Apparent Diffusion Coefficient of Normal Abdominal Organs and Bone Marrow From Whole-Body DWI at 1.5 T: The Effect of Sex and Age.

OBJECTIVE: The objectives of this study were to define the range of apparent diffusion coefficients (ADCs) from whole-body DWI in normal abdominal organs and bone marrow, to identify ADC differences between sexes and changes occurring with age, and to evaluate the effect of the fat fraction (FF) on the ADC of normal liver parenchyma and bone marrow. MATERIALS AND METHODS: Fifty-one healthy volunteers (mean age = 38 years; age range = 23-68 years) underwent whole-body DWI using single-shot echo-planar imaging (b = 0, 150, 400, 750, and 1000 s/mm(2)). A two-point Dixon technique was used to evaluate the FF. Perfusion-sensitive ADCs, which we refer to as "ADCALL," and perfusion-insensitive ADCs, which we refer to as "ADCHIGH," of the liver and renal parenchyma, spleen, pancreatic tail, and red and yellow bone marrow were calculated. The relationships between ADC and sex, age, and FF were examined. RESULTS: ADCALL and ADCHIGH were significantly higher in female volunteers for the pancreatic tail (p = 0.046 and 0.008, respectively), red bone marrow (p = 0.029 and 0.001), and yellow bone marrow (p < 0.001 for both) but with considerable overlap. There were significant negative correlations between ADCALL and ADCHIGH and age in the liver parenchyma (p = 0.008 and 0.01, respectively) and in the yellow bone marrow (p = 0.013 and 0.039) for all subjects. ADCALL and ADCHIGH were also negatively correlated with FF in the liver parenchyma (p = 0.006 and 0.008, respectively) and in yellow bone marrow (p < 0.001 and p = 0.001) in all subjects. CONCLUSION: The ADCs of normal liver parenchyma and bone marrow change significantly with age. The ADCs of bone marrow in women are significantly higher than those of men and correlate strongly with FF. These effects may have an impact on image interpretation when using whole-body DWI to assess disease burden and treatment response.

Comparison between diffusion-weighted MRI (DW-MRI) at 1.5 and 3 tesla: a phantom study.

PURPOSE: To compare DW-MRI between 1.5 and 3 Tesla (T) in terms of image quality, apparent diffusion coefficient (ADC), reproducibility, lesion-to-background contrast and signal-to-noise ratio (SNR), using a test object. MATERIALS AND METHODS: A spherical diffusion phantom was used for qualitatively assessing image quality and performing quantitative measurements between the two field strengths. RESULTS: Distortions and signal losses degraded image quality at 3T even when the protocols were optimized for minimum TE. The ADC, in the majority of the phantom compartments, was significantly different between 1.5T and 3T (P < 0.009), while the average coefficient of variation, excluding the phantom compartments affected by artifacts, was <1.3% at both field strengths. The lesion-to-background contrast was improved at 1.5T for images acquired with b = 1000 s/mm(2) and comparable contrast was achieved at 3T with higher b-values. The SNR gain at 3T could, in theory, be balanced by the increased number of signal excitations one can accommodate at 1.5T to perform DW-MRI within the same acquisition time and possibly improved image quality, when 3T systems with no parallel transmission are used. CONCLUSION: Further phantom and in vivo studies are required to investigate the utility of DW-MRI at 3T, if image quality and acquisition times comparable to the ones from 1.5T are assumed.

CXCR4-targeted and MMP-responsive iron oxide nanoparticles for enhanced magnetic resonance imaging.

MRI offers high spatial resolution with excellent tissue penetration but it has limited sensitivity and the commonly administered contrast agents lack specificity. In this study, two sets of iron oxide nanoparticles (IONPs) were synthesized that were designed to selectively undergo copper-free click conjugation upon sensing of matrix metalloproteinase (MMP) enzymes, thereby leading to a self-assembled superparamagnetic nanocluster network with T2 signal enhancement properties. For this purpose, IONPs with bioorthogonal azide and alkyne surfaces masked by polyethylene glycol (PEG) layers tethered to CXCR4-targeted peptide ligands were synthesized and characterized. The IONPs were tested in vitro and T2 signal enhancements of around 160 % were measured when the IONPs were incubated with cells expressing MMP2/9 and CXCR4. Simultaneous systemic administration of the bioorthogonal IONPs in tumor-bearing mice demonstrated the signal-enhancing ability of these 'smart' self-assembling nanomaterials.

A phantom for diffusion-weighted MRI (DW-MRI).

PURPOSE: To develop tissue-equivalent diffusivity materials and build a spherical diffusion phantom which mimics the conditions typically found in biological tissues. Also, to assess the reproducibility of ADC measurements from a whole-body diffusion protocol. MATERIALS AND METHODS: Nickel-doped agarose/sucrose gels were manufactured and used to build a spherical diffusion phantom with tissue-equivalent relaxation and diffusion compartments. The temporal stability of the gels was monitored for a period of 8 weeks and, using the same measurements, the reproducibility of ADC was assessed in a 1.5 Tesla (T) clinical system. RESULTS: The temporal stability of the nickel-doped agarose/sucrose gels diffusion properties was excellent (average coefficient of variation [CV] for ADC in all phantom compartments = 1.27%). The average CV for ADC measurements, excluding the phantom compartments affected by artifacts, was 0.76% showing that the reproducibility of ADC measurements using an EPI DW-MRI protocol is very good. CONCLUSION: Nickel-doped agarose/sucrose gels can be used as reference materials for MRI diffusion measurements and show excellent short-term stability with respect to ADC. A phantom made of these materials can be invaluable in optimizing DW-MRI protocols, developing novel pulse sequences for DW-MRI, or comparing ADC values among field strengths, vendors, and imaging centers.

Development and Evaluation of Machine Learning in Whole-Body Magnetic Resonance Imaging for Detecting Metastases in Patients With Lung or Colon Cancer: A Diagnostic Test Accuracy Study.

OBJECTIVES: Whole-body magnetic resonance imaging (WB-MRI) has been demonstrated to be efficient and cost-effective for cancer staging. The study aim was to develop a machine learning (ML) algorithm to improve radiologists' sensitivity and specificity for metastasis detection and reduce reading times. MATERIALS AND METHODS: A retrospective analysis of 438 prospectively collected WB-MRI scans from multicenter Streamline studies (February 2013-September 2016) was undertaken. Disease sites were manually labeled using Streamline reference standard. Whole-body MRI scans were randomly allocated to training and testing sets. A model for malignant lesion detection was developed based on convolutional neural networks and a 2-stage training strategy. The final algorithm generated lesion probability heat maps. Using a concurrent reader paradigm, 25 radiologists (18 experienced, 7 inexperienced in WB-/MRI) were randomly allocated WB-MRI scans with or without ML support to detect malignant lesions over 2 or 3 reading rounds. Reads were undertaken in the setting of a diagnostic radiology reading room between November 2019 and March 2020. Reading times were recorded by a scribe. Prespecified analysis included sensitivity, specificity, interobserver agreement, and reading time of radiology readers to detect metastases with or without ML support. Reader performance for detection of the primary tumor was also evaluated. RESULTS: Four hundred thirty-three evaluable WB-MRI scans were allocated to algorithm training (245) or radiology testing (50 patients with metastases, from primary 117 colon [n = 117] or lung [n = 71] cancer). Among a total 562 reads by experienced radiologists over 2 reading rounds, per-patient specificity was 86.2% (ML) and 87.7% (non-ML) (-1.5% difference; 95% confidence interval [CI], -6.4%, 3.5%; P = 0.39). Sensitivity was 66.0% (ML) and 70.0% (non-ML) (-4.0% difference; 95% CI, -13.5%, 5.5%; P = 0.344). Among 161 reads by inexperienced readers, per-patient specificity in both groups was 76.3% (0% difference; 95% CI, -15.0%, 15.0%; P = 0.613), with sensitivity of 73.3% (ML) and 60.0% (non-ML) (13.3% difference; 95% CI, -7.9%, 34.5%; P = 0.313). Per-site specificity was high (>90%) for all metastatic sites and experience levels. There was high sensitivity for the detection of primary tumors (lung cancer detection rate of 98.6% with and without ML [0.0% difference; 95% CI, -2.0%, 2.0%; P = 1.00], colon cancer detection rate of 89.0% with and 90.6% without ML [-1.7% difference; 95% CI, -5.6%, 2.2%; P = 0.65]). When combining all reads from rounds 1 and 2, reading times fell by 6.2% (95% CI, -22.8%, 10.0%) when using ML. Round 2 read-times fell by 32% (95% CI, 20.8%, 42.8%) compared with round 1. Within round 2, there was a significant decrease in read-time when using ML support, estimated as 286 seconds (or 11%) quicker ( P = 0.0281), using regression analysis to account for reader experience, read round, and tumor type. Interobserver variance suggests moderate agreement, Cohen κ = 0.64; 95% CI, 0.47, 0.81 (with ML), and Cohen κ = 0.66; 95% CI, 0.47, 0.81 (without ML). CONCLUSIONS: There was no evidence of a significant difference in per-patient sensitivity and specificity for detecting metastases or the primary tumor using concurrent ML compared with standard WB-MRI. Radiology read-times with or without ML support fell for round 2 reads compared with round 1, suggesting that readers familiarized themselves with the study reading method. During the second reading round, there was a significant reduction in reading time when using ML support.

Stripline resonator and preamplifier for preclinical magnetic resonance imaging at 4.7 T.

OBJECT: To design and evaluate a fully shielded, λ/4 stripline resonator as a receive-only surface coil for preclinical MRI at 4.7 T. MATERIALS AND METHODS: A 20 mm diameter stripline surface coil was fabricated from double-sided Duroid 5880 PCB material and was directly coupled to the input of a MOSFET preamplifier, without requiring a matching network. The new coil was compared with a conventional 20 mm, wire loop, receive-only surface coil in imaging experiments with a separate transmit-only saddle coil. RESULTS: The stripline surface coil exhibits a loaded Q-factor of 132 at 200 MHz, compared to 138 for a conventional wire loop coil and its resonant frequency drops by 0.2 MHz under loading, rather than 0.5 MHz for the wire loop. The stripline coil displays a more symmetrical B1 map compared to the wire loop, but the SNR falls off more rapidly with depth so it is 30% poorer 8 mm from the coil plane. It should be possible, however, to reduce this difference by using a thicker dielectric in future versions of the stripline coil. CONCLUSION: Compared to a conventional surface coil, the stripline coil is easy to manufacture, requires shorter set-up times and shows reduced dielectric interaction with conductive samples.

Fully automatic, multiorgan segmentation in normal whole body magnetic resonance imaging (MRI), using classification forests (CFs), convolutional neural networks (CNNs), and a multi-atlas (MA) approach.

PURPOSE: As part of a program to implement automatic lesion detection methods for whole body magnetic resonance imaging (MRI) in oncology, we have developed, evaluated, and compared three algorithms for fully automatic, multiorgan segmentation in healthy volunteers. METHODS: The first algorithm is based on classification forests (CFs), the second is based on 3D convolutional neural networks (CNNs) and the third algorithm is based on a multi-atlas (MA) approach. We examined data from 51 healthy volunteers, scanned prospectively with a standardized, multiparametric whole body MRI protocol at 1.5 T. The study was approved by the local ethics committee and written consent was obtained from the participants. MRI data were used as input data to the algorithms, while training was based on manual annotation of the anatomies of interest by clinical MRI experts. Fivefold cross-validation experiments were run on 34 artifact-free subjects. We report three overlap and three surface distance metrics to evaluate the agreement between the automatic and manual segmentations, namely the dice similarity coefficient (DSC), recall (RE), precision (PR), average surface distance (ASD), root-mean-square surface distance (RMSSD), and Hausdorff distance (HD). Analysis of variances was used to compare pooled label metrics between the three algorithms and the DSC on a 'per-organ' basis. A Mann-Whitney U test was used to compare the pooled metrics between CFs and CNNs and the DSC on a 'per-organ' basis, when using different imaging combinations as input for training. RESULTS: All three algorithms resulted in robust segmenters that were effectively trained using a relatively small number of datasets, an important consideration in the clinical setting. Mean overlap metrics for all the segmented structures were: CFs: DSC = 0.70 ± 0.18, RE = 0.73 ± 0.18, PR = 0.71 ± 0.14, CNNs: DSC = 0.81 ± 0.13, RE = 0.83 ± 0.14, PR = 0.82 ± 0.10, MA: DSC = 0.71 ± 0.22, RE = 0.70 ± 0.34, PR = 0.77 ± 0.15. Mean surface distance metrics for all the segmented structures were: CFs: ASD = 13.5 ± 11.3 mm, RMSSD = 34.6 ± 37.6 mm and HD = 185.7 ± 194.0 mm, CNNs; ASD = 5.48 ± 4.84 mm, RMSSD = 17.0 ± 13.3 mm and HD = 199.0 ± 101.2 mm, MA: ASD = 4.22 ± 2.42 mm, RMSSD = 6.13 ± 2.55 mm, and HD = 38.9 ± 28.9 mm. The pooled performance of CFs improved when all imaging combinations (T2w + T1w + DWI) were used as input, while the performance of CNNs deteriorated, but in neither case, significantly. CNNs with T2w images as input, performed significantly better than CFs with all imaging combinations as input for all anatomical labels, except for the bladder. CONCLUSIONS: Three state-of-the-art algorithms were developed and used to automatically segment major organs and bones in whole body MRI; good agreement to manual segmentations performed by clinical MRI experts was observed. CNNs perform favorably, when using T2w volumes as input. Using multimodal MRI data as input to CNNs did not improve the segmentation performance.

Reverse Classification Accuracy: Predicting Segmentation Performance in the Absence of Ground Truth.

When integrating computational tools, such as automatic segmentation, into clinical practice, it is of utmost importance to be able to assess the level of accuracy on new data and, in particular, to detect when an automatic method fails. However, this is difficult to achieve due to the absence of ground truth. Segmentation accuracy on clinical data might be different from what is found through cross validation, because validation data are often used during incremental method development, which can lead to overfitting and unrealistic performance expectations. Before deployment, performance is quantified using different metrics, for which the predicted segmentation is compared with a reference segmentation, often obtained manually by an expert. But little is known about the real performance after deployment when a reference is unavailable. In this paper, we introduce the concept of reverse classification accuracy (RCA) as a framework for predicting the performance of a segmentation method on new data. In RCA, we take the predicted segmentation from a new image to train a reverse classifier, which is evaluated on a set of reference images with available ground truth. The hypothesis is that if the predicted segmentation is of good quality, then the reverse classifier will perform well on at least some of the reference images. We validate our approach on multi-organ segmentation with different classifiers and segmentation methods. Our results indicate that it is indeed possible to predict the quality of individual segmentations, in the absence of ground truth. Thus, RCA is ideal for integration into automatic processing pipelines in clinical routine and as a part of large-scale image analysis studies.

RGD-targeted MnO nanoparticles as T1 contrast agents for cancer imaging - the effect of PEG length in vivo.

As magnetic resonance imaging (MRI) contrast agents, T1 Gd3+ chelates are generally the preferred option for radiologists over T2 iron oxide nanoparticles. The main reason for the popularity of T1 agents is the easier interpretation of T1-weighted MR images. However, the chemical versatility of nanoparticulate platforms makes them ideal candidates for the next generation of targeted MRI contrast agents. In this context, we present herein the design and preparation of a nanoparticulate contrast agent based on MnO, which presents T1 contrast enhancement properties as well as nanoparticle formulation. Functionalization of MnO nanoparticles with the extensively studied RGD peptide was used to target tumours over-expressing the αvß3 integrin. PEG (polyethylene glycol) molecules were used to increase the blood half-life of the nanoparticles in vivo, and the effect of different PEG lengths on the final contrast on MR images was investigated.

Temporal and spatial evolution of therapy-induced tumor apoptosis detected by caspase-3-selective molecular imaging.

PURPOSE: Induction of apoptosis in tumors is considered a desired goal of anticancer therapy. We investigated whether the dynamic temporal and spatial evolution of apoptosis in response to cytotoxic and mechanism-based therapeutics could be detected noninvasively by the caspase-3 radiotracer [(18)F]ICMT-11 and positron emission tomography (PET). EXPERIMENTAL DESIGN: The effects of a single dose of the alkylating agent cyclophosphamide (CPA or 4-hydroperoxycyclophosphamide), or the mechanism-based small molecule SMAC mimetic birinapant on caspase-3 activation was assessed in vitro and by [(18)F]ICMT-11-PET in mice bearing 38C13 B-cell lymphoma, HCT116 colon carcinoma, or MDA-MB-231 breast adenocarcinoma tumors. Ex vivo analysis of caspase-3 was compared to the in vivo PET imaging data. RESULTS: Drug treatment increased the mean [(18)F]ICMT-11 tumor uptake with a peak at 24 hours for CPA (40 mg/kg; AUC40-60: 8.04 ± 1.33 and 16.05 ± 3.35 %ID/mL × min at baseline and 24 hours, respectively) and 6 hours for birinapant (15 mg/kg; AUC40-60: 20.29 ± 0.82 and 31.07 ± 5.66 %ID/mL × min, at baseline and 6 hours, respectively). Voxel-based spatiotemporal analysis of tumor-intrinsic heterogeneity suggested that discrete pockets of caspase-3 activation could be detected by [(18)F]ICMT-11. Increased tumor [(18)F]ICMT-11 uptake was associated with caspase-3 activation measured ex vivo, and early radiotracer uptake predicted apoptosis, distinct from the glucose metabolism with [(18)F]fluorodeoxyglucose-PET, which depicted continuous loss of cell viability. CONCLUSION: The proapoptotic effects of CPA and birinapant resulted in a time-dependent increase in [(18)F]ICMT-11 uptake detected by PET. [(18)F]ICMT-11-PET holds promise as a noninvasive pharmacodynamic biomarker of caspase-3-associated apoptosis in tumors.