A Flexible RF Coil Design (OxyTrack) for Oxygen Measurements in Body Cavities

Abstract

Purpose:

Electron paramagnetic resonance (EPR) spectroscopy enables quantitative measurement of tissue oxygen levels. The conventional single-loop EPR resonator designs limit the oxygen measurements to superficial tissues within 1–3 cm depth and inadequately address clinical requirements for deep-tissue oxygen monitoring in anatomically complex regions and confined body cavities. The aim of this study was to develop a flexible RF coil-based sensor (OxyTrack) designed for real-time oxygen measurements in complex anatomical environments that are typically inaccessible to conventional rigid coil configurations.

Methods:

The RF coil configuration of the OxyTrack included a catheter-like, flexible design that incorporates the OxyChip (oxygen sensor) in the resonant loop. A modified coaxial cable arrangement with braided shielding was used for cavity measurements. The constructed coil/sensor was evaluated for power saturation thresholding, oxygen sensitivity (calibration), mechanical stability, and integrity of the coil under various stress conditions. Biological validation studies were performed to test dynamic oxygen variations in the gastrointestinal tract (GI) of murine subjects.

Results:

The flexible OxyTrack exhibited an oxygen sensitivity of 14.8 mG/mmHg with a linear response across physiological ranges (0–160 mmHg), maintaining signal integrity under various mechanical stresses. In vivo validation experiments in mice GI tracts demonstrated statistically significant discrimination of rectal tissue oxygenation between normoxic (0.52 ± 0.04 mmHg) and hyperoxic conditions (6.43 ± 0.24 mmHg) with p < 0.001. Pre-clinical imaging compatibility established the absence of significant artifacts.

Conclusion:

This flexible RF coil sensor enables minimally invasive, real-time oxygen monitoring in complex anatomical locations, with implications for pre-clinical research and potential clinical translation in oxygen-related pathophysiology assessment.

1 |. Introduction

Tissue hypoxia at diverse anatomical sites is linked to pathophysiological progression in cardiovascular disease, inflammatory conditions, and malignancies [1, 2]. Hypoxia in the gastrointestinal (GI) tract contributes to inflammatory bowel disease progression and ischemic complications, while myocardial oxygen deficiency drives coronary artery disease pathogenesis [3, 4]. The measurement of intracavitary tissue hypoxia, which characterizes these diseases, presents significant pre-clinical and clinical challenges due to limited accessibility and the need for quantitative oxygenation measurements during treatment [5].

Current clinical hypoxia assessment methodologies possess fundamental limitations restricting real-time pO₂ measurements for intracavitary applications. Imaging-based hypoxia detection techniques, including ¹⁸F-fluoromisonidazole positron emission tomography and blood oxygen level-dependent (BOLD) magnetic resonance imaging, provide regional hypoxia mapping but lack the spatial and temporal resolution needed for real-time localized tissue assessment [6, 7]. Immunohistochemical biomarker approaches utilizing HIF-1α expression or exogenous hypoxia markers provide cellular-level hypoxia detection but necessitate tissue sampling and cannot enable real-time monitoring during therapeutic interventions [8, 9]. Other perfusion-based hypoxia assessment techniques offer functional evaluation but only provide indirect hypoxia estimation through hemodynamics rather than direct quantification of tissue oxygenation [10, 11].

Electron paramagnetic resonance (EPR) oximetry provides direct hypoxia quantification through radiofrequency-based measurements, enabling absolute pO₂ determination via a paramagnetic oxygen probe [12, 13]. Operating in L-band frequency ranges, EPR oximetry demonstrates exceptional sensitivity for hypoxic tissue detection with measurement repeatability from identical locations over extended periods [14]. However, conventional EPR oximetry employs surface-loop resonators that constrain measurement depth to shallow tissues, including the skin and tissues directly below the skin (< 2 cm), preventing direct assessment of deep intracavitary tissues where some pathophysiological hypoxia develops [15, 16]. The geometry of surface resonators restricts access to tortuous anatomical structures, including coronary vessels, gastrointestinal tract segments, and bronchial pathways, where quantitative hypoxia evaluation requires a flexible sensor architecture.

This investigation evaluates a novel flexible EPR sensor (flexible OxyTrack) for feasibility in intracavitary in vivo applications. The sensor incorporates a flexible coaxial cable resonator design that addresses current depth penetration and anatomical accessibility limitations by permitting direct measurement of difficult-to-reach intracavity tissues, specifically for gastrointestinal oxygen monitoring, but also potentially including oximetry of coronary vessels, bronchial pathways, genital tracts, and the urinary tract.

2 |. Methods

2.1 |. Flexible OxyTrack Design and Construction

The flexible OxyTrack (Figure 1a,b) operates on a similar principle to the rigid OxyTrack described by O’Connell et al. [17]. In this instance, the main body uses a coaxial cable with a flexible braided shield to navigate tissue cavities without suffering from noise artifacts due to the bending wire. This design uses a modified Taoglas (CAB.616) 50-Ω coaxial cable assembly (Taoglas; Enniscorthy, Ireland) (Figure 1b) consisting of a silvered copper center wire, fluorinated ethylene propylene (FEP) insulator, a tinned copper braided shield, and an FEP jacket. A small loop is formed between the center wire and the braided shield forming the resonator. The loop houses the EPR oxygen probe LiNc-BuO incorporated into polydimethylsiloxane (PDMS) (in combination known as OxyChip) [18, 19], a highly oxygen-sensitive and oxygen-permeable material. It should be noted that the FEP insulation jacket covering the braided shield can prevent direct contact of the conducting shield with the tissue, thereby avoiding any potential effect on the resonator performance. Furthermore, the loop itself is encased in the OxyChip material, which isolates the conductive wire from the tissue as well.

FIGURE 1 |.

Flexible OxyTrack design, construction, and EPR characterization of oxygen sensitivity. (a) Schematic of flexible OxyTrack with OxyChip sensor. This design uses a modified Taoglas (CAB.616) coaxial cable assembly consisting of a silver-coated 50-Ω copper center wire, fluorinated ethylene propylene (FEP) insulator, a tin-coated copper-braided shield, and an FEP jacket. The coax cable (8-cm long) forms the flexible part of the device for insertion into body cavities. The schematic representation shows the modified cable with the center wire forming a loop with the braided shield. The OxyChip material is placed in and around the loop to create the oxygen-sensing probe. (b) Photographic image of the constructed flexible OxyTrack showing dimensionality. (c) EPR spectrum of the flexible OxyTrack equilibrated with 15.2 (gray) and 91.2 (black) mmHg of oxygen (pO₂) measured using L-Band (1.1 GHz) EPR spectrometer. The spectra were acquired in continuous mode using 0.1 mW RF power and 10 mG modulation amplitude. The SNR and linewidth of the spectra are shown. (d) Effect of oxygen on the linewidth of the flexible OxyTrack (calibration curve) showing a linear relationship in the 0–160 mmHg (21% oxygen) range. The oxygen sensitivity of the sensor is 14.8 mG/mmHg with an estimated SEIP of 0.634 mmHg.

EPR characterization of the probe and murine measurements were conducted using a custom-built continuous-wave EPR spectrometer equipped with an RF bridge operating at L-band frequency (~1.1 GHz). The spectrometer utilized continuous-wave mode with a 410 G permanent magnet (Sumitomo Special Metals; Japan) fitted with sweep coils (maximum B₀ sweep: 30 G) and field-modulation coils operating at 21 kHz (maximum modulation: 1.3 G). Field modulation was maintained at 10 mG for consistency across all measurements. Unless otherwise specified, all data points represent individual EPR spectrum scans (6 s/scan).

2.2 |. Power Saturation

The effects of RF power saturation were used to determine the optimal operating conditions for the resonator to maximize SNR while preventing artificial line broadening. The incident RF power was controlled using variable attenuators ranging from −43 dBm (50 nW) up to 16.5 dBm (44 mW) to fully characterize the saturation profile. Signal amplitude (mV) and linewidth (mG) were recorded at each power increment. Measurements were performed under anoxic conditions using nitrogen gas (Airgas Nitrogen Services; Radnor, PA).

2.3 |. Calibration of Oxygen Sensitivity

The flexible OxyTrack was calibrated using mixed nitrogen and oxygen gases with known pO₂ values, from 0 mmHg (0%) to 159.6 mmHg (21%) with 7.6 mmHg (1%) increments at room temperature and ambient pressure. Gas concentrations were verified using an FDO2 Optical Oxygen Sensor (Pyro Science GmbH; Aachen, Germany). The measured EPR linewidths were plotted against pO₂ and fitted with a linear function to obtain oxygen sensitivity (mG/mmHg) values.

2.4 |. Physical Characterization

The true physical build of the device was visualized by using the Quantum GX3 microCT system (Revvity; Waltham, MA) to reconstruct the probe head in three dimensions using Dragonfly 3D World. A 4.3^′′ digital microscope (GOCHIFIX; Hong Kong, China) was used to obtain a zoomed-in bright field view of the probe head. The electromagnetic profile of the probe was simulated using High-frequency Simulation Software (HFSS) (ANSYS; Canonsburg, PA) eigenmode analysis to visualize the relative magnetic field distribution in the L-band frequency.

2.5 |. Deformation Tests

2.5.1 |. Rotation and Isocenter Deviation

The flexible OxyTrack was fixed in a custom rotatable mount and systematically rotated through a complete 360° revolution in 30° angular increments relative to the main magnetic field B₀. To assess both rotational sensitivity and positional effects relative to the external magnetic field isocenter, measurements were conducted at multiple radial positions displaced from the isocenter, as illustrated by the cross-marked sampling positions in the experimental setup. Signal amplitude (mV) and linewidth (mG) were recorded at each of the 12 angular positions after allowing sufficient time for sensor stabilization to ensure no movement, providing a characterization of orientational dependence. The experiment was repeated four times for each incremental change while randomly deviating the sensor head from the isocenter to simulate real-world measurement variation. Measurements were performed under anoxic conditions.

2.5.2 |. Variable Bend Angle

Controlled mechanical bending was applied to the flexible OxyTrack with the total bend angles from 0° to 90°. The probe was progressively bent through discrete angles of 18°, 36°, 54°, 72°, and 90°, with the bending plane oriented perpendicular to the main magnetic field B₀ to isolate the effects on signal characteristics due to curvature-induced stress and B₁ orientation with respect to the main magnetic field. At each bend configuration, the probe was held firmly in each position to ensure no movement, after which both signal amplitude (mV) and linewidth (mG) measurements were recorded to quantify the relationship between mechanical deformation and signal response. Datapoints represent an average of five repeated EPR spectra. Measurements were performed under anoxic conditions.

2.5.3 |. Buckling Test

An axial compressive load was applied to the flexible OxyTrack with one rigidly fixed end, while the opposite end was free to translate under applied load. Compression was incrementally increased to achieve 0.5 cm in lateral deflection from the previous position. Signal amplitude (mV) and linewidth (mG) were measured at specific maximum lateral displacement (maximum displacement off-axis, DoA) values of 0, 0.5, 1.0, 1.5, 2.0, and 2.5 cm, corresponding to progressively increasing degrees of buckling deformation, while maintaining the magnetic field B₀ orientation relative to the cable’s original straight configuration to isolate the effects of buckling-induced stress and geometric changes on the magnetic resonance signal. Datapoints represent an average of five repeated EPR spectra. Measurements were performed under anoxic conditions.

2.6 |. pO₂ Measurements in Mouse

Measurement of pO₂ levels was carried out in the healthy gastrointestinal (GI) tract of a male C57BL/6 mouse using animal protocol approved by Dartmouth IACUC (#00002092(m24a)). The preparation involved subject isolation with a brief period of food deprivation to verify evacuation of fecal matter to minimize the risk of encountering interfering material with the probe. The mouse was anesthetized with 1.5% isoflurane and placed on a heating pad maintained at 37°C. The probe was inserted ~1.5 cm into the rectum. The oxygen challenge involved the mouse breathing medical air (21% O₂, normoxic) and switching to 100% O₂ (hyperoxic). The oxygen probe on the flexible OxyTrack was designed with a larger diameter than the coaxial cable to ensure secure contact with the GI lumen for stable readings and to prevent sensor movement. The quality factor of the resonator was evaluated using N9923A FieldFox portable network analyzer (Agilent Technologies; Santa Clara, CA) with the resonator in a cadaver mouse and in air. The resonator’s compatibility with CT and MRI was evaluated with a Revvity Quantum GX3 microCT and a 9.4T MRI (Bruker BioSpec; Ettlingen, Germany) in a cadaver mouse.

2.7 |. Data Analysis

Standard error of inverse prediction (SEIP) was used to calculate the error in oxygen measurement [20]. Briefly, SEIP is a measure of error in the x-direction (pO₂) instead of the typical y-direction (linewidth) error measurement and can be calculated as the root-mean-square error of the data to the fit divided by the slope of the linear fit. All statistical analyses were performed using Statistical Package for the Social Sciences (SPSS) (IBM SPSS; Chicago, IL). Pearson correlation was used to determine statistical significance between various deformation tests and the EPR signal linewidth. A two-tailed paired t-test was used to determine the statistical significance of pO₂ measurements in vivo.

3 |. Results

3.1 |. EPR and Oxygen-Sensitivity Characterization of the Flexible OxyTrack

Representative EPR spectra of the flexible OxyTrack equilibrated with 15.2 and 91.2 mmHg of oxygen (pO₂) measured using a CW L-Band (1.1 GHz) EPR spectrometer using 0.1 mW RF power and 10 mG modulation amplitude are shown in Figure 1c. The coil orientation was not optimized for these measurements. Field modulation was set to 10 mG for experimental consistency, though optimal modulation (proportional to linewidth) could have improved SNR for broader linewidths. The signal-to-noise ratio (SNR) for each spectrum was 68 and 50, respectively. The effect of oxygen-induced line-broadening was evident from the measured linewidths. Calibration of the coil demonstrated a highly linear relationship between EPR linewidth and pO₂ across the physiological range of oxygen levels. Using mixed nitrogen-oxygen gas standards spanning 0 to 159.6 mmHg (0%–21% O₂) in 7.6 mmHg (1% O₂) increments, measured EPR linewidths showed strong linear correlation with pO₂ values (Figure 1d). Linear regression analysis yielded an oxygen sensitivity of 14.8 mG/mmHg with a coefficient of determination of R² = 0.999. The calibration curve exhibited no deviation from linearity across the tested range, confirming the sensor’s suitability for quantitative oxygen measurements under ambient conditions. The SEIP was calculated to be 0.634 mmHg.

3.2 |. Physical Characterization

OxyChip is transparent to x-ray irradiation, permitting examination of the constructed flexible OxyTrack head design using high-resolution microCT (Figure 2a). A zoomed-in bright-field optical image of the coil head shows OxyChip placement in the center of the resonator loop and coating the transmission line wire (Figure 2b). Ansys HFSS FEM eigenmode simulation of the coil head demonstrated the maximum magnetic field strength distributed within the resonant loop (Figure 2c,d).

FIGURE 2 |.

Resonator construction. (a) High-resolution microCT image of the flexible OxyTrack sensing head. The inner diameter of the loop is approximately 1 mm, and the outer diameter is 1.6 mm. OxyChip material appears transparent in this image. (b) A zoomed-in bright field optical image of the sensing head showing the EPR sensitivity material. (c, d) Ansys HFSS FEM eigenmode simulation of the relative magnetic field strength due to the resonant loop; sagittal and axial views, respectively.

3.3 |. Power Saturation Studies

Power saturation studies revealed expected behavior for both signal amplitude and linewidth across the tested power range. Signal amplitude increased linearly (on the semi-log plot) with incident RF power before reaching a peak amplitude at approximately 1.0 mW (Figure 3a). Beyond this threshold, further power increases yielded a negative impact on signal amplitude. Power saturation starts to occur at ~0.173 mW. Linewidth measurements revealed no broadening for incident power levels < −0.173 mW, while progressive broadening was observed at higher powers (Figure 3b). At the maximum tested power of 44 mW, the linewidth increased by 100% compared to the optimal conditions. Based on these saturation profiles, an optimal incident RF power of 0.1 mW was selected for all subsequent experiments to maximize the SNR without power saturation-induced line broadening.

FIGURE 3 |.

Power saturation curves. (a) Power saturation curve of signal amplitude (mV) versus applied RF power (mW) on a log–log plot. The peak amplitude occurs around 1.0 mW. (b) Power saturation curve of linewidth (mG) versus applied RF power (mW) on a semi-log plot. Artificial linewidth broadening occurs when power is greater than 0.173 mW.

3.4 |. Deformation Tests

3.4.1 |. Rotation and Isocenter Deviation

Rotational measurements revealed an expected orientation dependency of signal amplitude across the 360° rotation (Figure 4a). Signal amplitude varied by an average of ~24 mV from maximum to minimum values, with the peak signal occurring at 120° and the minimum signal at 30°. The 90° separation between maximum and minimum reflects the loop’s response to the orientation of the B₁ relative to the static magnetic field (B₀). The specific angular positions (120° for maximum, 30° for minimum) reflect our arbitrary choice of 0° reference position on the rotatable mount during experimental setup, rather than any orientation relative to the magnetic field. The amplitude profile exhibited a characteristic sinusoidal pattern with 2-fold symmetry. Linewidth measurements demonstrated lesser sensitivity to angular position, with an average standard deviation of 1.4 mG across the rotation range. Positional displacement from the magnetic field isocenter introduced additional real-world variability. The Pearson correlation showed a negligible, positive correlation between rotational orientation, with isocenter deviation and linewidth. The correlation was not statistically significant, r(46) = 0.00842, p = 0.955, where 46 represents the degree of freedom (sample size—2) (Figure 5a). Despite the rotational and isocenter deviation dependence on SNR, the linewidth measurements remain largely unaffected.

FIGURE 4 |.

Deformation test. (a) Rotation and isocenter deviation test. The objective was to simulate real-world positional variation. On the left, the black circle represents the tip of the resonator at the isocenter. The gray circles with x represent repeated measurements that are randomly deviated from the isocenter. The data from x position shown x on the top graph. (Top) signal amplitude versus rotation angle. (Bottom) linewidth versus rotation angle. Amplitude varied by an average of ~24 mV from maximum to minimum. The linewidth average standard deviation was 1.4 mG across the range of rotation. (b) Variable bend angle test. The objective of this test was to see if flexing the resonator affected the linewidth. (Top) signal amplitude versus bend angle. (Bottom) linewidth versus bend angle. The linewidth average standard deviation was 1.72 mG across the range of deformation. (c) Buckling test. This test was designed to determine the flexible limits of the resonator. (Top) signal amplitude versus DoA (max displacement off-axis). (Bottom) linewidth versus DoA.

FIGURE 5 |.

Deformation test data analysis. Correlation analysis of data shown in Figure 4. (a) Rotation and isocenter deviation test. The Pearson correlation showed a negligible, positive correlation between rotational orientation, with isocenter deviation and linewidth. The correlation was not statistically significant, r(46) = 0.00842, p = 0.955, where 46 represents the degree of freedom (sample size—2). (b) Variable bend angle test. The Pearson correlation showed a negligible, negative correlation between linewidth and bend angle, which was not statistically significant, r(28) = −0.31, p = 0.099. (c) Buckling test. The Pearson correlation showed a moderate, positive correlation, r(28) = 0.37, p = 0.046, up to 2.5 cm DoA. If DoA does not exceed 2 cm, the Pearson’s correlation results show only a low, positive correlation that is not statistically significant, r(23) = 0.12, p = 0.564.

3.4.2 |. Variable Bend Angle

Controlled mechanical bending of the OxyTrack demonstrated robust linewidth stability across the tested deformation range, but a noticeable SNR dependence (Figure 4b). The Pearson correlation showed a statistically significant, r(28) = 0.84, p = < 0.001, positive correlation between the bend angle and signal amplitude (Figure 5b). The amplitude showed no statistical correlation up to a bend angle of 36 degrees (r(13) = −0.34, p = 0.215). However, from 36 degrees to 90 degrees, the amplitude response exhibited a strong positively correlated linear relationship with bend angle, r(18) = 0.985, R² = 0.97, p < 0.001. Linewidth measurements revealed much less sensitivity to mechanical deformation, with a standard deviation of 1.72 mG across the full bending range (Figure 4b). The Pearson correlation showed a trend between linewidth and bend angle, which was not statistically significant, r(28) = −0.31, p = 0.099 (Figure 5b). Linewidth remained stable despite the pronounced bend angles. The five replicate measurements at each bend configuration yielded an average standard deviation of 0.5 mV (coefficient of variation, CV = 0.75%) for signal amplitude and 1.0 mG (CV = 1.2%) for linewidth, demonstrating good measurement reproducibility under this type of mechanical stress.

3.4.3 |. Buckling Test

Axial compression testing was the most stress-inducing test performed and was designed to test the flexible limits of the resonator. The test revealed the coil’s response to post-buckling deformation across lateral displacements up to 2.5 cm. Signal amplitude demonstrated dependence under compressive loading for most of the test (Figure 4c). From 0 cm DoA to 2.5 cm DoA, there was a very high, positive correlation between DoA and signal amplitude. The Pearson correlation showed a statistical significance between the two variables, r(28) = 0.76, p < 0.001. Linewidth measurements exhibited a lesser, but still significant sensitivity to buckling deformation from 0 to 2.5 cm DoA (Figure 5c). The Pearson correlation showed a moderate, positive correlation, r(28) = 0.37, p = 0.046. However, 2.5 cm DoA exceeds the manufacturer’s bend radius specification for the cable (9 mm min.). If DoA does not exceed 2 cm, the Pearson’s correlation results show only a low, positive correlation that is not statistically significant, r(23) = 0.12, p = 0.564 (Figure 5c). Statistical analysis of the five replicate measurements at each displacement (from 0 to 2.0 cm DoA) yielded average standard deviations of 0.4 mV (CV = 1.9%) for signal amplitude and 1.4 mG (CV = 1.6%) for linewidth, indicating good measurement precision under this type of stress.

3.5 |. Measurement of Dynamic Changes of Oxygen Levels in the Rectal Cavity of a Mouse

For in vivo applications, the flexible OxyTrack was designed to easily fit into the rectum of a mouse (Figure 6a) to measure dynamic changes in rectal wall oxygenation. The Q of the resonator was measured to be ~558 in air and ~147 in a mouse (Figure 6b). Compatibility with two common pre-clinical imaging modalities was evaluated through the acquisition of a microCT fluoroscopic image (Figure 6c) and T1-weighted MRI minimum intensity projected 2D slice stack (Figure 6d). Analysis of these images revealed no significant image-distorting artifacts. In vivo demonstration of quantitative oximetry in a healthy mouse gastrointestinal system was performed. A sample spectrum of a pO₂ measurement at 3.7 mmHg (Figure 6e) shows an SNR of 222. After inserting the coil approximately 1.5 cm into the rectal cavity, the EPR spectrum was continuously acquired to monitor oxygen levels while the anesthetized mouse was breathing room air for 8 min to reach equilibrium with the surrounding GI lumen (Figure 6f). The pO₂ of the GI lumen at this time was extremely low (< 1 mmHg). The EPR spectra were continuously acquired while simultaneously administering an oxygen challenge to hyperoxic (100% oxygen) breathing; the succeeding measurements showed a slow but steady rise in the available oxygen in the tissue during hyperoxygenation. After approximately 8 min, the oxygen level reached equilibrium. The normal oxygen state had lower pO₂ values ($\overline{{pO}_2}=0.516\text{mmHg}$, σ = 0.0427 mmHg, from 8 to 10 min) than the hyper-oxygenated state ($\overline{{pO}_2}=6.43\text{mmHg}$, σ = 0.237 mmHg, from 18 to 20 min). A t-test for paired samples showed that this difference was statistically significant, t(20) = −107.228, p = < 0.001, with a Cohen’s d of 23.4.

FIGURE 6 |.

Murine colorectal measurement. (a) Cartoon representation of the experimental setup with the flexible OxyTrack inserted into the rectal cavity of a mouse. (b) S-parameter plot displaying the resonant absorption peak of the resonator in the mouse (black) and in air (gray) conditions. The quality factor (Q) of the resonator was measured to be ~558 in air and ~147 in the mouse. (c) microCT fluoroscopic image of the flexible OxyTrack in the colorectal cavity of a deceased mouse. The white arrows identify the outline of the flexible OxyTrack as a reference for panels 6 (c) and 6 (d). (d) MRI T1-weighted minimum intensity projected image of the flexible OxyTrack in the same deceased mouse. (e) Sample EPR spectrum of a pO₂ (3.7 mmHg) measurement of the mouse rectal wall. (f) Dynamic changes of oxygen levels inside the mouse rectal cavity during different breathing conditions. (Left) medical air breathing. The measured value of oxygen slowly decreases as the sensor equalizes with the rectal cavity after having been exposed to room air. (Right) After the sensor equalized with the rectal wall, the breathing gas was switched to 100% O₂. After a small delay, the oxygen level of the rectal wall slowly increases until it reaches equilibrium after about 8 min. Normoxic (0.52 ± 0.04 mmHg) and hyperoxic conditions (6.43 ± 0.24 mmHg), p < 0.001.

4 |. Discussion

4.1 |. Quantitative Performance

The flexible OxyTrack design presented in this report differs from our previously reported rigid OxyTrack [17] in both coaxial cable flexibility and resonator geometry. While both devices achieve comparable oxygen sensitivity for the LiNc-BuO sample volume, the flexible design uses a more optimal circular loop configuration unconstrained by the needle-tip geometry required for the rigid system. This circular geometry provides improved sensitivity compared to the extended loop configuration necessitated by the rigid device’s needle shape, while the flexible coaxial design allows deployment in anatomically complex cavities that are inaccessible to rigid resonators. For the flexible OxyTrack, the effective B₁ field region extends approximately 80% of the loop diameter on either side (~1.6-mm for our 1-mm loop), the highly oxygen-permeable nature of the PDMS material ensures oxygen equilibration throughout the entire OxyChip volume, allowing accurate oxygen measurements of the surrounding tissue despite only ~51% of the material being in the effective sensing region. The inhomogeneous distribution of the B₁ field in the OxyChip material does, however, manifest itself as a broadening of the anoxic spectrum due to uneven excitation, which in this case is ~120 mG. In contrast, when the OxyChip is uniformly excited in a homogenous field, the anoxic linewidth is ~60 mG [21].

The power saturation analysis highlighted the excellent sensitivity of the flexible OxyTrack, which enables reproducible quantitative measurements at very low power levels. This sensitivity stems from the small loop geometry, approximately 1-mm diameter, working in concert with the inherently strong EPR signal from the OxyChip material. The compact loop concentrates the microwave field while substantially reducing the required power. Mechanical-stress testing confirmed robust performance of the sensor under realistic experimental conditions. Flexing the coil at controlled increments showed no statistically significant effect on significant linewidth, even under 90° deflection, covering the expected deformation during intracavitary use. The buckling analysis verified that operational limits align with cable specifications, demonstrating reliability under mechanical conditions well beyond typical experimental requirements. It should be noted that cables with smaller bend radius are available and may be employed to address translational needs.

The rotational studies revealed an expected angular dependence in signal amplitude. In EPR spectroscopy, the maximum signal occurs when the oscillating magnetic field (B₁) generated by the resonator loop is perpendicular (90°) to the static magnetic field (B₀). For a planar loop resonator with 180° rotational symmetry about its axis, rotation through 360° produces two orientations where B₁ is optimally perpendicular to B₀, resulting in a sinusoidal amplitude pattern with two maxima separated by approximately 180°. The observed maximum and minimum peaks at 120° and 30° (separated by 90°) confirm this expected behavior and reflect the loop plane’s orientation relative to B₀ in our experimental setup. Importantly, linewidth data, the critical parameter for oxygen determination, demonstrated minimal sensitivity to angular position (average standard deviation of 1.4 mG), confirming that measurement accuracy persists regardless of sensor orientation. This stability is crucial because the linewidth directly determines oxygen concentration. The data show that while optimizing signal-to-noise ratio may require careful positioning at very low power, the core measurement accuracy persists regardless of sensor orientation. This finding has immediate practical significance: anatomical constraints that require suboptimal sensor positioning need not compromise measurement quality, which provides a substantial advantage for potential pre-clinical and clinical oxygen monitoring, where precise placement may be anatomically restricted.

4.2 |. Biological Validation of the Methodology

The murine gastrointestinal tract experiments established the sensor’s capacity for continuous monitoring of tissue oxygenation dynamics within a physiologically relevant research model. Baseline measurements revealed markedly low colonic pO₂ values, findings that align with other reports using a different approach [22]. The controlled oxygen challenge protocol, transitioning from medical air to 100% oxygen administration, induced measurable tissue oxygenation increases with temporal kinetics consistent with physiological patterns. These experimental results suggest immediate translational applications for pre-clinical research, which might be particularly useful within radiation-induced gastrointestinal injury assessments. The capacity for quantitative pre- and post-radiation oxygen measurements positions this technology as a potential non-invasive biomarker platform for evaluating tissue damage or pathology.

Compatibility with standard pre-clinical imaging modalities, including microCT and MRI was verified and revealed minimal artifact generation. This imaging compatibility opens pathways for simultaneous real-time tissue oxygen monitoring alongside anatomical and functional imaging, potentially enabling comprehensive tissue characterization protocols that integrate metabolic and structural assessment capabilities. The absence of significant imaging interference suggests that concurrent measurements could provide interesting insight into the temporal relationship between dynamic tissue oxygenation and corresponding anatomy during experimental interventions.

4.3 |. Methodological Innovation and Pre-Clinical Research Potential

The flexible OxyTrack is an approach to addressing fundamental limitations of in vivo tissue oximetry, specifically the microwave depth penetration constraints that restrict conventional EPR spectroscopic oximetry to superficial tissue. This technological development enables quantitative, real-time monitoring of the partial oxygen pressure of intracavitary tissues. By employing a small resonant loop with an oxygen-permeable paramagnetic probe, contact-mediated oxygen diffusion from the local tissue environment could correlate with tissue pO₂ levels. Requiring only surface contact, rather than invasive tissue penetration, avoids tissue trauma while enabling reproducible measurements at precise anatomical locations.

The technology demonstrates particular relevance for experimental radiotherapy applications, where tissue hypoxia represents a well-established mechanism of treatment resistance, with hypoxic conditions reducing therapeutic effectiveness by factors of 2–3. From a practical pre-clinical perspective, this sensor could facilitate a transition from population-based treatment protocols toward individualized therapeutic strategies through quantitative assessment of specimen-specific tissue oxygenation. This proof-of-concept investigation demonstrates that the fundamental microwave depth penetration limitations in biological tissues can be systematically addressed through innovative resonator design, enabling continuous-wave L-band EPR spectroscopy for tissue cavity measurements in previously inaccessible anatomical regions.

4.4 |. Technical Limitations

Several limitations should be considered for future applications. The sensor’s key feature, which measures partial oxygen pressure, may be sensitive to various factors in active tissue environments. During times of increased digestive activity, the peristaltic contractions of the gastrointestinal tract can create fluctuating mechanical pressure on the device. This pressure might introduce small changes in measurements. Careful experimental design is needed, and it may be necessary to monitor physiological changes at the same time to differentiate between oxygen concentration and changes in pressure. Other factors include the need for direct contact with tissue, which may limit its use in some situations. The current design of the sensor is tailored for specific measurements, thus it might require modifications for pre-clinical applications not discussed in this paper. Further evaluation may be necessary to determine the suitability of this technology based on how it will be used. While the metallic resonator loop is encapsulated within the oxygen-permeable PDMS/LiNc-BuO material and not in direct contact with tissue, we acknowledge that RF-induced heating during MRI remains a potential concern that requires formal assessment.

4.5 |. Research Implications and Technological Extensions

The flexible OxyTrack addresses key limitations in current oxygen monitoring techniques. It provides absolute quantitative measurements for deep intracavitary environments. This marks a major improvement over older polarographic electrodes, which consume oxygen during measurements and lack sensitivity at low oxygen levels, and surface-based optical sensors, which have limited depth. This development fills a gap in the quantitative assessment of deep-tissue hypoxia with a relevant timeframe for clinical use. EPR oximetry is often an overlooked standard for precise tissue oxygen measurements in preclinical studies. The flexible design opens access to areas that were previously difficult to reach, making a valuable contribution to the methodology. Its high sensitivity to low oxygen levels is especially useful for monitoring regional hypoxic tissue. This includes areas where treatment resistance often arises, like radioresistant tumor cores or ischemic heart regions. It may enable real-time tracking of oxygen levels during treatment to improve effectiveness and reduce complications.

Possible clinical applications include cardiac catheterization for real-time oxygen monitoring in the heart during coronary procedures, tracking tumor hypoxia in radiation oncology, and monitoring gastrointestinal issues in inflammatory bowel disease and ischemic conditions, along with potential neurological assessment.

5 |. Conclusions

This study demonstrates the successful development and validation of the flexible OxyTrack EPR sensor, which represents an innovation in EPR oximetry technology for biomedical applications. By using a flexible coaxial cable design, we’ve overcome the key limitations of traditional surface EPR sensors, allowing us to noninvasively measure tissue oxygenation in complex internal anatomical structures that were previously out of reach for EPR oximetry. We achieved several important technical milestones, which include preserving the high sensitivity (14.8 mG/mmHg sensitivity, R² = 0.999) of the OxyChip despite the flexible coaxial cable design, while using low power. The sensor also showed robust mechanical performance under real-world deformation conditions and worked seamlessly with common imaging techniques without creating any artifacts. The tests confirm that the flexible OxyTrack maintains measurement accuracy even when subjected to various mechanical stresses, which provides evidence for its suitability in dynamic biological environments. Furthermore, the in vivo validation studies using mice gastrointestinal tracts showed that the sensor can effectively monitor tissue oxygenation changes in real-time, which concurs with previously published values. The sensor’s ability to track how oxygenation responds during oxygen challenges demonstrates its potential for monitoring both normal physiological processes and disease states as they occur. These results lay the groundwork for future pre-clinical and clinical translation, where continuous tissue oxygenation monitoring could make a real difference in survival outcomes.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.