The effect of crystal composition and environment on the color Doppler ultrasound twinkling artifact

Objective. Pathological mineralizations form throughout the body and can be difficult to detect using conventional imaging methods. Color Doppler ultrasound twinkling highlights ∼60% of kidney stones with a rapid color shift and is theorized to arise from crevice microbubbles as twinkling disappears on kidney stones at elevated pressures and scratched acrylic balls in ethanol. Twinkling also sometimes appears on other pathological mineralizations; however, it is unclear whether the etiology of twinkling is the same as for kidney stones. Approach. In this study, five cholesterol, calcium phosphate, and uric acid crystals were grown in vitro and imaged in Doppler mode with a research ultrasound system and L7-4 transducer in water. To evaluate the influence of pressure on twinkling, the same crystals were imaged in a high-pressure chamber. Then, the effect of surface tension on twinkling was evaluated by imaging crystals in different concentrations of surfactant (1%, 2%, 3%, 4%) and ethanol (10%, 30%, 50%, 70%), artificial urine, bovine blood, and a tissue-mimicking phantom. Main results. Results showed that all crystals twinkled in water, with cholesterol twinkling significantly more than calcium phosphate and uric acid. When the ambient pressure was increased, twinkling disappeared for all tested crystals when pressures reached 7 MPa (absolute) and reappeared when returned to ambient pressure (0.1 MPa). Similarly, twinkling across all crystals decreased with surface tension when imaged in the surfactant and ethanol (statistically significant when surface tension <22 mN m−1) and decreased in blood (surface tension = 52.7 mN m−1) but was unaffected by artificial urine (similar surface tension to water). In the tissue-mimicking phantom, twinkling increased for cholesterol and calcium phosphate crystals with no change observed in uric acid crystals. Significance. Overall, these results support the theory that bubbles are present on crystals and cause twinkling, which could be leveraged to improve twinkling for the detection of other pathological mineralizations.


Introduction
Pathological mineralizations can form throughout the body and cause severe pain and disability or signify an underlying condition (Tsolaki and Bertazzo 2019). The gold standard for detecting mineralizations is computed tomography (CT) with sensitivities of 80%-93%; however, CT scans are expensive and expose patients to ionizing radiation (Smith andVaranelli 2000, Choi et al 2012). Blood and fluid tests have good sensitivity (80%-90%), but do not allow for localization of minerals and are only sensitive to mature mineralization (Sturrock 2000, Sirotti et al 2021. B-mode ultrasound is a cheap, safe, and noninvasive alternative that can be used to detect mineralizations, but its sensitivity is highly dependent on the operator (19%-93%), reducing its usefulness (Sorensen et al 2003, Filippucci et al 2009, Perez-Ruiz et al 2009. The color Doppler ultrasound twinkling artifact, or twinkling, appears as a rapid color shift on nearly 60% of kidney stones and can supplement traditional B-mode imaging (Aytaç and Ozcan 1999, Lee et  Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. Dillman et al 2011, Kielar et al 2012, Winkel et al 2012, Korkmaz et al 2014, Masch et al 2016. Twinkling has also appeared on some other mineralization sites; however, the cause of twinkling on pathological mineralizations remains under investigation. Our objective is to evaluate the effects of crystal composition, hydraulic pressure, and environment on twinkling.
Twinkling was first observed on kidney stones and was initially attributed to scattering off the rough stone surface (Rahmouni et al 1996). Since then, scientists have debated as to whether twinkling was an artifact arising from signal processing or a physical interaction between the ultrasound and kidney stone (Chelfouh et al 1998, Aytaç and Ozcan 1999, Kamaya et al 2003, Alan et al 2011, Wang et al 2011, Gao et al 2012, Tanabe et al 2014, Shang et al 2017, Wood and Urban 2020. After almost two decades of research, Lu et al (2013) suggested scattering off microbubbles in crevices on the kidney stone surface causes twinkling. The primary support for this theory was that imaging kidney stones in decreased or elevated hydrostatic pressures resulted in an increase or decrease of twinkling, respectively (Lu et al 2013, Simon et al 2018. Researchers then sought specific stone features that may harbor the microbubbles and, in addition to surface features, found gas pockets in the internal microstructure that could be contributing to twinkling (Simon et al 2018, Rokni et al 2021. These results continue to support the microbubble theory of twinkling on kidney stones; however, it is not clear whether similar microbubbles are present on pathological mineralizations that cause twinkling.
Common pathological mineralizations include uric acid crystals in joints which forms gout, cholesterol crystallization that forms in some atherosclerotic plaques, and calcium phosphate that forms in heterotopic ossification or breast microcalcifications (Grebe and Latz 2013, Ragab et al 2017, Mujtaba et al 2019, Tsolaki and Bertazzo 2019. Unlike kidney stones, which are heterogenous structures of crystals interwoven in organic matrices, these other pathological mineralizations are more homogeneous crystals with minimal organic material (Tsolaki and Bertazzo 2019). Previously, researchers have tried to correlate twinkling with kidney stone composition, but results have been mixed likely due to the complicated structures and mixed compositions of kidney stones (Shang et al 2017, Wood and Urban 2020, Rokni et al 2021. However, the homogeneity of pathological mineralizations may allow for more consistency in the chemical properties like hydrophobicity and physical properties like surface roughness, which may allow twinkling to predict composition. Pathological mineralizations also form in various environments (e.g. urine, synovial fluid, blood, and soft tissue); the interaction between crystals with different chemical properties and the environment will affect the presence and distribution of potential crevice microbubbles. Theoretical works describe how surface tension stabilizes crevice bubbles against dissolution (Apfel 1970, Crum 1979, Atchley and Prosperetti 1989. Surface tension also influences the bubble dynamics when the crystal is imaged with ultrasound. Lu et al (2013) demonstrated this experimentally by imaging a scratched acrylic sphere in water and 70% ethanol with Doppler ultrasound. Twinkling was observed when the sphere was imaged in water but disappeared when imaged in 70% ethanol, which has a much lower surface tension than water (72 mN m −1 versus 22 mN m −1 , respectively). Therefore, our objective is to determine whether pathological mineralizations contain microbubbles that cause twinkling. Crystals of three different compositions mimicking those of common pathological mineralizations (i.e. cholesterol, calcium phosphate and uric acid) were grown in vitro and scanned with Doppler ultrasound. The effect on twinkling was evaluated for crystals imaged in elevated hydrostatic pressures, reduced surface tensions through surfactant/ethanol solutions, and in in situ-mimicking environments (blood, artificial urine, and a tissue-mimicking phantom). We hypothesize that twinkling will be strongest in cholesterol crystals and will decrease as pressure or surface tension increases.

Crystal growth
Cholesterol, calcium phosphate, and uric acid crystals were grown in vitro following recipes modified from Abela and Aziz (2005), Rinaudo and Boistelle (1980), and Francis et al (1969), respectively. Briefly, for cholesterol crystals, 3 g of cholesterol powder (92.5%, Sigma Aldrich, St. Louis, MO, USA) was melted and then cooled to room temperature to allow for crystal precipitation. For uric acid crystals, 0.3 g of crystalline uric acid (99%, Sigma Aldrich, St. Louis, MO, USA) were dissolved in 300 ml of boiling deionized water. The pH was adjusted to ∼4.5 by adding 0.1 molar sodium hydroxide solution (Belle Chemical, Billings, MT, USA) and measured with an EcoSense ® pH 10A meter (YSI, Yellow Springs, OH, USA). Wooden balls with diameters of 8 mm were added to the solution as a precipitation site due to the small crystal size. Before precipitation, the wooden balls were soaked in water and imaged with ultrasound to confirm they did not influence the twinkling artifact. Precipitation occurred over 5-10 d at room temperature. For calcium phosphate crystals, 1 g of sodium phosphate monobasic (99%, Sigma Aldrich, St. Louis, MO, USA) and 1 g of calcium chloride (93%, Sigma Aldrich, St. Louis, MO, USA) were added to 500 ml of boiling deionized water. The pH was adjusted to ∼6.4 adding by adding 0.1 molar sodium hydroxide solution. The solution was left for >1 week at room temperature for crystals to precipitate.
Crystals were evaluated for inorganic similarities to the in vivo pathological mineralization using Raman Spectroscopy (Horiba LabRam HR evolution) using one representative crystal of each composition. A 488 nm excitation laser with maximum power of 4.5 mW was focused through a 50x LWD (NA 0.5) objective lens and confocal hole of 100 um and a grating of 600 gr mm −1 . The spectrometer was calibrated using the 1st order response of a single crystal Si sample. Spectral peaks of grown crystals were compared to reference spectra of similar pure minerals or pathological mineralizations (RRUFF, Tuscon, AZ, USA) ( Effect of crystal composition on the twinkling artifact Ultrasound imaging was performed in filtered (<5 μm) and deionized water. Following previous twinkling studies by Lu et al (2013) and Simon et al (2018), the water was degassed to <20% O 2 concentration (Extech D0210 Dissolved Oxygen Meter, Extech, Waltham, MA, USA) using a custom-built flow system with a gas contactor membrane (PDMSXA-2.1, PermSelect©, Ann Arbor, Michigan, USA). A research ultrasound system (Vantage-128, Verasonics ® , Kirkland, WA, USA) with an L7-4 imaging transducer (Philips/ATL, Bothell, WA, USA) was used with a center frequency of 5.2 MHz, elevation focus of 30 mm, and −6 dB azimuthal angle of 1.7°. The Doppler waveform was measured in degassed and deionized water using a golden capsule hydrophone (HGL-Series, Onda, Sunnyvale, CA, USA) to have peak pressures of p + ≈ 3.6 MPa and p -≈ 3.1 MPa. Ultrasound parameters were held constant across all scans with 7 Doppler ensembles consisting of 12 cycles each repeated at 3000 Hz.
Five crystals of each composition with similar max dimensions (cholesterol ≈ 10 mm, calcium phosphate ≈ 5 mm, uric acid on 8 mm wooden balls ≈ 0.3 mm) were chosen to be imaged with Doppler ultrasound. Crystals were submerged in water and degassed in a desiccant chamber at ∼0.01 MPa absolute for >2 h prior to imaging to remove excess bubbles. Crystals were individually imaged in the water tank on an 8 × 8 × 1.5 cm block of neoprene with the transducer located 30 mm above. The demodulated radio-frequency (RF) data, or in-phase quadrature (IQ) data, was saved at approximately 1 frame per second (fps). The transducer was mechanically scanned across the crystal with a step size of 0.5 mm based on the measured −6 dB azimuthal angle. At each step, 2 frames were averaged to create 1 representative frame of Doppler power. Then, twinkling was quantified by summing the Doppler power inside a region of interest (ROI) box around the crystal and dividing by a depthand size-matched background ROI (figure 1). Total twinkling on each crystal was calculated by summing across all frames and dividing by the number of frames to account for differences in crystal size. This value was then converted to a dB scale using the background ROI as the reference.
Effect of hydrostatic pressure on the twinkling artifact An aluminum hydraulic pressure chamber similar to the one used in Lu et al (2013) was custom-built to evaluate the effect of pressure on twinkling (figure 2). The same L7-4 ultrasound transducer was used to image each crystal at one location through a 2-cm thick acrylic window in the chamber lid. Peak pressures were measured using a golden capsule hydrophone (HGL-series, Onda, Sunnyvale, CA, USA) to be ∼1 MPa after propagation through the acrylic window. Hydraulic pressure was increased using a hand pump (AMETEK, Berwyn, PA, USA) until pressure reached 14 MPa absolute. The pressure increase was paused every 3.5 MPa for approximately 1 min to allow IQ data to be saved at constant pressure. Twinkling was quantified by dividing the summed Doppler power in the ROI including the crystal by the background ROI and plotted versus time and pressure.

Effect of surface tension on the twinkling artifact
To evaluate the effect of surface tension, the same crystals were mechanically scanned, as was previously done in water, in 1%, 2%, 3%, and 4% surfactant solutions (Palmolive, New York, NY, USA) and 10%, 30%, 50% and 70% ethanol (KOPTEC, King of Prussia, PA, USA). A 5 μl drop of each liquid was photographed on an acrylic sheet with a Canon Rebel T6 camera (Canon U.S.A, Inc., Melville, NY) and the contact angle was measured in ImageJ (NIH, Bethesda, MD, USA) using the procedure defined by Lamour et al (2010). Surface tension was calculated using the equation where s s is the surface energy (38 mJ m −2 for acrylic), sl s is the interstitial energy between the surface and liquid, l s is the surface tension of the liquid, and q is the measured contact angle.
To observe twinkling in more in situ-mimicking environment, crystals were also imaged in artificial urine (Shmaefsky 1995), heparinized whole bovine blood (Lampire, Pipersville, PA, USA), and tissue-mimicking polyacrylamide (PA) gel. Artificial urine was created by mixing 24.3 g urea (99.5%, GE Healthcare Life Sciences, Marlborough, MA, USA), 10 g crystalline sodium chloride (EMD Millipore Corp., Billerica, MA, USA), 6 g potassium chloride (Sigma-Aldrich, St. Louis, MO, USA), and 6.4 g sodium phosphate monobasic (99.0%, Sigma-Aldrich, St. Louis, MO, USA) in 1 l deionized and degassed water. The PA gel was created according to Lafon et al (2005). In brief, 71.6 ml of deionized and degassed water was added to 17.5 ml of acrylamide (Sigma-Aldrich, St. Louis, MO, USA) and degassed in the desiccant chamber for >2 h. Half of the PA solution was crosslinked using 0.24 ml of 10% ammonium persulfate (98.0%, Sigma-Aldrich, St. Louis, MO, USA) and 0.0025 ml tetramethylethylenediamine (99%, Sigma-Aldrich, St. Louis, MO, USA). Crystals were submerged and degassed using a desiccant chamber at ∼0.01 MPa absolute in the PA solution for >2 h prior to adding the crosslinking agents.

Statistical analysis
All statistical analyses were performed using Minitab (Minitab, State College, PA, USA). Normality was confirmed using a Ryan-Joiner test. One-way ANOVAs were used to determine statistical significance in all studies. Post-hoc Tukey tests were used to compare the effect of composition and hydraulic pressures while post-hoc Dunnett tests were used to compare the effects of each environment to water. Values of P 0.05 indicate statistical significance.

Results
Effect of crystal composition on the twinkling artifact All crystals were successfully grown with uric acid crystals covering the upper surface of the wooden ball. Raman spectra for each crystal showed peaks consistent with inorganic minerals found in vivo ( 3). Every crystal displayed the twinkling artifact with cholesterol crystals producing the most twinkling (5.8 ± 0.5 dB frame −1 ), followed by calcium phosphate crystals (4.5 ± 0.4 dB frame −1 ), and finally uric acid crystals (4.4 ± 0.2 dB frame −1 ) (figure 4). Twinkling was significantly higher on cholesterol crystals compared to calcium phosphate and uric acid crystals (p = 0.002 and p = 0.005, respectively). There was no statistical difference between calcium phosphate and uric acid crystals (p = 0.8).
Effect of hydrostatic pressure on the twinkling artifact When crystals were scanned in the pressure chamber, twinkling was found to decrease as the absolute pressure increased (figures 5 and 6). In all 5 cholesterol crystals, twinkling significantly decreased when the pressure increased from 0.1 to 7 MPa (p < 0.001) and significantly increased when the pressure decreased from 14 to 0.1 MPa (p < 0.001). The mean Doppler power across all 5 calcium phosphate crystals decreased significantly when the pressure increased from 0.1 to 3.5 MPa (p < 0.001) as well as when the pressure increased from 3.5 to 7 MPa (p < 0.001). In 3/5 calcium phosphate crystals, twinkling decreased at 3.5 MPa, whereas 2/5 crystals did not show a reduction in twinkling until the pressure reached 7 MPa. The Doppler power again increased for calcium phosphate crystals when the pressure decreased from 14 to 0.1 MPa (p < 0.001). For all uric acid crystals,  twinkling decreased significantly when the pressure increased from 0.1 MP to 7 MPa (p < 0.001) and significantly increased when the pressure decreased from 14 to 0.1 MPa (p < 0.001). Interestingly, for uric acid crystals, the mean Doppler power at initial ambient conditions (0.1 MPa) was significantly higher than when the crystal was returned to ambient pressure after being exposed to 14 MPa (p < 0.001). When crystals were imaged in artificial urine (75.8 mN m −1 ) no significant differences in twinkling were noted compared to water for any crystal composition. However, when crystals were imaged in blood (surface   tension of 52.8 mN m −1 ), twinkling was significantly lower compared to imaging in water for cholesterol crystals (p = 0.02), calcium phosphate crystals (p = 0.02), and uric acid crystals (p = 0.006). Finally, when crystals were imaged in the polyacrylamide gel, twinkling increased significantly compared to water for cholesterol (7.2 ± 0.6 dB frame −1 ; p −1 = 0.03) and calcium phosphate (6.2 ± 0.9 dB frame −1 ; p −1 = 0.01) crystals; no difference between twinkling in water and the polyacrylamide gel was observed for uric acid crystals (4.4 ± 0.4 dB frame −1 ; p −1 = 0.8).

Discussion
These results demonstrate that crystal composition, hydraulic pressure, and surface tension affect twinkling, which supports the theory that microbubbles on or within crystals cause twinkling. Ultrasound imaging in water produced twinkling on all crystals with cholesterol crystals showing consistently higher Doppler power for all environments compared to uric acid or calcium phosphate crystals. Doppler power decreased in all crystals at elevated hydrostatic pressures and increased when the pressure was reduced to ambient levels. The Doppler power was found to decrease with surface tension when imaged in surfactant or ethanol. When the environment was modified to simulate in situ conditions, no change in twinkling compared to water was observed when all crystals were imaged in artificial urine, but twinkling decreased when all crystals were imaged in blood and increased when cholesterol and calcium phosphate crystals were imaged in the tissue-mimicking phantom. Cholesterol crystals were found to twinkle more than calcium phosphate or uric acid crystals, regardless of the ambient medium. As cholesterol is a lipid and thus very hydrophobic, it is likely that more or larger bubbles are present as previous experimental work has shown that bubbles are more likely to form in micron sized hydrophobic crevices (Ryan andHemmingsen 1998, Yang et al 2003). Calcium phosphate and uric acid crystals are significantly less hydrophobic than cholesterol, although some ionic charge may contribute to the distribution of bubbles on the surface (Bell et al 1972, Ghosh et al 2015, Ma et al 2017. Quantifying hydrophobicity is typically achieved by measuring the contact angle which proved difficult with the crystals used here as it requires a smooth surface (Raichur et al 2000). Future work could include surface characterization of the crystals with scanning electron microscopy or 3D coherence scanning interferometry to provide more insight into the specific surface features on the crystals that could harbor bubbles (Lee et al 2021, Rokni et al 2021. Twinkling was found to decrease for all crystals when the ambient pressure increased beyond 7 MPa; twinkling returned when the pressure was again reduced to ambient levels. For cholesterol and uric acid crystals, the effect of pressure was consistent for all 5 tested crystals, with calcium phosphate showing more variation amongst individual crystals. These results were not surprising as variations were noted in the hydraulic pressures needed to decrease twinkling in kidney stones (Lu et al 2013, Simon et al 2018. Lu et al (2013) found pressures of 8.5 MPa reduced twinkling on all tested stones while Simon et al (2018) found pressures of 9.7 MPa only reduced twinkling on 69/90 tested stones. Our results with lab-grown crystals were very consistent compared to the kidney stone results. It is likely that our in vitro crystals were more consistent in structure and composition compared to kidney stones, reducing the variability in the response to overpressure. Interestingly, the calcium phosphate crystals showed some variation in the maximum pressure needed to reduce twinkling. While the exact reason is unclear, it is possible that the formulation of calcium phosphate, which involved mixing three chemicals, caused structural variations that influenced the size and/or distribution of bubbles. Additionally, twinkling on uric acid crystals did not recover to the same levels when imaged at ambient pressures. As uric acid is known to be soluble in water (8 mg/100 ml; Cicerello 2018), it is possible that partial dissolution of the already small crystals disrupted any surface crevice bubbles. This phenomenon was also suspected to occur in uric acid kidney stones (Simon et al 2018).
Decreasing the surface tension of the liquid surrounding the crystals using surfactant or ethanol was found to reduce twinkling on all tested crystals. As surface tension stabilizes bubbles, it is likely that reducing the surface tension caused some bubbles to dissolve into solution, thus reducing twinkling. When the crystals were imaged in artificial urine, which has similar surface tension to water, no significant differences in twinkling were observed. However, when crystals were imaged in blood, which had a surface tension only slightly lower than water, twinkling significantly decreased for all crystals. Attenuation is the most likely reason for the reduction in twinkling as imaging through 30 mm of blood reduces the peak pressures to p + ≈ 2.7 MPa and p -≈ 2.3 MPa (attenuation ≈ 0.7 dB cm −1 , Treeby et al 2011). Indeed, when the crystals were re-scanned in water with similar lower pressure amplitudes, twinkling was reduced to levels similar to what was noted in blood. However, when crystals were embedded in the tissue-mimicking polyacrylamide gel, twinkling increased significantly for cholesterol and calcium phosphate crystals; uric acid crystals showed no difference between water and tissue-mimicking gel. This could again relate to the surface chemistry or physical properties like surface roughness which could cause pockets of air to be trapped around calcium phosphate and cholesterol crystals during gelation. In vivo studies of pathological mineralizations in tissue, such as heterotopic ossification, would provide a more realistic model to understand how tissue affects twinkling.
Although results were relatively consistent for the small sample size of crystals used for ultrasound imaging (n = 5) and Raman spectroscopy (n = 1), small variations in surface roughness and exact crystal chemistry may not be fully captured in this present study. While the Raman spectra of the grown crystals aligned well with those of pathological mineralizations, some extraneous peaks or slight shifts in peak values indicate small differences in composition makeup or structure which may or may not influence twinkling. Additionally, we should note that changing the surface tension of the surrounding medium using surfactant and ethanol, will influence factors besides surface tension. For example, viscosity will also be influenced by adding surfactants or ethanol; however, previous work suggests viscosity has minimal or no effect on surface bubbles (Mirsandi et al 2020). Surfactants are also commonly used to stabilize microbubbles, so increased surfactant concentrations could result in unintended bubble stabilization (Khan et al 2019). Our results suggest this did not occur in our study, as the surface tension threshold for reductions in twinkling were similar for ethanol and the surfactant. While all crystal-environment combinations were investigated for completeness, not all cases are clinically relevant. Finally, while these results provide empirical evidence that bubbles are present on crystals, they do not provide visual evidence. Future work could involve visualizing bubbles using μCT or environmental electron scanning microscopy, as has been done previously with kidney stones, which would provide further support for the microbubble theory of twinkling on crystals (Simon et al 2018, Rokni et al 2021.

Conclusions
Overall, the results suggest microbubbles are present on crystals and cause twinkling. All lab-grown crystals twinkled in all environments, with cholesterol consistently showing the most twinkling. Twinkling disappeared in crystals when the ambient pressure increased; twinkling returned when the pressure was again reduced to ambient levels. Similarly, twinkling decreased when surface tension was decreased using surfactant or ethanol. Twinkling was not affected by imaging in artificial urine, decreased when imaging in blood, and increased for cholesterol and calcium phosphate crystals when imaging in a tissue-mimicking phantom. These results suggest the etiology of twinkling on crystals mimicking pathological mineralizations is similar to that of kidney stones, which may allow for the development of methods to improve twinkling on pathological mineralizations.