Non-destructive classification of unlabeled cells: Combining an automated benchtop magnetic resonance scanner and artificial intelligence.

In order to treat degenerative diseases, the importance of advanced therapy medicinal products has increased in recent years. The newly developed treatment strategies require a rethinking of the appropriate analytical methods. Current standards are missing the complete and sterile analysis of the product of interest to make the drug manufacturing effort worthwhile. They only consider partial areas of the sample or product while also irreversibly damaging the investigated specimen. Two-dimensional T1 / T2 MR relaxometry meets these requirements and is therefore a promising in-process control during the manufacturing and classification process of cell-based treatments. In this study a tabletop MR scanner was used to perform two-dimensional MR relaxometry. Throughput was increased by developing an automation platform based on a low-cost robotic arm, resulting in the acquisition of a large dataset of cell-based measurements. Two-dimensional inverse Laplace transformation was used for post-processing, followed by data classification performed with support vector machines (SVM) as well as optimized artificial neural networks (ANN). The trained networks were able to distinguish non-differentiated from differentiated MSCs with a prediction accuracy of 85%. To increase versatility, an ANN was trained on 354 independent, biological replicates distributed across ten different cell lines, resulting in a prediction accuracy of up to 98% depending on data composition. The present study provides a proof of principle for the application of T1 / T2 relaxometry as a non-destructive cell classification method. It does not require labeling of cells and can perform whole mount analysis of each sample. Since all measurements can be performed under sterile conditions, it can be used as an in-process control for cellular differentiation. This distinguishes it from other characterization techniques, as most are destructive or require some type of cell labeling. These advantages highlight the technique's potential for preclinical screening of patient-specific cell-based transplants and drugs.

Myocardial T1: quantification by using an ECG-triggered radial single-shot inversion-recovery MR imaging sequence.

PURPOSE: To develop and validate a fast cardiac magnetic resonance imaging T1 mapping technique with high spatial resolution based on a radial inversion-recovery (IR) spoiled gradient-echo acquisition. MATERIALS AND METHODS: Approval for the study was granted by the local institutional review board, and all subjects gave written informed consent. An electrocardiographically triggered radial single-shot IR (TRASSI) sequence was developed in conjunction with a custom-written fitting algorithm. The proposed imaging technique was validated in phantom measurements and then used for cardiac T1 mapping in 62 subjects with or without cardiac disease. The study population included 51 healthy subjects, three patients with arrhythmia, and eight patients with myocardial infarction. The potential heart rate dependency of the TRASSI method was tested by using linear regression analysis. Statistically significant differences between the sexes and various section orientations were analyzed with a Student t test for independent groups and a repeated-measures analysis of variance for dependent groups. RESULTS: High-spatial-resolution T1 maps (1.17 × 1.17 mm) without motion artifacts and without heart rate dependency (slope = -0.0303, R(2) = 0.0000887, P = .899) were acquired with an acquisition time of less than 6 seconds in all subjects. The mean T1 of healthy left ventricular myocardium across all examined subjects was 1031 msec ± 33 (standard deviation). Testing for reproducibility in three individuals with 34 repetitive measurements revealed a mean standard deviation of 4.1 msec (0.412%). Subacute and chronic myocardial infarction could be detected in all eight patients. T1 disturbances due to arrhythmia proved to be minimal in three patients (standard deviation, <1.2%). CONCLUSION: Fast and accurate cardiac T1 mapping is feasible within a single-shot IR experiment.

Correlating quantitative MR measurements of standardized tumor lines with histological parameters and tumor control dose.

PURPOSE: To correlate non-invasively acquired radiobiologically relevant magnetic resonance (MR) parameters with functional histology and tumor control doses (TCD(50)). MATERIALS AND METHODS: The MR parameters relative perfusion, re-oxygenation and lactate (Lac) concentration from eight human xenograft squamous tumor lines were compared with the histologically acquired pimonidazole hypoxic fraction, the perfused vessel area and TCD(50). RESULTS: Good spatial correlation in the parameter maps could be observed between the pimonidazole staining and tumor regions, which can be reoxygenated when breathing carbogen. A strong positive correlation (R=0.74) was found between whole tumor pimonidazole hypoxic fraction and re-oxygenation, as one would expect. A good correlation was also observed between Lac concentration and re-oxygenation (R=0.71) and between TCD(50) and re-oxygenation (R=0.64), whereas Lac and TCD(50) showed a moderate relation (R=0.44). The in vivo measurement of relative perfusion could be validated to reflect the perfused vessel area (R=0.63). No correlation was detected between perfusion and re-oxygenation or TCD(50). CONCLUSIONS: Lac and re-oxygenation were shown to be pretreatment predictive markers independent from the pathophysiological changes induced during a fractionated course of radiotherapy. These parameters hold promise to be acquired non-invasively with results just a few minutes after measurement and to tailor radiotherapy to individual patterns of a tumor microenvironment.

Sensitive J-coupled metabolite mapping using Sel-MQC with selective multi-spin-echo readout.

The selective multiple quantum coherence technique is combined with a read gradient to accelerate the measurement of a specific scalar-coupled metabolite. The sensitivities of the localization using pure phase encoding and localization with the read gradient are compared in experiments at high magnetic field strength (17.6 T). Multiple spin-echoes of the selective multiple quantum coherence edited metabolite are acquired using frequency-selective refocusing of the specified molecule group. The frequency-selective refocusing does not affect the J-modulation of a coupled spin system, and the echo time is not limited to a multiple of 1/J to acquire pure in-phase or antiphase signal. The multiple echoes can be used to accelerate the metabolite imaging experiment or to measure the apparent transverse relaxation T(2). A simple phase-shifting scheme is presented, which enables the suppression of editing artifacts resulting from the multiple spin-echoes of the water resonance. The experiments are carried out on phantoms, in which lactate and polyunsaturated fatty acids are edited, and in vivo on tumors, in which lactate content and T(2) are imaged. The method is of particular interest when a fast and sensitive selective multiple quantum coherence editing is necessary, e.g., for spatial three dimensional experiments.

In vivo measurement of local aortic pulse-wave velocity in mice with MR microscopy at 17.6 Tesla.

Transgenic mouse models of human diseases have gained increasing importance in the pathophysiology of cardiovascular diseases (CVD). As an indirect measure of vascular stiffness, aortic pulse-wave velocity (PWV) is an important predictor of cardiovascular risk. This study presents an MRI approach that uses a flow area method to estimate local aortic pulse-wave velocity at different sites in the murine aorta. By simultaneously measuring the cross-sectional area and the through-plane velocity with a phase-contrast CINE method, it was possible to measure average values for the PWV in the ascending and descending aorta within the range of 2.4-4.3 m/s for C57BL/6J mice (ages 2 and 8 months) and apoE-knockout mice (age 8 months). Statistically significant differences of the mean values of the PWV of both groups could be determined. By repeating CINE measurements with a time delay of 1 ms between two subsequent data sets, an effective temporal resolution of 1000 frames/s (fps) could be achieved.

Short-echo spectroscopic imaging combined with lactate editing in a single scan.

A short-echo spectroscopic imaging sequence extended with a frequency-selective multiple-quantum- coherence technique (Sel-MQC) is presented. The method enables acquisition of a complete water-suppressed proton spectrum with a short echo time and filtering of the J-coupling metabolite, lactate, from co-resonant lipids in one scan. The purpose of the study was to validate this combined pulse sequence in vitro and in vivo. Measurements on phantoms confirmed the feasibility of the method, and, for a practical in vivo application, experiments were carried out on eight tumors from two different tumor models [UT-SCC-8 (n = 4) and SAS (n = 4)]. T(1)- and T(2)-weighted metabolite and lipid ratios were calculated, and the tumors showed different values in the central and outer regions. The ratio of the lipid methylene peak area (1.30 ppm) to choline peak area (3.20 ppm) was significantly (p < 0.01) different in the central tumor area between the two models, and lactate was detected in only three out of four tumors in the SAS tumor line. The present approach of combining short-echo spectroscopic imaging and lactate editing allows the characterization of tumor-specific metabolites such as choline, lipid methylene and methyl resonances as well as lactate in a single scan.

Lung MRI using an MR-compatible active breathing control (MR-ABC).

This work introduces an MR-compatible active breathing control device (MR-ABC) that can be applied to lung imaging. An MR-ABC consists of a pneumotachograph for respiratory monitoring and an airway-sealing unit. Using an MR-ABC, the subjects were forced to suspend breathing for short time intervals, which were used in turn for data acquisition. While the breathing flow was stopped, data acquisition was triggered by ECG to achieve simultaneous cardiac and respiratory synchronization and thus avoid artifacts from blood flow or heart movement. The flow stoppage allowed a prolonged acquisition window of up to 1.5 sec. To evaluate the potential of an MR-ABC for segmented k-space acquisition, diaphragm displacement was investigated in five volunteers and compared with images acquired using breath-holding, a respiratory belt, and free breathing. Respiratory movement was comparatively low using the breath-hold approach, a respiratory belt or an MR-ABC. During free-breathing diaphragm displacement was comparatively large. To demonstrate the potential of an MR-ABC, lung MRI was performed using whole-chest 3D gradient-echo imaging, multislice turbo spin-echo (TSE) imaging, and short tau inversion recovery TSE (STIR-TSE). Cardiorespiratory synchronization was used for each sequence. None of the volunteers reported any discomfort or inconvenience when using an MR-ABC. Flow stoppage of up to 2.5 sec per breathing cycle was well tolerated, therefore allowing for a reduction of the total imaging time as compared to usage of a respiratory belt or MR navigator.

In vivo quantitative three-dimensional motion mapping of the murine myocardium with PC-MRI at 17.6 T.

This work presents a method that allows for the assessment of 3D murine myocardial motion in vivo at microscopic resolution. Phase-contrast (PC) magnetic resonance imaging (MRI) at 17.6 T was applied to map myocardial motion in healthy mice along three gradient directions. High-resolution velocity maps were acquired at three different levels in the murine myocardium with an in-plane resolution of 98 mum, a slice thickness of 0.6 mm, and a temporal resolution of 6 ms. The applied PC-MRI method was validated with phantom experiments that confirmed the correctness of the method with deviations of <1.7%. Myocardial in-plane velocities between 0.5 cm/s and 2.2 cm/s were determined for the healthy murine myocardium. Through-plane velocities of 0.1-0.83 cm/s were measured. Velocity data was also used to calculate the myocardial twist angle during systole at different slices in the short-axis view.