Oxygen-enhanced MRI and radiotherapy in patients with oropharyngeal squamous cell carcinoma.

Background and purpose: This study aimed to assess the role of T1 mapping and oxygen-enhanced MRI in patients undergoing radical dose radiotherapy for HPV positive oropharyngeal cancer, which has not yet been examined in an OE-MRI study. Materials and methods: Variable Flip Angle T1 maps were acquired on a 3T MRI scanner while patients (n = 12) breathed air and/or 100 % oxygen, before and after fraction 10 of the planned 30 fractions of chemoradiotherapy ('visit 1' and 'visit 2', respectively). The analysis aimed to assess to what extent (1) native R1 relates to patient outcome; (2) OE-MRI response relates to patient outcome; (3) changes in mean R1 before and after radiotherapy related to clinical outcome in patients with oropharyngeal squamous cell carcinoma. Results: Due to the radiotherapy being largely successful, the sample sizes of non-responder groups were small, and therefore it was not possible to properly assess the predictive nature of OE-MRI. The tumour R1 increased in some patients while decreasing in others, in a pattern that was overall consistent with the underlying OE-MRI theory and previously reported tumour OE-MRI responses. In addition, we discuss some practical challenges faced when integrating this technique into a clinical trial, with the aim that sharing this is helpful to researchers planning to use OE-MRI in future clinical studies. Conclusion: Altogether, these results suggest that further clinical OE-MRI studies to assess hypoxia and radiotherapy response are worth pursuing, and that there is important work to be done to improve the robustness of the OE-MRI technique in human applications in order for it to be useful as a widespread clinical technique.

Challenges for machine learning in clinical translation of big data imaging studies.

Combining deep learning image analysis methods and large-scale imaging datasets offers many opportunities to neuroscience imaging and epidemiology. However, despite these opportunities and the success of deep learning when applied to a range of neuroimaging tasks and domains, significant barriers continue to limit the impact of large-scale datasets and analysis tools. Here, we examine the main challenges and the approaches that have been explored to overcome them. We focus on issues relating to data availability, interpretability, evaluation, and logistical challenges and discuss the problems that still need to be tackled to enable the success of "big data" deep learning approaches beyond research.

Modeling the Effect of Hyperoxia on the Spin-Lattice Relaxation Rate R1 of Tissues.

PURPOSE: Inducing hyperoxia in tissues is common practice in several areas of research, including oxygen-enhanced MRI (OE-MRI), which attempts to use the resulting signal changes to detect regions of tumor hypoxia or pulmonary disease. The linear relationship between PO2 and R1 has been reproduced in phantom solutions and body fluids such as vitreous fluid; however, in tissue and blood experiments, factors such as changes in deoxyhemoglobin levels can also affect the ΔR1. THEORY AND METHODS: This manuscript proposes a three-compartment model for estimating the hyperoxia-induced changes in R1 of tissues depending on B0, SO2 , blood volume, hematocrit, oxygen extraction fraction, and changes in blood and tissue PO2 . The model contains two blood compartments (arterial and venous) and a tissue compartment. This model has been designed to be easy for researchers to tailor to their tissue of interest by substituting their preferred model for tissue oxygen diffusion and consumption. A specific application of the model is demonstrated by calculating the resulting ΔR1 expected in healthy, hypoxic and necrotic tumor tissues. In addition, the effect of sex-based hematocrit differences on ΔR1 is assessed. RESULTS: The ΔR1 values predicted by the model are consistent with reported literature OE-MRI results: with larger positive changes in the vascular periphery than hypoxic and necrotic regions. The observed sex-based differences in ΔR1 agree with findings by Kindvall et al. suggesting that differences in hematocrit levels may sometimes be a confounding factor in ΔR1. CONCLUSION: This model can be used to estimate the expected tissue ΔR1 in oxygen-enhanced MRI experiments.

Determination of oxygen relaxivity in oxygen nanobubbles at 3 and 7 Tesla.

OBJECTIVE: Oxygen-loaded nanobubbles have shown potential for reducing tumour hypoxia and improving treatment outcomes, however, it remains difficult to noninvasively measure the changes in partial pressure of oxygen (PO2) in vivo. The linear relationship between PO2 and longitudinal relaxation rate (R1) has been used to noninvasively infer PO2 in vitreous and cerebrospinal fluid, and therefore, this experiment aimed to investigate whether R1 is a suitable measurement to study oxygen delivery from such oxygen carriers. METHODS: T1 mapping was used to measure R1 in phantoms containing nanobubbles with varied PO2 to measure the relaxivity of oxygen (r1Ox) in the phantoms at 7 and 3 T. These measurements were used to estimate the limit of detection (LOD) in two experimental settings: preclinical 7 T and clinical 3 T MRI. RESULTS: The r1Ox in the nanobubble solution was 0.00057 and 0.000235 s-1/mmHg, corresponding to a LOD of 111 and 103 mmHg with 95% confidence at 7 and 3 T, respectively. CONCLUSION: This suggests that T1 mapping could provide a noninvasive method of measuring a > 100 mmHg oxygen delivery from therapeutic nanobubbles.

Using Variable Flip Angle (VFA) and Modified Look-Locker Inversion Recovery (MOLLI) T1 mapping in clinical OE-MRI.

BACKGROUND AND PURPOSE: The imaging technique known as Oxygen-Enhanced MRI is under development as a noninvasive technique for imaging hypoxia in tumours and pulmonary diseases. While promising results have been shown in preclinical experiments, clinical studies have mentioned experiencing difficulties with patient motion, image registration, and the limitations of single-slice images compared to 3D volumes. As clinical studies begin to assess feasibility of using OE-MRI in patients, it is important for researchers to communicate about the practical challenges experienced when using OE-MRI on patients to help the technique advance. MATERIALS AND METHODS: We report on our experience with using two types of T1 mapping (MOLLI and VFA) for a recently completed OE-MRI clinical study on oropharyngeal squamous cell carcinoma. RESULTS: We report: (1) the artefacts and practical difficulties encountered in this study; (2) the difference in estimated T1 from each method used - the VFA T1 estimation was higher than the MOLLI estimation by 27% on average; (3) the standard deviation within the tumour ROIs - there was no significant difference in the standard deviation seen within the tumour ROIs from the VFA versus MOLLI; and (4) the OE-MRI response collected from either method. Lastly, we collated the MRI acquisition details from over 45 relevant manuscripts as a convenient reference for researchers planning future studies. CONCLUSION: We have reported our practical experience from an OE-MRI clinical study, with the aim that sharing this is helpful to researchers planning future studies. In this study, VFA was a more useful technique for using OE-MRI in tumours than MOLLI T1 mapping.

A General Model to Calculate the Spin-Lattice Relaxation Rate (R1) of Blood, Accounting for Hematocrit, Oxygen Saturation, Oxygen Partial Pressure, and Magnetic Field Strength Under Hyperoxic Conditions.

BACKGROUND: Under normal physiological conditions, the spin-lattice relaxation rate (R1) in blood is influenced by many factors, including hematocrit, field strength, and the paramagnetic effects of deoxyhemoglobin and dissolved oxygen. In addition, techniques such as oxygen-enhanced magnetic resonance imaging (MRI) require high fractions of inspired oxygen to induce hyperoxia, which complicates the R1 signal further. A quantitative model relating total blood oxygen content to R1 could help explain these effects. PURPOSE: To propose and assess a general model to estimate the R1 of blood, accounting for hematocrit, SO2 , PO2 , and B0 under both normal physiological and hyperoxic conditions. STUDY TYPE: Mathematical modeling. POPULATION: One hundred and twenty-six published values of R1 from phantoms and animal models. FIELD STRENGTH/SEQUENCE: 5-8.45 T. ASSESSMENT: We propose a two-compartment nonlinear model to calculate R1 as a function of hematocrit, PO2 , and B0. The Akaike Information Criterion (AIC) was used to select the best-performing model with the fewest parameters. A previous model of R1 as a function of hematocrit, SO2 , and B0 has been proposed by Hales et al, and our work builds upon this work to make the model applicable under hyperoxic conditions (SO2  > 0.99). Models were assessed using the AIC, mean squared error (MSE), coefficient of determination (R2 ), and Bland-Altman analysis. The effect of volume fraction constants WRBC and Wplasma was assessed by the SD of resulting R1. The range of the model was determined by the maximum and minimum B0, hematocrit, SO2 , and PO2 of the literature data points. STATISTICAL TESTS: Bland-Altman, AIC, MSE, coefficient of determination (R2 ), SD. RESULTS: The model estimates agreed well with the literature values of R1 of blood (R2  = 0.93, MSE = 0.0013 s-2 ), and its performance was consistent across the range of parameters: B0 = 1.5-8.45 T, SO2  = 0.40-1, PO2  = 30-700 mmHg. DATA CONCLUSION: Using the results from this model, we have quantified and explained the contradictory decrease in R1 reported in oxygen-enhanced MRI and oxygen-delivery experiments. LEVEL OF EVIDENCE: 3 TECHNICAL EFFICACY: Stage 1.

A simplified empirical model to estimate oxygen relaxivity at different magnetic fields.

The change in longitudinal relaxation rate (R1 ) produced by oxygen has been used as a means of inferring oxygenation levels in magnetic resonance imaging in numerous applications. The relationship between oxygen partial pressure (pO2 ) and R1 is linear and reproducible, and the slope represents the relaxivity of oxygen (r1Ox ) in that material. However, there is considerable variability in the values of r1Ox reported, and they have been shown to vary by field strength and temperature. Therefore, we have compiled 28 reported empirical values of the relaxivity of oxygen as a resource for researchers. Furthermore, we provide an empirical model for estimating the relaxivity of oxygen in water, saline, plasma, and vitreous fluids, accounting for magnetic field strength and temperature. The model agrees well (R2  = 0.93) with the data gathered from the literature for fields ranging from 0.011 to 8.45 T and temperatures of 21-40 °C. This provides a useful resource for researchers seeking to quantify pO2 in simple fluids in their studies, such as water and saline phantoms, or bodily fluids such as vitreous fluids, cerebrospinal fluids, and amniotic fluids.

Learning patterns of the ageing brain in MRI using deep convolutional networks.

Both normal ageing and neurodegenerative diseases cause morphological changes to the brain. Age-related brain changes are subtle, nonlinear, and spatially and temporally heterogenous, both within a subject and across a population. Machine learning models are particularly suited to capture these patterns and can produce a model that is sensitive to changes of interest, despite the large variety in healthy brain appearance. In this paper, the power of convolutional neural networks (CNNs) and the rich UK Biobank dataset, the largest database currently available, are harnessed to address the problem of predicting brain age. We developed a 3D CNN architecture to predict chronological age, using a training dataset of 12,802 T1-weighted MRI images and a further 6,885 images for testing. The proposed method shows competitive performance on age prediction, but, most importantly, the CNN prediction errors ΔBrainAge=AgePredicted-AgeTrue correlated significantly with many clinical measurements from the UK Biobank in the female and male groups. In addition, having used images from only one imaging modality in this experiment, we examined the relationship between ΔBrainAge and the image-derived phenotypes (IDPs) from all other imaging modalities in the UK Biobank, showing correlations consistent with known patterns of ageing. Furthermore, we show that the use of nonlinearly registered images to train CNNs can lead to the network being driven by artefacts of the registration process and missing subtle indicators of ageing, limiting the clinical relevance. Due to the longitudinal aspect of the UK Biobank study, in the future it will be possible to explore whether the ΔBrainAge from models such as this network were predictive of any health outcomes.

Oxygen-enhanced MRI MOLLI T1 mapping during chemoradiotherapy in anal squamous cell carcinoma.

BACKGROUND AND PURPOSE: Oxygen-enhanced magnetic resonance imaging (MRI) and T1-mapping was used to explore its effectiveness as a prognostic imaging biomarker for chemoradiotherapy outcome in anal squamous cell carcinoma. MATERIALS AND METHODS: T2-weighted, T1 mapping, and oxygen-enhanced T1 maps were acquired before and after 8-10 fractions of chemoradiotherapy and examined whether the oxygen-enhanced MRI response relates to clinical outcome. Patient response to treatment was assessed 3 months following completion of chemoradiotherapy. A mean T1 was extracted from manually segmented tumour regions of interest and a paired two-tailed t-test was used to compare changes across the patient population. Regions of subcutaneous fat and muscle tissue were examined as control ROIs. RESULTS: There was a significant increase in T1 of the tumour ROIs across patients following the 8-10 fractions of chemoradiotherapy (paired t-test, p < 0.001, n = 7). At baseline, prior to receiving chemoradiotherapy, there were no significant changes in T1 across patients from breathing oxygen (n = 9). In the post-chemoRT scans (8-10 fractions), there was a significant decrease in T1 of the tumour ROIs across patients when breathing 100% oxygen (paired t-test, p < 0.001, n = 8). Out of the 12 patients from which we successfully acquired a visit 1 T1-map, only 1 patient did not respond to treatment, therefore, we cannot correlate these results with clinical outcome. CONCLUSIONS: These clinical data demonstrate feasibility and potential for T1-mapping and oxygen enhanced T1-mapping to indicate perfusion or treatment response in tumours of this nature. These data show promise for future work with a larger cohort containing more non-responders, which would allow us to relate these measurements to clinical outcome.

Correction: Rapid determination of the tumour stroma ratio in squamous cell carcinomas with desorption electrospray ionization mass spectrometry (DESI-MS): a proof-of-concept demonstration.

Correction for 'Rapid determination of the tumour stroma ratio in squamous cell carcinomas with desorption electrospray ionization mass spectrometry (DESI-MS): a proof-of-concept demonstration' by Michael Woolman et al., Analyst, 2017, 142, 3250-3260.

Rapid determination of the tumour stroma ratio in squamous cell carcinomas with desorption electrospray ionization mass spectrometry (DESI-MS): a proof-of-concept demonstration.

Squamous cell carcinomas constitute a major class of head & neck cancers, where the tumour stroma ratio (TSR) carries prognostic information. Patients affected by stroma-rich tumours exhibit a poor prognosis and a higher chance of relapse. As such, there is a need for a technology platform that allows rapid determination of the tumour stroma ratio. In this work, we provide a proof-of-principle demonstration that Desorption Electrospray Ionization Mass Spectrometry (DESI-MS) can be used to determine tumour stroma ratios. Slices from three independent mouse xenograft tumours from the human FaDu cell line were subjected to DESI-MS imaging, staining and detailed analysis using digital pathology methods. Using multivariate statistical methods we compared the MS profiles with those of isolated stromal cells. We found that m/z 773.53 [PG(18:1)(18:1) - H]-, m/z 835.53 [PI(34:1) - H]- and m/z 863.56 [PI(18:1)(18:0) - H]- are biomarker ions that can distinguish FaDu cancer from cancer associated fibroblast (CAF) cells. A comparison with DESI-MS analysis of controlled mixtures of the CAF and FaDu cells showed that the abundance of the biomarker ions above can be used to determine, with an error margin of close to 5% compared with quantitative pathology estimates, TSR values. This proof-of-principle demonstration is encouraging and must be further validated using human samples and a larger sample base. At maturity, DESI-MS thus may become a stand-alone molecular pathology tool providing an alternative rapid cancer assessment without the need for time-consuming staining and microscopy methods, potentially further conserving human resources.

Optimized Mass Spectrometry Analysis Workflow with Polarimetric Guidance for ex vivo and in situ Sampling of Biological Tissues.

Spatially Targeted Mass Spectrometry (MS) analysis using survey scans with an imaging modality often requires consecutive tissue slices, because of the tissue damage during survey scan or due to incompatible sample preparation requirements between the survey modality and MS. We report two spatially targeted MS analysis workflows based on polarized light imaging guidance that use the same tissue sample for survey and targeted analysis. The first workflow is applicable for thin-slice analysis, and uses transmission-polarimetry-guided Desorption ElectroSpray Ionization Mass Spectrometry (DESI-MS), and confirmatory H&E histopathology analysis on the same slice; this is validated using quantitative digital pathology methods. The second workflow explores a polarimetry-guided MS platform for thick tissue assessment by developing reflection-mode polarimetric imaging coupled with a hand-held Picosecond InfraRed Laser (PIRL) MS ablation probe that requires minimal tissue removal to produce detectable signal. Tissue differentiation within 5-10 s of sampling with the hand-held probe is shown using multivariate statistical methods of the MS profiles. Both workflows were tasked with differentiating necrotic cancer sites from viable cancers using a breast tumour model, and their performance was evaluated. The use of the same tissue surface addresses mismatches in guidance due to intrinsic changes in tissue morphology over consecutive sections.

An Assessment of the Utility of Tissue Smears in Rapid Cancer Profiling with Desorption Electrospray Ionization Mass Spectrometry (DESI-MS).

Mass spectrometry imaging with desorption electrospray ionization mass spectrometry (DESI-MS) is used to characterize cancer from ex vivo slices of tissues. The process is time-consuming. The use of tissue smears for DESI-MS analysis has been proposed as it eliminates the time required to snap-freeze and section the tissue. To assess the utility of tissue smears for rapid cancer characterization, principal component analysis (PCA) was performed to evaluate the concordance between DESI-MS profiles of breast cancer from tissue slices and smears prepared on various surfaces. PCA suggested no statistical discrimination between DESI-MS profiles of tissue sections and tissue smears prepared on glass, polytetrafluoroethylene (PTFE), and porous PTFE. However, the abundances of cancer biomarker ions varied between sections and smears, with DESI-MS analysis of tissue sections yielding higher ion abundances of cancer biomarkers compared with smears. Coefficient of variance (CV) analysis suggests DESI-MS profiles from tissue smears are as reproducible as the ones from tissue sections. The limit of detection with smear samples from single pixel analysis is comparable to tissue sections that average the signal from a tissue area of 0.01 mm2. The smears prepared on the PTFE surface possessed a higher degree of homogeneity compared with the smears prepared on the glass surface. This allowed single MS scans (~1 s) from random positions across the surface of the smear to be used in rapid cancer typing with good reproducibility, providing pathologic information for cancer typing at speeds suitable for clinical utility. Graphical Abstract ᅟ.

Rapid Detection of Necrosis in Breast Cancer with Desorption Electrospray Ionization Mass Spectrometry.

Identification of necrosis in tumors is of prognostic value in treatment planning, as necrosis is associated with aggressive forms of cancer and unfavourable outcomes. To facilitate rapid detection of necrosis with Mass Spectrometry (MS), we report the lipid MS profile of necrotic breast cancer with Desorption Electrospray Ionization Mass Spectrometry (DESI-MS) imaging validated with statistical analysis and correlating pathology. This MS profile is characterized by (1) the presence of the ion of m/z 572.48 [Cer(d34:1) + Cl]- which is a ceramide absent from the viable cancer subregions; (2) the absence of the ion of m/z 391.25 which is present in small abundance only in viable cancer subregions; and (3) a slight increase in the relative intensity of known breast cancer biomarker ions of m/z 281.25 [FA(18:1)-H]- and 303.23 [FA(20:4)-H]-. Necrosis is accompanied by alterations in the tissue optical depolarization rate, allowing tissue polarimetry to guide DESI-MS analysis for rapid MS profiling or targeted MS imaging. This workflow, in combination with the MS profile of necrosis, may permit rapid characterization of necrotic tumors from tissue slices. Further, necrosis-specific biomarker ions are detected in seconds with single MS scans of necrotic tumor tissue smears, which further accelerates the identification workflow by avoiding tissue sectioning and slide preparation.

Variations in the Abundance of Lipid Biomarker Ions in Mass Spectrometry Images Correlate to Tissue Density.

While mass spectrometry (MS) imaging is widely used to investigate the molecular composition of ex vivo slices of cancerous tumors, little is known about how variations in the cellular properties of cancer tissue can influence cancer biomarker ion images. To better understand the basis for variations in the abundances of cancer biomarker ions seen in MS images of relatively homogeneous ex vivo tumor samples, sections of snap frozen human breast cancer murine xenografts were subjected to desorption electrospray ionization mass spectrometry (DESI-MS) imaging. Serial sections were then stained with hematoxylin and eosin (H&E) and subjected to detailed morphometric cellular analysis, using a commercial digital pathology platform augmented with custom-tailored image analysis algorithms developed in-house. Gross morphological heterogeneities due to stroma, vasculature, and noncancer cells were mapped in the tumor and found to not correlate with the areas of suppressed cancer biomarker abundance. Instead, the ion abundances of major breast cancer biomarkers were found to correlate with the cytoplasmic area of cancer cells that comprised the tumor tissue. Therefore, detailed cellular analyses can be used to rationalize subtle heterogeneities in ion abundance in MS images, not explained by the presence of gross morphological heterogeneities such as stroma.

Wide-field tissue polarimetry allows efficient localized mass spectrometry imaging of biological tissues.

While mass spectrometers can detect chemical signatures within milliseconds of data acquisition time, the non-targeted nature of mass spectrometry imaging (MSI) necessitates probing the entire surface of the sample to reveal molecular composition even if the information is only sought from a sample subsection. This leads to long analysis times. Here, we used polarimetry to identify, within a biological tissue, areas of polarimetric heterogeneity indicative of cancer. We were then able to target our MS analysis using polarimetry results to either the cancer region itself or to the cancer margin. A tandem of polarimetry and Desorption Electrospray Ionization Mass Spectrometry Imaging (DESI-MSI) enables fast (10 fold compared to non-targeted imaging), and accurate pathology assessment (cancer typification in less than 2 minutes compared to 30 minutes for histopathology) of ex vivo tissue slices, without additional sample preparation. This workflow reduces the overall analysis time of MSI as a research tool.