characterized by intrinsic challenges in predicting drug response patterns, often leading to ineffective treatment. Current clinical practice for treatment decision-making is commonly based on primary or secondary tumour biopsies, yet when disease progression accelerates, tissue biopsies are not performed on a regular basis. It is in this context that liquid biopsies may offer a unique window to uncover key vulnerabilities, providing valuable information about previously underappreciated treatment opportunities. Here, we present MyCTC chip, a novel microfluidic device enabling the isolation, culture and drug susceptibility testing of cancer cells derived from liquid biopsies. Cancer cell capture is achieved through a label-free, antigen-agnostic enrichment method, and it is followed by cultivation in dedicated conditions, allowing on-chip expansion of captured cells. Upon growth, cancer cells are then transferred to drug screen chambers located within the same device, where multiple compounds can be tested simultaneously. We demonstrate MyCTC chip performance by means of spike-in experiments with patient-derived breast circulating tumour cells, enabling >95% capture rates, as well as prospective processing of blood from breast cancer patients and ascites fluid from patients with ovarian, tubal and endometrial cancer, where sensitivity to specific chemotherapeutic agents was identified. Together, we provide evidence that MyCTC chip may be used to identify personalized drug response patterns in patients with advanced metastatic disease and with limited treatment opportunities.

Shear wave elastography imaging in a porcine tendinopathy model

OBJECTIVE

To quantify the effect of structural damage in an ex vivo animal tendinopathy model using shear wave elastography (SWE).

MATERIALS AND METHODS

Sixteen porcine flexor tendons were injected with a 0.05 mL bolus of 1.5% collagenase solution to induce focal structural damage without surfacing tears. Control tendons were injected with saline (n = 16). Eight tendons from each group were incubated at 37 °C for 3.5 h while the remaining 8 from each group were incubated for 7 h. Tendons were mechanically stretched to 0% and 1% strain. Simultaneously, SWE was acquired proximal to, at, and distal to the injection site using a clinical ultrasound scanner.

RESULTS

There were significant differences in SWS (saline > collagenase) at 1% strain and 7-h incubation for all three locations (PROX p = 0.0031, ROI p = 0.001, DIST p = 0.0043). There were also significant differences at 0% strain and 7 h, but only at (p = 0.0005), and distal to (p = 0.0035), the injection site. No statistically significant differences were observed for 3.5-h incubation, at 0% or 1% strain.

CONCLUSIONS

Collagenase-mediated structural damage does appear to convey decreased tissue elasticity on SWE when ex vivo tendons are incubated for 7 h. These findings suggest that SWE may be a useful tool for predicting ultimate tissue strength in tendinopathic tissues. Pull-to-failure testing should be performed in the future and are expected to show that tendons with decreased SWS, and, therefore, decreased elasticity, rupture at lower pulls forces.

Quantitative MRI of chronic pancreatitis: results from a multi-institutional prospective study, magnetic resonance imaging as a non-invasive method for assessment of pancreatic fibrosis (MINIMAP)

PURPOSE

To determine if quantitative MRI techniques can be helpful to evaluate chronic pancreatitis (CP) in a setting of multi-institutional study.

METHODS

This study included a subgroup of participants (n = 101) enrolled in the Prospective Evaluation of Chronic Pancreatitis for Epidemiologic and Translational Studies (PROCEED) study (NCT03099850) from February 2019 to May 2021. MRI was performed on 1.5 T using Siemens and GE scanners at seven clinical centers across the USA. Quantitative MRI parameters of the pancreas included T1 relaxation time, extracellular volume (ECV) fraction, apparent diffusion coefficient (ADC), and fat signal fraction. We report the diagnostic performance and mean values within the control (n = 50) and CP (n = 51) groups. The T1, ECV and fat signal fraction were combined to generate the quantitative MRI score (Q-MRI).

RESULTS

There was significantly higher T1 relaxation time; mean 669 ms (± 171) vs. 593 ms (± 82) (p = 0.006), ECV fraction; 40.2% (± 14.7) vs. 30.3% (± 11.9) (p < 0.001), and pancreatic fat signal fraction; 12.2% (± 5.5) vs. 8.2% (± 4.4) (p < 0.001) in the CP group compared to controls. The ADC was similar between groups (p = 0.45). The AUCs for the T1, ECV, and pancreatic fat signal fraction were 0.62, 0.72, and 0.73, respectively. The composite Q-MRI score improved the diagnostic performance (cross-validated AUC: 0.76).

CONCLUSION

Quantitative MR parameters evaluating the pancreatic parenchyma (T1, ECV fraction, and fat signal fraction) are helpful in the diagnosis of CP. A Q-MRI score that combines these three MR parameters improves diagnostic performance. Further studies are warranted with larger study populations including patients with acute and recurrent acute pancreatitis and longitudinal follow-ups.

Application of a novel T1 retrospective quantification using internal references (T1-REQUIRE) algorithm to derive quantitative T1 relaxation maps of the brain

Most MRI sequences used clinically are qualitative or weighted. While such images provide useful information for clinicians to diagnose and monitor disease progression, they lack the ability to quantify tissue damage for more objective assessment. In this study, an algorithm referred to as the T1-REQUIRE is presented as a proof-of-concept which uses nonlinear transformations to retrospectively estimate T1 relaxation times in the brain using T1-weighted MRIs, the appropriate signal equation, and internal, healthy tissues as references. T1-REQUIRE was applied to two T1-weighted MR sequences, a spin-echo and a MPRAGE, and validated with a reference standard T1 mapping algorithm in vivo. In addition, a multiscanner study was run using MPRAGE images to determine the effectiveness of T1-REQUIRE in conforming the data from different scanners into a more uniform way of analyzing T1-relaxation maps. The T1-REQUIRE algorithm shows good agreement with the reference standard (Lin's concordance correlation coefficients of 0.884 for the spin-echo and 0.838 for the MPRAGE) and with each other (Lin's concordance correlation coefficient of 0.887). The interscanner studies showed improved alignment of cumulative distribution functions after T1-REQUIRE was performed. T1-REQUIRE was validated with a reference standard and shown to be an effective estimate of T1 over a clinically relevant range of T1 values. In addition, T1-REQUIRE showed excellent data conformity across different scanners, providing evidence that T1-REQUIRE could be a useful addition to big data pipelines.

Liver T1 relaxation time of the 'normal liver' in healthy Asians: measurement with MOLLI and B1-corrected VFA methods at 3T

OBJECTIVES

Liver T1 is a potential magnetic resonance imaging biomarker for liver diseases. This study aimed to determine the T1 relaxation time of the normal liver (PDFF<5%) in healthy Asian volunteers using modified look-locker inversion recovery (MOLLI) and B1 inhomogeneity-corrected variable flip angle (B1-corrected VFA).

METHODS

60 healthy Asian volunteers without focal or diffuse liver disease underwent a liver scan at 3T magnetic resonance. Proton density fat fraction (PDFF) and liver stiffness measurements were applied for the quantification of liver fat and fibrosis. T1 mapping was performed with MOLLI and B1-corrected VFA sequences. Bland-Altman, linear regression, Student t-test, and one-way analysis of variance were used for statistical analysis.

RESULTS

The mean T1 relaxation times of the whole liver were 901 ± 34 ms by MOLLI, and 948 ± 29 ms by B1-corrected VFA in healthy volunteers. There was a strong correlation (r = 0.86, p < 0.0001) for liver T1 between two T1 mapping methods. There were significant differences between the right and left lobes in liver T1 relaxation times using both methods (p < 0.05). Gender and Asian ethnic disparities had no impact on liver T1 relaxation times.

CONCLUSION

T1 relaxation times of the normal liver (PDFF<5%) in healthy volunteers were established by MOLLI and B1-corrected VFA T1 mapping methods at 3T. It may provide suitable and robust baseline values for the assessment of liver diseases.

ADVANCES IN KNOWLEDGE

Gender and Asian ethnic disparities do not impact liver T1 relaxation time measurements.

Challenges in ensuring the generalizability of image quantitation methods for MRI

Image quantitation methods including quantitative MRI, multiparametric MRI, and radiomics offer great promise for clinical use. However, many of these methods have limited clinical adoption, in part due to issues of generalizability, that is, the ability to translate methods and models across institutions. Researchers can assess generalizability through measurement of repeatability and reproducibility, thus quantifying different aspects of measurement variance. In this article, we review the challenges to ensuring repeatability and reproducibility of image quantitation methods as well as present strategies to minimize their variance to enable wider clinical implementation. We present possible solutions for achieving clinically acceptable performance of image quantitation methods and briefly discuss the impact of minimizing variance and achieving generalizability towards clinical implementation and adoption.

Automated lung vessel segmentation reveals blood vessel volume redistribution in viral pneumonia

PURPOSE

It is known from histology studies that lung vessels are affected in viral pneumonia. However, their diagnostic potential as a chest CT imaging parameter has only rarely been exploited. The purpose of this study is to develop a robust method for automated lung vessel segmentation and morphology analysis and apply it to a large chest CT dataset.

METHODS

In total, 509 non-enhanced chest CTs (NECTs) and 563 CT pulmonary angiograms (CTPAs) were included. Sub-groups were patients with healthy lungs (group_NORM, n = 634) and those RT-PCR-positive for Influenza A/B (group_INF, n = 159) and SARS-CoV-2 (group_COV, n = 279). A lung vessel segmentation algorithm (LVSA) based on traditional image processing was developed, validated with a point-of-interest approach, and applied to a large clinical dataset. Total blood vessel volume in lung (TBV) and the blood vessel volume percentage (BV%) of three blood vessel size types were calculated and compared between groups: small (BV5%, cross-sectional area < 5 mm²), medium (BV5-10%, 5-10 mm²) and large (BV10%, >10 mm²).

RESULTS

Sensitivity of the LVSA was 84.6% (95 %CI: 73.9-95.3) for NECTs and 92.8% (95 %CI: 90.8-94.7) for CTPAs. In viral pneumonia, besides an increased TBV, the main finding was a significantly decreased BV5% in group_COV (n = 14%) and group_INF (n = 15%) compared to group_NORM (n = 18%) [p < 0.001]. At the same time, BV10% was increased (group_COV n = 15% and group_INF n = 14% vs. group_NORM n = 11%; p < 0.001).

CONCLUSION

In COVID-19 and Influenza, the blood vessel volume is redistributed from small to large vessels in the lung. Automated LSVA allows researchers and clinicians to derive imaging parameters for large amounts of CTs. This can enhance the understanding of vascular changes, particularly in infectious lung diseases.

No-gold-standard evaluation of quantitative imaging methods in the presence of correlated noise

Objective evaluation of quantitative imaging (QI) methods with patient data is highly desirable, but is hindered by the lack or unreliability of an available gold standard. To address this issue, techniques that can evaluate QI methods without access to a gold standard are being actively developed. These techniques assume that the true and measured values are linearly related by a slope, bias, and Gaussian-distributed noise term, where the noise between measurements made by different methods is independent of each other. However, this noise arises in the process of measuring the same quantitative value, and thus can be correlated. To address this limitation, we propose a no-gold-standard evaluation (NGSE) technique that models this correlated noise by a multi-variate Gaussian distribution parameterized by a covariance matrix. We derive a maximum-likelihood-based approach to estimate the parameters that describe the relationship between the true and measured values, without any knowledge of the true values. We then use the estimated slopes and diagonal elements of the covariance matrix to compute the noise-to-slope ratio (NSR) to rank the QI methods on the basis of precision. The proposed NGSE technique was evaluated with multiple numerical experiments. Our results showed that the technique reliably estimated the NSR values and yielded accurate rankings of the considered methods for 83% of 160 trials. In particular, the technique correctly identified the most precise method for ∼ 97% of the trials. Overall, this study demonstrates the efficacy of the NGSE technique to accurately rank different QI methods when correlated noise is present, and without access to any knowledge of the ground truth. The results motivate further validation of this technique with realistic simulation studies and patient data.

Exploring the Interobserver Agreement in Computer-Aided Radiologic Tumor Measurement and Evaluation of Tumor Response

The accurate, objective, and reproducible evaluation of tumor response to therapy is indispensable in clinical trials. This study aimed at investigating the reliability and reproducibility of a computer-aided contouring (CAC) tool in tumor measurements and its impact on evaluation of tumor response in terms of RECIST 1.1 criteria. A total of 200 cancer patients were retrospectively collected in this study, which were randomly divided into two sets of 100 patients for experiential learning and testing. A total of 744 target lesions were identified by a senior radiologist in distinctive body parts, of which 278 lesions were in data set 1 (learning set) and 466 lesions were in data set 2 (testing set). Five image analysts were respectively instructed to measure lesion diameter using manual and CAC tools in data set 1 and subsequently tested in data set 2. The interobserver variability of tumor measurements was validated by using the coefficient of variance (CV), the Pearson correlation coefficient (PCC), and the interobserver correlation coefficient (ICC). We verified that the mean CV of manual measurement remained constant between the learning and testing data sets (0.33 vs. 0.32, p = 0.490), whereas it decreased for the CAC measurements after learning (0.24 vs. 0.19, p < 0.001). The interobserver measurements with good agreement (CV < 0.20) were 29.9% (manual) vs. 49.0% (CAC) in the learning set (p < 0.001) and 30.9% (manual) vs. 64.4% (CAC) in the testing set (p < 0.001). The mean PCCs were 0.56 ± 0.11 mm (manual) vs. 0.69 ± 0.10 mm (CAC) in the learning set (p = 0.013) and 0.73 ± 0.07 mm (manual) vs. 0.84 ± 0.03 mm (CAC) in the testing set (p < 0.001). ICCs were 0.633 (manual) vs. 0.698 (CAC) in the learning set (p < 0.001) and 0.716 (manual) vs. 0.824 (CAC) in the testing set (p < 0.001). The Fleiss’ kappa analysis revealed that the overall agreement was 58.7% (manual) vs. 58.9% (CAC) in the learning set and 62.9% (manual) vs. 74.5% (CAC) in the testing set. The 80% agreement of tumor response evaluation was 55.0% (manual) vs. 66.0% in the learning set and 60.6% (manual) vs. 79.7% (CAC) in the testing set. In conclusion, CAC can reduce the interobserver variability of radiological tumor measurements and thus improve the agreement of imaging evaluation of tumor response.

Optimizing the protocol for retinal vascular permeability mapping from fluorescein videoangiography data

An intact blood-retinal barrier is critical to maintaining the function of the retina. Many diseases of the eye have been directly associated with impairment in vascular permeability, and methods to measure vascular permeability could offer a window into early detection of disease; however, there exist no direct measures of vascular permeability that have be translated to the clinic. This work details a complete clinical workflow to quantify vascular permeability and volumetric blood flow from fluorescein videoangiography data, with validation through realistic simulations. For optimizing the protocol, this study carried on frame rate of fluorescein videoangiography to generate a high-resolution image while minimizing the error.

Radiomics as an emerging tool in the management of brain metastases

Brain metastases (BM) are associated with significant morbidity and mortality in patients with advanced cancer. Despite significant advances in surgical, radiation, and systemic therapy in recent years, the median overall survival of patients with BM is less than 1 year. The acquisition of medical images, such as computed tomography (CT) and magnetic resonance imaging (MRI), is critical for the diagnosis and stratification of patients to appropriate treatments. Radiomic analyses have the potential to improve the standard of care for patients with BM by applying artificial intelligence (AI) with already acquired medical images to predict clinical outcomes and direct the personalized care of BM patients. Herein, we outline the existing literature applying radiomics for the clinical management of BM. This includes predicting patient response to radiotherapy and identifying radiation necrosis, performing virtual biopsies to predict tumor mutation status, and determining the cancer of origin in brain tumors identified via imaging. With further development, radiomics has the potential to aid in BM patient stratification while circumventing the need for invasive tissue sampling, particularly for patients not eligible for surgical resection.

Quantitative multiparametric MRI predicts response to neoadjuvant therapy in the community setting

BACKGROUND

The purpose of this study was to determine whether advanced quantitative magnetic resonance imaging (MRI) can be deployed outside of large, research-oriented academic hospitals and into community care settings to predict eventual pathological complete response (pCR) to neoadjuvant therapy (NAT) in patients with locally advanced breast cancer.

METHODS

Patients with stage II/III breast cancer (N = 28) were enrolled in a multicenter study performed in community radiology settings. Dynamic contrast-enhanced (DCE) and diffusion-weighted (DW)-MRI data were acquired at four time points during the course of NAT. Estimates of the vascular perfusion and permeability, as assessed by the volume transfer rate (K^trans) using the Patlak model, were generated from the DCE-MRI data while estimates of cell density, as assessed by the apparent diffusion coefficient (ADC), were calculated from DW-MRI data. Tumor volume was calculated using semi-automatic segmentation and combined with K^trans and ADC to yield bulk tumor blood flow and cellularity, respectively. The percent change in quantitative parameters at each MRI scan was calculated and compared to pathological response at the time of surgery. The predictive accuracy of each MRI parameter at different time points was quantified using receiver operating characteristic curves.

RESULTS

Tumor size and quantitative MRI parameters were similar at baseline between groups that achieved pCR (n = 8) and those that did not (n = 20). Patients achieving a pCR had a larger decline in volume and cellularity than those who did not achieve pCR after one cycle of NAT (p < 0.05). At the third and fourth MRI, changes in tumor volume, K^trans, ADC, cellularity, and bulk tumor flow from baseline (pre-treatment) were all significantly greater (p < 0.05) in the cohort who achieved pCR compared to those patients with non-pCR.

CONCLUSIONS

Quantitative analysis of DCE-MRI and DW-MRI can be implemented in the community care setting to accurately predict the response of breast cancer to NAT. Dissemination of quantitative MRI into the community setting allows for the incorporation of these parameters into the standard of care and increases the number of clinical community sites able to participate in novel drug trials that require quantitative MRI.

Development, validation, qualification, and dissemination of quantitative MR methods: Overview and recommendations by the ISMRM quantitative MR study group

On behalf of the International Society for Magnetic Resonance in Medicine (ISMRM) Quantitative MR Study Group, this article provides an overview of considerations for the development, validation, qualification, and dissemination of quantitative MR (qMR) methods. This process is framed in terms of two central technical performance properties, i.e., bias and precision. Although qMR is confounded by undesired effects, methods with low bias and high precision can be iteratively developed and validated. For illustration, two distinct qMR methods are discussed throughout the manuscript: quantification of liver proton-density fat fraction, and cardiac T₁ . These examples demonstrate the expansion of qMR methods from research centers toward widespread clinical dissemination. The overall goal of this article is to provide trainees, researchers, and clinicians with essential guidelines for the development and validation of qMR methods, as well as an understanding of necessary steps and potential pitfalls for the dissemination of quantitative MR in research and in the clinic.

Development of a standardized MRI protocol for pancreas assessment in humans

Magnetic resonance imaging (MRI) has detected changes in pancreas volume and other characteristics in type 1 and type 2 diabetes. However, differences in MRI technology and approaches across locations currently limit the incorporation of pancreas imaging into multisite trials. The purpose of this study was to develop a standardized MRI protocol for pancreas imaging and to define the reproducibility of these measurements. Calibrated phantoms with known MRI properties were imaged at five sites with differing MRI hardware and software to develop a harmonized MRI imaging protocol. Subsequently, five healthy volunteers underwent MRI at four sites using the harmonized protocol to assess pancreas size, shape, apparent diffusion coefficient (ADC), longitudinal relaxation time (T1), magnetization transfer ratio (MTR), and pancreas and hepatic fat fraction. Following harmonization, pancreas size, surface area to volume ratio, diffusion, and longitudinal relaxation time were reproducible, with coefficients of variation less than 10%. In contrast, non-standardized image processing led to greater variation in MRI measurements. By using a standardized MRI image acquisition and processing protocol, quantitative MRI of the pancreas performed at multiple locations can be incorporated into clinical trials comparing pancreas imaging measures and metabolic state in individuals with type 1 or type 2 diabetes.

Normative placental structure in pregnancy using quantitative Magnetic Resonance Imaging

INTRODUCTION

To characterize normative morphometric, textural and microstructural placental development by applying advanced and quantitative magnetic resonance imaging (qMRI) techniques to the in-vivo placenta.

METHODS

We enrolled 195 women with uncomplicated, healthy singleton pregnancies in a prospective observational study. Women underwent MRI between 16- and 40-weeks' gestation. Morphometric and textural metrics of placental growth were calculated from T2-weighted (T2W) images, while measures of microstructural development were calculated from diffusion-weighted images (DWI). Normative tables and reference curves were constructed for each measured index across gestation and according to fetal sex.

RESULTS

Data from 269 MRI studies from 169 pregnant women were included in the analyses. During the study period, placentas undergo significant increases in morphometric measures of volume, thickness, and elongation. Placental texture reveals increasing variability with advancing gestation as measured by grey level non uniformity, run length non uniformity and long run high grey level emphasis. Placental microstructure did not vary with gestational age. Placental elongation was the only metric that differed significantly between male and female fetuses.

DISCUSSION

We report quantitative metrics of placental morphometry, texture and microstructure in a large cohort of healthy controls during the second and third trimesters of pregnancy. These measures can serve as normative references of in-vivo placental development to better understand placental function in high-risk conditions and allow for the early detection of placental mal-development.

Validation and Clinical Applicability of Whole-Volume Automated Segmentation of Optical Coherence Tomography in Retinal Disease Using Deep Learning

Question

Is deep learning–based segmentation of macular disease in optical coherence tomography (OCT) suitable for clinical use?

Findings

In this diagnostic study of OCT data from 173 patients with age-related macular degeneration or diabetic macular edema, model segmentations qualitatively ranked better or comparable for clinical applicability to 1 or more expert grader segmentations in 127 scans (73%) by a panel of 3 retinal specialists. Scans with high quantitative accuracy scores were not reliably associated with higher rankings.

Meaning

These findings suggest that qualitative evaluation adds to quantitative approaches when assessing clinical applicability of segmentation tools and clinician satisfaction in practice.

Importance

Quantitative volumetric measures of retinal disease in optical coherence tomography (OCT) scans are infeasible to perform owing to the time required for manual grading. Expert-level deep learning systems for automatic OCT segmentation have recently been developed. However, the potential clinical applicability of these systems is largely unknown.

Objective

To evaluate a deep learning model for whole-volume segmentation of 4 clinically important pathological features and assess clinical applicability.

Design, Setting, Participants

This diagnostic study used OCT data from 173 patients with a total of 15 558 B-scans, treated at Moorfields Eye Hospital. The data set included 2 common OCT devices and 2 macular conditions: wet age-related macular degeneration (107 scans) and diabetic macular edema (66 scans), covering the full range of severity, and from 3 points during treatment. Two expert graders performed pixel-level segmentations of intraretinal fluid, subretinal fluid, subretinal hyperreflective material, and pigment epithelial detachment, including all B-scans in each OCT volume, taking as long as 50 hours per scan. Quantitative evaluation of whole-volume model segmentations was performed. Qualitative evaluation of clinical applicability by 3 retinal experts was also conducted. Data were collected from June 1, 2012, to January 31, 2017, for set 1 and from January 1 to December 31, 2017, for set 2; graded between November 2018 and January 2020; and analyzed from February 2020 to November 2020.

Main Outcomes and Measures

Rating and stack ranking for clinical applicability by retinal specialists, model-grader agreement for voxelwise segmentations, and total volume evaluated using Dice similarity coefficients, Bland-Altman plots, and intraclass correlation coefficients.

Results

Among the 173 patients included in the analysis (92 [53%] women), qualitative assessment found that automated whole-volume segmentation ranked better than or comparable to at least 1 expert grader in 127 scans (73%; 95% CI, 66%-79%). A neutral or positive rating was given to 135 model segmentations (78%; 95% CI, 71%-84%) and 309 expert gradings (2 per scan) (89%; 95% CI, 86%-92%). The model was rated neutrally or positively in 86% to 92% of diabetic macular edema scans and 53% to 87% of age-related macular degeneration scans. Intraclass correlations ranged from 0.33 (95% CI, 0.08-0.96) to 0.96 (95% CI, 0.90-0.99). Dice similarity coefficients ranged from 0.43 (95% CI, 0.29-0.66) to 0.78 (95% CI, 0.57-0.85).

Conclusions and Relevance

This deep learning–based segmentation tool provided clinically useful measures of retinal disease that would otherwise be infeasible to obtain. Qualitative evaluation was additionally important to reveal clinical applicability for both care management and research.

Qualitative and quantitative magnetic resonance imaging evaluation of bone tissue vaso-occlusive events in patients with sickle cell disease

OBJECTIVE

To evaluate the association between bone changes due to vaso-occlusive events in sickle cell disease (SCD) revealed by conventional MRI sequences and the fat fraction obtained using a 6-point DIXON technique (FFdix), in an attempt to use quantitative data as a biomarker for bone complications.

METHODS

Cross-sectional study, with 48 SCD patients, 26-homozygous (HbSS), and 22-compound heterozygous (HbSC). Forty-eight healthy individuals paired by age, weight, and sex with SCD patients. All participants underwent lumbar spine and pelvis MRI. Conventional sequences: bone complications related to vaso-occlusive events-femoral head avascular necrosis, bone infarctions, "H"-shaped vertebrae, bone marrow necrosis. Six-point DIXON technique: quantitative evaluation of the bone marrow at pre-established sites (lumbar vertebrae, sacrum, iliacs, femoral heads, greater femoral trochanters, femoral necks). Pearson's correlation, ROC curve, and binary logistic regression analysis were performed.

RESULTS

The most frequent findings in the SCD group included femoral head avascular necrosis (75%), bone infarctions (58.3%), "H"-shaped vertebrae (58.3%), and typical imaging findings of bone marrow necrosis (8.3%). Cortical bone thickness in the proximal femoral diaphysis in patients with SCD was moderately negatively correlated with FFdix in lumbar vertebrae, iliacs, femoral necks, and first sacral vertebrae. The ROC curves and odds ratios demonstrated excellent performance of FFdix in all the evaluated anatomical sites and identified patients having bone complications.

CONCLUSIONS

FFdix could serve as a potential biomarker in SCD because of its association with bone complications secondary to vaso-occlusive events in patients with SCD, especially in femoral heads, femoral necks, and iliacs.

Artificial Intelligence and Radiomics in Head and Neck Cancer Care: Opportunities, Mechanics, and Challenges

The advent of large-scale high-performance computing has allowed the development of machine-learning techniques in oncologic applications. Among these, there has been substantial growth in radiomics (machine-learning texture analysis of images) and artificial intelligence (which uses deep-learning techniques for "learning algorithms"); however, clinical implementation has yet to be realized at scale. To improve implementation, opportunities, mechanics, and challenges, models of imaging-enabled artificial intelligence approaches need to be understood by clinicians who make the treatment decisions. This article aims to convey the basic conceptual premises of radiomics and artificial intelligence using head and neck cancer as a use case. This educational overview focuses on approaches for head and neck oncology imaging, detailing current research efforts and challenges to implementation.

Quantification of Pancreas Surface Lobularity on CT: A Feasibility Study in the Normal Pancreas

Objective

To assess the feasibility and reproducibility of pancreatic surface lobularity (PSL) quantification derived from abdominal computed tomography (CT) in a population of patients free from pancreatic disease.

Materials and Methods

This retrospective study included 265 patients free from pancreatic disease who underwent contrast-enhanced abdominal CT between 2017 and 2019. A maximum of 11 individual PSL measurements were performed by two abdominal radiologists (head [5 measurements], body, and tail [3 measurements each]) using dedicated software. The influence of age, body mass index (BMI), and sex on PSL was assessed using the Pearson correlation and repeated measurements. Inter-reader agreement was assessed using the intraclass correlation coefficient (ICC) and Bland Altman (BA) plots.

Results

CT images of 15 (6%) patients could not be analyzed. A total of 2750 measurements were performed in the remaining 250 patients (143 male [57%], mean age 45 years [range, 18–91]), and 2237 (81%) values were obtained in the head 951/1250 (76%), body 609/750 (81%), and tail 677/750 (90%). The mean ± standard deviation PSL was 6.53 ± 1.37. The mean PSL was significantly higher in male than in female (6.89 ± 1.30 vs. 6.06 ± 1.31, respectively, p < 0.001). PSL gradually increased with age (r = 0.32, p < 0.001) and BMI (r = 0.32, p < 0.001). Inter-reader agreement was excellent (ICC 0.82 [95% confidence interval 0.72–0.85], with a BA bias of 0.30 and 95% limits of agreement of −1.29 and 1.89).

Conclusion

CT-based PSL quantification is feasible with a high success rate and inter-reader agreement in subjects free from pancreatic disease. Significant variations were observed according to sex, age, and BMI. This study provides a reference for future studies.

Assessment of Renal Cell Carcinoma by Texture Analysis in Clinical Practice: A Six-Site, Six-Platform Analysis of Reliability

Background: Multiple commercial and open-source software applications are available for texture analysis. Nonstandard techniques can cause undesirable variability that impedes result reproducibility and limits clinical utility. Objective: The purpose of this study is to measure agreement of texture metrics extracted by 6 software packages. Methods: This retrospective study included 40 renal cell carcinomas with contrast-enhanced CT from The Cancer Genome Atlas and Imaging Archive. Images were analyzed by 7 readers at 6 sites. Each reader used 1 of 6 software packages to extract commonly studied texture features. Inter and intra-reader agreement for segmentation was assessed with intra-class correlation coefficients. First-order (available in 6 packages) and second-order (available in 3 packages) texture features were compared between software pairs using Pearson correlation. Results: Inter- and intra-reader agreement was excellent (ICC 0.93-1). First-order feature correlations were strong (r>0.8, p<0.001) between 75% (21/28) of software pairs for mean and standard deviation, 48% (10/21) for entropy, 29% (8/28) for skewness, and 25% (7/28) for kurtosis. Of 15 second-order features, only co-occurrence matrix correlation, grey-level non-uniformity, and run-length non-uniformity showed strong correlation between software packages (0.90-1, p<0.001). Conclusion: Variability in first and second order texture features was common across software configurations and produced inconsistent results. Standardized algorithms and reporting methods are needed before texture data can be reliably used for clinical applications. Clinical Impact: It is important to be aware of variability related to texture software processing and configuration when reporting and comparing outputs.

Importance of incorporating quantitative imaging biomarker technical performance characteristics when estimating treatment effects

BACKGROUND/AIMS

Quantitative imaging biomarkers have the potential to detect change in disease early and noninvasively, providing information about the diagnosis and prognosis of a patient, aiding in monitoring disease, and informing when therapy is effective. In clinical trials testing new therapies, there has been a tendency to ignore the variability and bias in quantitative imaging biomarker measurements. Unfortunately, this can lead to underpowered studies and incorrect estimates of the treatment effect. We illustrate the problem when non-constant measurement bias is ignored and show how treatment effect estimates can be corrected.

METHODS

Monte Carlo simulation was used to assess the coverage of 95% confidence intervals for the treatment effect when non-constant bias is ignored versus when the bias is corrected for. Three examples are presented to illustrate the methods: doubling times of lung nodules, rates of change in brain atrophy in progressive multiple sclerosis clinical trials, and changes in proton-density fat fraction in trials for patients with nonalcoholic fatty liver disease.

RESULTS

Incorrectly assuming that the measurement bias is constant leads to 95% confidence intervals for the treatment effect with reduced coverage (<95%); the coverage is especially reduced when the quantitative imaging biomarker measurements have good precision and/or there is a large treatment effect. Estimates of the measurement bias from technical performance validation studies can be used to correct the confidence intervals for the treatment effect.

CONCLUSION

Technical performance validation studies of quantitative imaging biomarkers are needed to supplement clinical trial data to provide unbiased estimates of the treatment effect.

Validation and estimation of spleen volume via computer-assisted segmentation on clinically acquired CT scans

Purpose: Deep learning is a promising technique for spleen segmentation. Our study aims to validate the reproducibility of deep learning-based spleen volume estimation by performing spleen segmentation on clinically acquired computed tomography (CT) scans from patients with myeloproliferative neoplasms. Approach: As approved by the institutional review board, we obtained 138 de-identified abdominal CT scans. A sum of voxel volume on an expert annotator's segmentations establishes the ground truth (estimation 1). We used our deep convolutional neural network (estimation 2) alongside traditional linear estimations (estimation 3 and 4) to estimate spleen volumes independently. Dice coefficient, Hausdorff distance,R 2 coefficient, Pearson R coefficient, the absolute difference in volume, and the relative difference in volume were calculated for 2 to 4 against the ground truth to compare and assess methods' performances. We re-labeled on scan-rescan on a subset of 40 studies to evaluate method reproducibility. Results: Calculated against the ground truth, theR 2 coefficients for our method (estimation 2) and linear method (estimation 3 and 4) are 0.998, 0.954, and 0.973, respectively. The Pearson R coefficients for the estimations against the ground truth are 0.999, 0.963, and 0.978, respectively (paired t -tests produced p < 0.05 between 2 and 3, and 2 and 4). Conclusion: The deep convolutional neural network algorithm shows excellent potential in rendering more precise spleen volume estimations. Our computer-aided segmentation exhibits reasonable improvements in splenic volume estimation accuracy.

Deep convolution neural networks based artifact suppression in under-sampled radial acquisitions of myocardial T 1 mapping images

We developed a deep convolutional neural network (CNN) based method to remove streaking artefact from accelerated radial acquisitions of myocardial T ₁-mapping images. A deep CNN based on a modified U-Net architecture was developed and trained to remove the streaking artefacts from under-sampled T ₁ mapping images. A total of 2090 T ₁-weighted images for 33 patients (55 ± 15 years, 19 males) and five healthy subjects (30 ± 14 years, 2 males) were used for training and testing the network. The images were acquired using radial slice interleaved T ₁ mapping sequence (STONE) and retrospectively under-sampled to achieve acceleration rate of 4 (corresponding to 48 spokes). The dataset was split into training and testing subsets with 23 subjects (60%) and 15 subjects (40%), respectively. For generating voxel-wise T ₁ maps, a two-parameter fitting model was used. Network performance was evaluated using normalized mean square error (NMSE), structural similarity index (SSIM) and peak signal-to-noise ratio (PSNR) metrics. The proposed network allowed fast (<0.3 s/image) removal of the artefact from all T ₁-weighted testing images and the corresponding T ₁ maps with PSNR = 64.3 ± 1.02, NMSE = 0.2 ± 0.09 and SSIM = 0.9 ± 0.3 × 10^-4. There was no statistically significant difference between the measured T ₁ maps for both per-subject (reference: 1085 ± 37 ms, CNN: 1088 ± 37 ms, p = 0.4) and per-segment (reference: 1084 ± 48 ms, CNN: 1083 ± 58 ms, p = 0.9) analyses. In summary, deep CNN allows fast and reliable removal of streaking artefact from under-sampled radial T ₁ mapping images. Our results show that the highly non-linear operations of deep CNN processing of T ₁ mapping images do not impact accurate reconstruction of myocardial T ₁ maps.

A Multiscale Deep Learning Method for Quantitative Visualization of Traumatic Hemoperitoneum at CT: Assessment of Feasibility and Comparison with Subjective Categorical Estimation

PURPOSE

To evaluate the feasibility of a multiscale deep learning algorithm for quantitative visualization and measurement of traumatic hemoperitoneum and to compare diagnostic performance for relevant outcomes with categorical estimation.

MATERIALS AND METHODS

This retrospective, single-institution study included 130 patients (mean age, 38 years; interquartile range, 25-50 years; 79 men) with traumatic hemoperitoneum who underwent CT of the abdomen and pelvis at trauma admission between January 2016 and April 2019. Labeled cases were separated into five combinations of training (80%) and test (20%) sets, and fivefold cross-validation was performed. Dice similarity coefficients (DSCs) were compared with those from a three-dimensional (3D) U-Net and a coarse-to-fine deep learning method. Areas under the receiver operating characteristic curve (AUCs) for a composite outcome, including hemostatic intervention, transfusion, and in-hospital mortality, were compared with consensus categorical assessment by two radiologists. An optimal cutoff was derived by using a radial basis function-based support vector machine.

RESULTS

Mean DSC for the multiscale algorithm was 0.61 ± 0.15 (standard deviation) compared with 0.32 ± 0.16 for the 3D U-Net method and 0.52 ± 0.17 for the coarse-to-fine method (P < .0001). Correlation and agreement between automated and manual volumes were excellent (Pearson r = 0.97, intraclass correlation coefficient = 0.93). The algorithm produced intuitive and explainable visual results. AUCs for automated volume measurement and categorical estimation were 0.86 and 0.77, respectively (P = .004). An optimal cutoff of 278.9 mL yielded accuracy of 84%, sensitivity of 82%, specificity of 93%, positive predictive value of 86%, and negative predictive value of 83%.

CONCLUSION

A multiscale deep learning method for traumatic hemoperitoneum quantitative visualization had improved diagnostic performance for predicting hemorrhage-control interventions and mortality compared with subjective volume estimation. Supplemental material is available for this article. © RSNA, 2020.

Prospects and Challenges of Radiomics by Using Nononcologic Routine Chest CT

Chest CT scans are one of the most common medical imaging procedures. The automatic extraction and quantification of imaging features may help in diagnosis, prognosis of, or treatment decision in cardiovascular, pulmonary, and metabolic diseases. However, an adequate sample size as a statistical necessity for radiomics studies is often difficult to achieve in prospective trials. By exploiting imaging data from clinical routine, a much larger amount of data could be used than in clinical trials. Still, there is only little literature on the implementation of radiomics in clinical routine chest CT scans. Reasons are heterogeneous CT scanning protocols and the resulting technical variability (eg, different slice thicknesses, reconstruction kernels or timings after contrast material administration) in routine CT imaging data. This review summarizes the recent state of the art of studies aiming to develop quantifiable imaging biomarkers at chest CT, such as for osteoporosis, chronic obstructive pulmonary disease, interstitial lung disease, and coronary artery disease. This review explains solutions to overcome heterogeneity in routine data such as the use of imaging repositories, the standardization of radiomic features, algorithmic approaches to improve feature stability, test-retest studies, and the evolution of deep learning for modeling radiomics features. Supplemental material is available for this article. © RSNA, 2020 See also the commentary by Kay in this issue.

Iodine quantification in limited angle tomography

Purpose

To develop and test the feasibility of a two‐pass iterative reconstruction algorithm with material decomposition designed to obtain quantitative iodine measurements in digital breast tomosynthesis.

Methods

Contrast‐enhanced mammography has shown promise as a cost‐effective alternative to magnetic resonance imaging for imaging breast cancer, especially in dense breasts. However, one limitation is the poor quantification of iodine contrast since the true three‐dimensional lesion shape cannot be inferred from the two‐dimensional (2D) projection. Use of limited angle tomography can potentially overcome this limitation by segmenting the iodine map generated by the first‐pass reconstruction using a convolutional neural network, and using this segmentation to restrict the iodine distribution in the second pass of the reconstruction. To evaluate the performance of the algorithms, a set of 2D digital breast phantoms containing targets with varying iodine concentration was used. In each breast phantom, a single simulated lesion with a random size (4 to 8 mm) was placed in a random location within each phantom, with the iodine distribution defined as either homogeneous or rim‐enhanced and blood iodine concentration set between 1.4 and 5.6 mg/mL. Limited angle projection data of these phantoms were simulated for wide and narrow angle geometries, and the proposed reconstruction and segmentation algorithms were applied.

Results

The median Dice similarity coefficient of the segmented masks was 0.975 for the wide angle data and 0.926 for the narrow angle data. Using these segmentations during the second reconstruction pass resulted in an improvement in the concentration estimates (mean estimated‐to‐true concentration ratio, before and after second pass: 48% to 73% for wide angle; 30% to 73% for narrow angle), and a reduction in the coefficient of variation of the estimates (55% to 27% for wide angle; 54% to 35% for narrow angle).

Conclusion

We demonstrate that the proposed two‐pass reconstruction can potentially improve accuracy and precision of iodine quantification in contrast‐enhanced tomosynthesis.

Quantitative Imaging Assessment for Clinical Trials in Oncology

BACKGROUND

Objective radiographic assessment is crucial for accurately evaluating therapeutic efficacy and patient outcomes in oncology clinical trials. Imaging assessment workflow can be complex; can vary with institution; may burden medical oncologists, who are often inadequately trained in radiology and response criteria; and can lead to high interobserver variability and investigator bias. This article reviews the development of a tumor response assessment core (TRAC) at a comprehensive cancer center with the goal of providing standardized, objective, unbiased tumor imaging assessments, and highlights the web-based platform and overall workflow. In addition, quantitative response assessments by the medical oncologists, radiologist, and TRAC are compared in a retrospective cohort of patients to determine concordance.

PATIENTS AND METHODS

The TRAC workflow includes an image analyst who pre-reviews scans before review with a board-certified radiologist and then manually uploads annotated data on the proprietary TRAC web portal. Patients previously enrolled in 10 lung cancer clinical trials between January 2005 and December 2015 were identified, and the prospectively collected quantitative response assessments by the medical oncologists were compared with retrospective analysis of the same dataset by a radiologist and TRAC.

RESULTS

This study enlisted 49 consecutive patients (53% female) with a median age of 60 years (range, 29-78 years); 2 patients did not meet study criteria and were excluded. A linearly weighted kappa test for concordance for TRAC versus radiologist was substantial at 0.65 (95% CI, 0.46-0.85; standard error [SE], 0.10). The kappa value was moderate at 0.42 (95% CI, 0.20-0.64; SE, 0.11) for TRAC versus oncologists and only fair at 0.34 (95% CI, 0.12-0.55; SE, 0.11) for oncologists versus radiologist.

CONCLUSIONS

Medical oncologists burdened with the task of tumor measurements in patients on clinical trials may introduce significant variability and investigator bias, with the potential to affect therapeutic response and clinical trial outcomes. Institutional imaging cores may help bridge the gap by providing unbiased and reproducible measurements and enable a leaner workflow.

A novel magnetic resonance imaging postprocessing technique for the assessment of intervertebral disc degeneration-Correlation with histological grading in a rabbit disc degeneration model

INTRODUCTION

Estimation of intervertebral disc degeneration on magnetic resonance imaging (MRI) is challenging. Qualitative schemes used in clinical practice correlate poorly with pain and quantitative techniques have not entered widespread clinical use.

METHODS

As part of a prior study, 25 New Zealand white rabbits underwent annular puncture to induce disc degeneration in 50 noncontiguous lumbar discs. At 16 weeks, the animals underwent multi-echo T2 MRI scanning and were euthanized. The discs were stained and examined histologically. Quantitative T2 relaxation maps were prepared using the nonlinear least squares method. Decay Variance maps were created using a novel technique of aggregating the deviation in the intensity of each echo signal from the expected intensity based on the previous rate of decay.

RESULTS

Decay Variance maps showed a clear and well demarcated nucleus pulposus with a consistent rate of decay (low Decay Variance) in healthy discs that showed progressively more variable decay (higher Decay Variance) with increasing degeneration. Decay Variance maps required significantly less time to generate (1.0 ± 0.0 second) compared with traditional T2 relaxometry maps (5 (±0.9) to 1788.9 (±116) seconds). Histology scores correlated strongly with Decay Variance scores (r = 0.82, P < .01) and weakly with T2 signal intensity (r = 0.32, P < .01) and quantitative T2 relaxometry (r = 0.39, P < .01). Decay Variance had superior sensitivity and specificity for the detection of degenerate discs when compared to T2 signal intensity or Quantitative T2 mapping.

CONCLUSION

Our results show that using a multi-echo T2 MRI sequence, Decay Variance can quantitatively assess disc degeneration more accurately and with less image-processing time than quantitative T2 relaxometry in a rabbit disc puncture model. The technique is a viable candidate for quantitative assessment of disc degeneration on MRI scans. Further validation on human subjects is needed.

Normal T1 relaxometry and extracellular volume of the pancreas in subjects with no pancreas disease: correlation with age and gender

OBJECTIVE

Determine normal T₁ and extracellular volume (ECV) of the pancreas in subjects with no pancreas disease and correlate with age and gender.

SUBJECTS AND METHODS

We imaged 120 healthy subjects (age range 20-78 years) who are on annual screening with MRI/MRCP for the possibility of pancreatic cancer. Subjects had a predisposition to develop pancreatic cancer, but no history of pancreas disease or acute symptoms. Equal number (n = 60) of subjects were scanned on either 1.5 T or 3 T scanner using dual flip angle spoiled gradient echo technique incorporating fat suppression and correction for B₁ field inhomogeneity. Optimization of imaging parameters was performed using a T₁ phantom. ECV was calculated using pre- and post-contrast T₁ of the pancreas and plasma. Regression analysis and Mann-Whitney tests were used for statistical analysis.

RESULTS

Median T₁ on 1.5 T was 654 ms (IQR 608-700); median T₁ on 3 T was 717 ms (IQR 582-850); median ECV on 1.5 T was 0.28 (IQR 0.21-0.33), and median ECV on 3 T was 0.25 (IQR 0.19-0.28). Age had a mild positive correlation with T₁ (r = 0.24, p = 0.009), but not with ECV (r = 0.06, p = 0.54). T₁ and ECV were similar in both genders (p > 0.05).

CONCLUSION

This study measured the median T₁ and ECV of the pancreas in subjects with no pancreas disease. Pancreas shows longer T₁ relaxation times in older population, whereas extracellular fraction remains unchanged. Median T₁ values were different between two magnet strengths; however, no difference was seen between genders and ECV fractions.

Gastric cancer and image-derived quantitative parameters: Part 2-a critical review of DCE-MRI and 18F-FDG PET/CT findings

Abstract

There is yet no consensus on the application of functional imaging and qualitative image interpretation in the management of gastric cancer. In this second part, we will discuss the role of image-derived quantitative parameters from dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) and ¹⁸F-fluorodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG PET/CT) in gastric cancer, as both techniques have been shown to be promising and useful tools in the clinical decision making of this disease. We will focus on different aspects including aggressiveness assessment, staging and Lauren type discrimination, prognosis prediction and response evaluation. Although both the number of articles and the patients enrolled in the studies were rather small, there is evidence that quantitative parameters from DCE-MRI such as K^trans, V_e, K_ep and AUC could be promising image-derived surrogate parameters for the management of gastric cancer. Data from ¹⁸F-FDG PET/CT studies showed that standardised uptake value (SUV) is significantly associated with the aggressiveness, treatment response and prognosis of this disease. Along with the results from diffusion-weighted MRI and contrast-enhanced multidetector computed tomography presented in Part 1 of this critical review, there are additional image-derived quantitative parameters from DCE-MRI and ¹⁸F-FDG PET/CT that hold promise as effective tools in the diagnostic pathway of gastric cancer.

Key Points

• Quantitative analysis from DCE-MRI and¹⁸F-FDG PET/CT allows the extrapolation of multiple image-derived parameters.

• Data from DCE-MRI (K^trans, V_e, K_ep and AUC) and ¹⁸F-FDG PET/CT (SUV) are non-invasive, quantitative image-derived parameters that hold promise in the evaluation of the aggressiveness, treatment response and prognosis of gastric cancer.

Pancreatic cancer detection and characterization-state of the art cross-sectional imaging and imaging data analysis

Pancreatic ductal adenocarcinoma (PDAC) represents a deadly disease, prognosticated to become the 2nd most common cause of cancer related death in the western world by 2030. State of the art radiologic high-resolution cross-sectional imaging by computed tomography (CT) and magnetic resonance imaging (MRI) represent advanced techniques for early lesion detection, pre-therapeutic patient staging and therapy response monitoring. In light of molecular taxonomies currently under development, the implementation of advanced imaging data post-processing pipelines and the integration of imaging and clinical data for the development of risk assessment and clinical decision support tools are required. This review will present the current state of cross-sectional radiologic imaging and image post-processing related to PDAC.

Acceleration of spleen segmentation with end-to-end deep learning method and automated pipeline

Delineation of Computed Tomography (CT) abdominal anatomical structure, specifically spleen segmentation, is useful for not only measuring tissue volume and biomarkers but also for monitoring interventions. Recently, segmentation algorithms using deep learning have been widely used to reduce time humans spend to label CT data. However, the computerized segmentation has two major difficulties: managing intermediate results (e.g., resampled scans, 2D sliced image for deep learning), and setting up the system environments and packages for autonomous execution. To overcome these issues, we propose an automated pipeline for the abdominal spleen segmentation. This pipeline provides an end-to-end synthesized process that allows users to avoid installing any packages and to deal with the intermediate results locally. The pipeline has three major stages: pre-processing of input data, segmentation of spleen using deep learning, 3D reconstruction with the generated labels by matching the segmentation results with the original image dimensions, which can then be used later and for display or demonstration. Given the same volume scan, the approach described here takes about 50 s on average whereas the manual segmentation takes about 30 min on the average. Even if it includes all subsidiary processes such as preprocessing and necessary setups, the whole pipeline process requires on the average 20 min from beginning to end.

Repeatability of dynamic contrast enhanced vp parameter in healthy subjects and patients with brain tumors

PURPOSE

To study the repeatability of plasma volume (v_p) extracted from dynamic-contrast-enhanced (DCE) MRI in order to define threshold values for significant longitudinal changes, and to assess changes in patients with high-grade-glioma (HGG).

METHODS

Twenty eight healthy subjects, of which eleven scanned twice, were used to assess the repeatability of v_p within the normal-appearing brain tissue and to define threshold values for significant changes based on least-detected-differences (LDD) of mean v_p values and histogram comparisons using earth-mover's-distance (EMD). Sixteen patients with HGG were scanned longitudinally with eight patients scanned before and following bevacizumab therapy. Longitudinal changes were assessed based on defined threshold values in comparison to RANO criteria.

RESULTS

The threshold values for significant changes were: LDD = 0.0024 (ml/100 ml, 21%) for mean v_p and EMD = 4.14. In patients, in 20/24 comparisons, no significant longitudinal changes were detected for v_p within the normal-appearing brain tissue. Concurring results were obtained between changes in lesion volume (RANO criteria) and LDD or EMD values in cases diagnosed with progressive-disease, yet in about 50% of cases diagnosed with partial-response preliminary results demonstrated significant increase in v_p despite significant reductions in lesion volume. In two patients, these changes preceded progression detected at follow-up scans. In general, a good concordance was obtained between LDD and EMD.

CONCLUSION

This study shows high repeatability of v_p and provides threshold values for significant changes in longitudinal assessment of patients with brain tumors. Preliminary results suggest the use of v_p-DCE parameter to improve assessment of therapy response in patients with high-grade-glioma.