Method for detecting circulating cells in superficial body fluids

Abstract

The present invention relates to a method for measuring circulating cells in superficial body fluids by means of high-frequency-based device. The method can be used for detecting circulating cells in the fluids of an individual without the necessity of extracting a sample of the individual, being useful as a diagnostic tool and for monitoring the effectiveness of a treatment administered to an individual suffering from a viral, protozoal, fungal and/or bacterial disease.

Claims

1. An in vitro method for detecting and quantifying circulating cells in superficial body fluids of a subject comprising a) placing an ultrasound-based device in contact with a liquid sample isolated from a superficial body fluid of the subject for obtaining a ultrasound data of said liquid sample; b) analyzing the ultrasound data obtained in step a) by a data processing system which provides information about cells and their concentration in the fluid, and c) correlating the information obtained in step b) with the presence and amount of circulating cells in the individual, wherein the ultrasound-based device operates at a central frequency in the range of 10-100 MHz and at a pulse duration between 25-1,000 ns, and at a wavelengths between 15 and 150 m.

2. The method according to claim 1, wherein the signal emitted by the ultrasound-based device operates at a central frequency of 20 MHz and/or at a pulse duration of 500 ns, and/or at a wavelength of 75 m.

3. The method according to claim 1, wherein the circulating cells are white blood cells, red blood cells or cancer cells.

4. The method according to claim 1, wherein the ultrasound-based device comprises a transducer.

5. The method according to claim 1, wherein the ultrasound data obtained in step a) is in the form of (i) collection of A-line data and/or of (ii) B-mode or C-mode type of 2D data or 3D data.

6. The method according to claim 1, wherein the information about cells is cell size, cell concentration and/or cell viability.

7. The method according to claim 1, wherein the superficial body fluid is selected from the group consisting of cerebrospinal fluid, blood, urine, pleural fluid, synovial fluid and pericardial fluid.

8. An in vitro method for the diagnosis of a viral, protozoal, fungal and/or bacterial disease in a subject or for identifying subjects who may be suffering from a viral, protozoal, fungal and/or bacterial disease comprising a) quantifying circulating cells in superficial body fluids of the subject by a method according to claim 1, and b) correlating the amount of said circulating cells with the presence of a viral, protozoal, fungal and/or bacterial disease in the subject, wherein an amount of circulating cells higher than a reference value is indicative of a viral, protozoal, fungal and/or bacterial disease, or is indicative that the subject may be suffering from a viral, protozoal, fungal and/or bacterial disease.

9. An in vitro method for monitoring the effectiveness of a treatment administered to a subject suffering from a viral, protozoal, fungal and/or bacterial disease, comprising a) quantifying circulating cells in a superficial body fluid of the subject before and after the treatment by a method according to claim 1, and b) correlating the amount of said circulating cells with the effectiveness of the treatment administered to said subject, wherein an amount of circulating cells in the subject after the treatment lower than the amount of circulating cells before the treatment is indicative that the treatment is being effective.

10. An in vitro method according to claim 8, wherein the viral, protozoal, fungal and/or bacterial disease is meningitis.

11. A non-invasive method for detecting and quantifying circulating cells in superficial body fluids comprising a) placing an ultrasound-based device on the skin of the subject for obtaining a ultrasound data of the a superficial body fluid under the skin; b) analyzing the ultrasound data obtained in step a) by a data processing system which provides information about cells and their concentration in the fluid, and c) correlating the information obtained in step b) with the presence and amount of circulating cells in the subject, wherein the ultrasound-based device operates at a central frequency in the range of 10-50 MHz, and at a pulse duration between 50-1,000 ns, and at a wavelength between 30 and 150 m.

12. The method according to claim 11, wherein the signal emitted by the ultrasound-based device operates at a central frequency of 20 MHz and/or at a pulse duration of 500 ns, and/or at a wavelength of 75 m.

13. The method according to claim 11, wherein the circulating cells are white blood cells, red blood cells or cancer cells.

14. The method according to claim 11, wherein the ultrasound-based device comprises a transducer.

15. The method according to claim 11, wherein the ultrasound data obtained in step a) is in the form of (i) collection of A-line data and/or of (ii) B-mode or C-mode type of 2D data or 3D data.

16. The method according to claim 11, wherein the information about cells is cell size, cell concentration and/or cell viability.

17. The method according to claim 11, wherein the superficial body fluid is selected from the group consisting of cerebrospinal fluid, blood, urine, pleural fluid, synovial fluid and pericardial fluid.

18. A method for the diagnosis of a viral, protozoal, fungal and/or bacterial disease in a subject or for identifying subjects who may be suffering from a viral, protozoal, fungal and/or bacterial disease, comprising a) quantifying circulating cells in superficial body fluids of the subject by a method according to claim 11, and b) correlating the amount of said circulating cells with the presence of a viral, protozoal, fungal and/or bacterial disease in the subject, wherein an amount of circulating cells higher than a reference value is indicative of a viral, protozoal, fungal and/or bacterial disease, is indicative that the subject may be suffering from a viral, protozoal, fungal and/or bacterial disease.

19. Method for monitoring the effectiveness of a treatment administered to a subject suffering from a viral, protozoal, fungal and/or bacterial disease comprising a) quantifying circulating cells in a superficial body fluid of the subject before and after the treatment by a method according to claim 11, and b) correlating the detection of said circulating cells with the effectiveness of the treatment administered to said subject, wherein an amount of circulating cells in the subject after the treatment lower than the amount of circulating cells before the treatment is indicative that the treatment is being effective.

20. Method according to claim 18, wherein the viral, protozoal, fungal and/or bacterial disease is meningitis.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1. Schematic diagram of the method used.

(2) FIG. 2. Ultrasound backscatter coefficient in relation to the WBC concentration measured with a Fuchs Rosenthal hemocytometer. The dashed lines represents the linear fit that explains the backscatter variability of the data when compared to WBC concentration (97%), as measured by means of the coefficient of determination, R{circumflex over ()}2. This figure shows the capabilities of high-frequency ultrasounds to confidently distinguish WBC concentrations below and above the meningitis diagnostic threshold of 10 cells/L and 20 cells/L set for 1-3 months old infants and neonates, respectively. For in vivo validation, the system needs to be adjusted to counteract the fontanelle attenuation to maintain diagnostic sensitivity.

(3) FIG. 3. 2D ultrasound image of CSF samples at WBC concentrations 0, 12 and 100 cells/L as measured with hemocytometer. An increased number of scatterers (cells) are observed in the image with increased WBC concentration. No scatterer (cell) is observed at 0 cells/L and that individual scatters can be observed up to 100 cells/L.

(4) FIG. 4. Diagrammatic description of cells detection in the CSF space. A first group of signals is reflected from the fontanelle layers, a second group of backscatters is obtained if cells are present in the fluid and, a third group of signals are reflected from the fluid-brain interface. This last group of signals can be used to indicate optimal alignment of the transducer, as it is orthogonal to all tissue layers.

(5) FIG. 5 shows a transducer array which can be used to focalize the ultrasound energy at different points.

(6) FIG. 6. Diagram showing a single element transducer in contact with a fluid. (Left) The transducer delivers maximum acoustic energy in a focal region. The sensitivity of the transducer is best in the focal region. In order to sample different regions of the liquid, the acoustic beam of the transducer can be mechanically displaced or electronically displaced if multiple elements are used. (Middle) The acoustic pressure exerted into the fluid (or acoustic radiation force) implies a fluid displacement that follows a pattern, i.e. a flow. (Right) If particles are confined in the fluid, they are trapped into the flow as long as their position is within the field of view of the transducer. The higher the energy and pulse repetition frequency, the higher the flow velocity and particles velocity when trapped within the flow.

(7) FIG. 7. Backscatter signal of a particle is that signal reflected back to the transducer by a scattering object. The closer the wavelength of the emitted signal is to the scatter diameter, the higher the backscattered signal. As observed in the right diagram, the emitted signal keeps propagating in the fluid at with a lower energy since part of the energy is reflected by the particle and part is being attenuated by the medium.

(8) FIG. 8. (Left) Backscatter spectra of cells of different size show an increase energy as the cell diameter increases. In a larger band implementation (ex-vivo application) the maximum of the frequency will be also be displaced towards higher frequencies for smaller cells. (Right) Backscatter spectra is shifted to lower frequencies and widens in bandwidth as the attenuation of the medium is increased by higher protein concentrations.

(9) FIG. 9. At each pulse emission, the acoustic pressure exerts a displacement of the particle at a velocity that depends on the energy and repetition frequency of the pulse as well as in the particle and medium properties. The backscatter signal velocity can be tracked, e.g. by correlating consecutive acquisitions (right side of the figure), and used to improve the detection sensitivity of the technique in compromised signal-to-noise conditions.

(10) FIG. 10. Schematic of a coding system to enhance cell sensitivity. (Top) In transmission, an electrical impulse is coded to a larger sequence with an overall higher energy. This sequence excites the transducer which converts this energy to a mechanical signal that is input into the system. (Bottom) In receiving mode, the signal may be at SNR close to 0 dB, but the coded pattern can still be recognized (deconvolved, decorrelated) by the decoder, which outputs a peak if the sequence is detected.

EXAMPLES

(11) There exists a clinical interest in noninvasively detecting changes in the composition of body fluids. A very clear example is that of the cerebrospinal fluid (CSF), where, currently, only invasive approaches provide accurate information of fluid composition or characteristics. In case of an infection, a major concern is to detect whether an external pathogen has accessed the CSF space. In such case, the immunologic system reacts and sends white blood cells (WBCs) to the site of infection to kill the external organism. In the process of leaving the peripheral blood stream, WBCs penetrate vessel walls and allow blood proteins to leak into the site of infection. The increase of cellularity and protein level in the fluid space progressively changes the bulk properties of the fluid like its density and viscosity.

Example 1

(12) Sample Preparation

(13) CSF mimicking samples were elaborated as an ultrafiltrate version of human blood plasma at varying concentration of white blood cells (WBCs) in the 0-100 cells/L range. Blood samples from patients with leukocytosis arriving at the fluid analysis laboratory in the hospital were collected in tubes and treated with disodium ethylenediaminetetraacetate (EDTA). The tubes were centrifuged at 300 g for 10 minutes and 1-2 mL of plasma from each EDTA tube was used to create a pool of plasma. Pool proteins were measured and saline serum was added to obtain a final plasma volume of 50 mL at 0.5 g/L proteins level. This protein matrix constituted our healthy mock CSF. From each EDTA tube, 1 mL of buffy coat was pipetted out with a Pasteur pipette and transferred to a Wintrobe tube. The ten Wintrobe tubes were then centrifuged at 300 g for 10 minutes. For each tube, the WBC layer was pipetted out and diluted in 1-mL of the protein matrix. From this 1 mL stock suspension, cell dilutions were prepared to obtain concentrations in the range 0-100 WBC/L. Cell count and differential was performed by means of a Fuchs Rosenthal hemocytometer.

(14) Methods

(15) Cell samples were ultrasonically scanned using a single element transducer (V3320, Olympus, Waltham, Mass., USA). The transducer centre frequency was 75 MHz with a 6 dB bandwidth ranging from 42.5 MHz to 101 MHz, a focus located at 12.5 mm and an f-number of 2. The transducer was excited with a Panametrics 5900PR pulser (Olympus, Waltham, Mass., USA) and connected to a Picoscope 6402D oscilloscope (Pico Technology, Cambridgeshire, United Kingdom). Sequential RF signals centred in the focus were acquired and stored in a workstation for post-processing. Cell samples were put in direct contact with the transducer and contained in a specifically designed PVC lid that was coiled onto the transducer. As a result of the attenuation through the liquid medium and the cell response, the pulse received is centered at 45 MHz, having a wavelength of 33 m and 45 ns pulse duration. Samples were pipetted in and out of the lid cavity through a 6 mm aperture (FIG. 1). Samples were kept at 4 C. and measured within three hours from sample elaboration to maintain cell viability levels above 90%. A 450 L sample volume was used to guarantee immersion of the acoustic focus in the sample. The backscatter coefficient was computed from the RF signals using the method described in FIG. 1 and compared to the hemocytometer results. The time and frequency response of the backscattered signal coming from cells was theoretically studied using Anderson's model with a Poisson modulus close to 0.5, the value used for cells in this frequency range.

(16) Results

(17) FIG. 2 shows two linear trends that put in relation the backscatter energy to the total WBC counts and to the WBC counts as measured with the hemocytometer. Results showed linear agreement of the backscatter signal with increasing WBC (R{circumflex over ()}2: 0.97).

Example 2

(18) Cell Sample Preparation

(19) See example 1 above.

(20) Methods

(21) Cell samples were ultrasonically scanned using a commercial ultrasound system (DermaScan C USB, Corex Technologies, Hadsund, Denmark). The transducer centre frequency was 20 MHz (pulse wavelength, 75 m) with a 6 dB bandwidth ranging from 17.5 MHz to 22.5 MHz, a system resolution of 60 m (axial)150 m (lateral), a focus located at 6 mm, a linear scan of 1.21 mm and a frame-rate of 5 fps.

(22) Results

(23) 2D images were obtained from the imaging system for concentrations from 0-100 cells/L. FIG. 3 shows three images corresponding to 0, 12 and 100 WBC/L samples. Individual echoes observed in the images may be attached to single cells despite of the system resolution-to-particle size mismatch. In the long wavelength limit where the wavelength of the acoustical energy being scattered may be greater than the size of a weakly scattering (nonresonant) particle, i.e., >2 a where may represent the wavelength and a may represent the particle radius, the backscattered energy may be due to Rayleigh scattering and it may depend on the contrast between the compressibility and density of the particle and that of the fluid suspension medium, and the volume of the particle. The axial resolution given by the manufacturer is 60 m and the radii of the WBC in the sample were of about 6 m. As a result, 2 a/ equals 0.63 (<1 with a/< 1/10) concluding that the backscattering effect is within the Rayleigh regime.

Example 3

(24) Non-Invasive Diagnosis of Meningitis

(25) Bacterial Meningitis (BM) is a very aggressive disease that affects the central nervous system and that still causes high morbidity and mortality among young infants and neonates. In developing countries mortality is about 40-58% and about 10% in developed countries. The clinical presentation of this disease is very specific being fever the most common and sometimes only symptom. Diagnosis requires a sample of the cerebrospinal fluid (CSF), a fluid that surrounds the brain and spinal cord. Currently, the lumbar puncture is the only way to obtain a sample of the CSF to analyze its composition characteristics. However, the lumbar puncture is an invasive procedure, difficult to perform in the infants and neonates and very often (up to 48%) traumatic, meaning that blood contaminates the sample leading to unreliable CSF results. In developing countries where the incidence of the disease is 10 (in endemic regions) or even 100 times (in epidemic seasons) than in developed countries (10 cases per 100.000 people), it is common that a lumbar puncture is never performed because of the lack of laboratory facilities to analyze the sample. Hence, diagnosis is heavily based on the clinical symptomatology perceived by the physician or in his/her absence the nurse, who follows a clinical protocol. In developed countries, where laboratory facilities are abundant and given the poor prognosis for an infected patient in case of delayed treatment, physicians have a low threshold to perform a lumbar puncture in patients with meningeal symptoms (e.g. fever). Moreover, patients with an increased CSF WBC count (over 20 cells/L in newborns <1 month, and over 10 cells/L in infants 1-3 months old) will immediately get empirical antibiotic treatment for BM until the CSF bacterial culture result is available (after 48-72 hours). Although a safe strategy, this results in up to 95% of infants without BM receiving a lumbar puncture and many of them also treated with antibiotics. Therefore, in developed countrieswhere the incidence of BM is lowthis strategy does not add any benefit to the non-BM patients's healthcare. So for under-resourced as developed countries we think that better and easy-to-use methods are needed.

(26) The method of the invention uses high-frequency ultrasounds to sense the presence of cells in the CSF through the fontanelle (FIG. 4) and signal processing to detect individual cells and provide CSF cell concentration to assist in the non-invasive diagnosis of meningitis as well as reducing the number of lumbar punctures to those that are only strictly necessary.

(27) High frequency ultrasound manufacturers:

(28) Pre-Clinical Imaging VisualSonics Inc. (Canada) Atys medical (France)

(29) Dermatology Atys medical (France) TPM, Taberna Pro Medicum (Germany) Cortex technologies (Denmark) Sonoscape (China) Sonosite, Inc (USA) Longwood, Inc (USA) Esaote (Italy) GE Healthcare (US)

(30) 1. Acustic Fundamental and Concept Development

(31) 1.1 Sampling

(32) 1.1.1. Volume Sampling

(33) The ultrasound device (or transducer) applies a pressure signal that is maximum in a small region, i.e. the focal region, of the liquid volume. If the device is comprised by one single element that remains static, only one-dimensional data is obtained. If multiple regions of the fluid volume are going to be sampled, the single element can be mechanically or electronically displaced and two- or three-dimensional data (or images) can be obtained. When using multiple elements, an acoustic beam can be electronically steered to image a portion or the entire liquid volume (FIG. 5).

(34) 1.1.2 Focused Acoustic Radiation Force

(35) The acoustic pressure applied by the transducer generates an acoustic radiation force that may induce a flow in the liquid volume. Such flow increases as the pulse repetition frequency (PRF) also increases, i.e. the frequency at which the pressure signal is emitted, and is maximum in the focal region (FIG. 6).

(36) 1.2 Fluid Characterization

(37) 1.2.1 Acoustic Backscatter of Single Cells

(38) If there are cells in the liquid volume (a suspension), part of the pressure signal is reflected by the cells back to the transducer (FIG. 7). These reflected signals from single cells are called acoustic backscattered signals or backscatter. Backscatter energy increases as the wavelength of the emitted signal approaches the size of the cell, i.e. the frequency of the emitted signal increases. Therefore, whenever we refer to the transducer in the text, it should be understood that we are referring to a high-frequency transducer. Signal and data processing is also needed to provide a measurement of cell concentration (see section 2).

(39) 1.2.2 Spectral Content of Backscatter from Cells

(40) The spectral content of a backscatter signal is a hallmark of the cell producing such backscatter signal. Cell properties like size and composition are related to the energy, frequency and bandwidth of the backscatter signal. Similarly, the viscosity of the suspension medium can also be related to frequency shifts and bandwidth changes of the backscatter spectrum as well as from the spectrum of the signal reflected at the interface between the fluid and the distal wall of the container. Such analysis is only possible at very low skin attenuation or ex-vivo liquid analysis (FIG. 8).

(41) 1.2.3 Cell Velocity

(42) Cell velocity can be induced by either natural fluid flow, the radiation force of the ultrasound signal or the pressure implied by the user on the tissue, which would displace inner structures.

(43) At low SNRs, single cell velocity increases the detection capabilities of single cell backscatter because it allows differentiating it from noise, which does not propagate. By measuring the velocity of single cells in the suspension, for instance by means of correlation between consecutive data acquisitions, the maximum displacement of the cell for a given time can be known. Hence, the detection capabilities or sensitivity can still be improved by gating or allowing a cell velocity range (FIG. 9).

(44) 1.2.4 Sensitivity Improvement by Means of Coded Sequences of Excitation Signals

(45) In situations when the SNR is compromised and the power of the signal cannot be increased, e.g. for safety concerns, a coded excitation sequence can be used to increase the SNR. On one hand, the amount of signal energy is larger because instead of a single pulse, multiple pulses are generated. Because the signal is transmitted throughout a longer time, the power is maintained but the spatial resolution is decreased. In order to recover the loss of spatial resolution, the pulse sequence is convolved with a match filter (or coder) that produces a unique signal pattern for the excitation sequence. This is the signal that is sent to the system. In receiving mode, the signal is deconvolved or decoded. For received signals originated from other sources, i.e. noise, the decoder outputs low signal values because the signal pattern is not recognized. Contrarily, when the coded signal is received, the signal is decoded and a high narrow (correlation) peak is produced. Even in simulations with SNR close to 0, this strategy has shown to nicely resolve backscatter from single cells (FIG. 10).

(46) 1.3 Structural Measurements

(47) 1.3.1 Liquid Thickness Calculation

(48) When the body fluid is confined between two tissues, the thickness of the fluid can be estimated based on the sound speed of the signal in the fluid and the time distance of the reflected signals at the interfaces of the tissues surrounding the fluid.

(49) The signal received by the transducer should be composed by a first set of reflections from the outer tissue, a second signal representative of the content in the fluid (backscatters of cells, if any), and a last set of reflected signals from the shallowest tissue surrounding the fluid.

(50) 1.4. Operational Factors

(51) 1.4.1. Alignment Check

(52) Continuing with the above description, when this received signal structure and organization is not given, this means that at least one of the tissue components is not in the field of view of the transducer. By repositioning or reorienting the transducer, the sets of reflected signals can be recognized, which would indicate that the alignment of the transducer is correct. At this point, single-cell backscatter signals coming from the fluid can be isolated or gated and further analyzed. Note the fact of not receiving a reflection from the deeper tissue surrounding the fluid does not necessarily prevent from receiving single-cell backscatters from the fluid.

(53) 2. Signal Processing and Analysis

(54) 2.1 Backscatter Energy Analysis in a Frequency Range of Interest

(55) The returned backscattered signal is time-gated to include only the acoustic focus, which is located at a distance between 5-15 mm and has a length ranging from the hundreds of micrometers up to 2 mm. The focal volume is the region with a maximum sensitivity to cells in the fluid. The signal is Fast Fourier transformed and the power spectrum in the 15-30 MHz frequency range is analysed to calculate cell concentration and cell properties.

(56) 2.2 Improved Cell Echo Detection

(57) 2.2.1 Cross-Correlation

(58) Consecutive RF signals (or its enveloped version) are cross-correlated to find signal linear coherence due to cells. After normalization of the cross-correlation signal to the energy of the RF signals, a threshold ranging from 0.3-0.9 is used to identify cross-correlation peaks that can be attached to the echoes from cells. At acquisition rates in the order of tenths of milliseconds, the shift of cross-correlation peaks corresponding to a moving cell through the focus is used to determine cell velocity. Alternatively, the phase shift of the spectral backscatter frequency peak can be used. If a 2D-3D image is obtained by means of a linearly scanned single element transducer or a linear array or a 2D array, the Hough transform can be applied to the data as an efficient technique to detect cells, their trajectory and speed. Hough-transformed data might be later thresholded before a cell detection algorithm is used to count the number of cells.

(59) 2.2.2 Coded Excitation

(60) Coded excitation techniques (chirp, golay, barker) are used to improve the signal-to-noise ratio of the received backscatter signal. A pulse is coded by a match filter into a sequence or combination of sequences and transmitted into the body. The received signal is then deconvoluted by a duplicate of the matched filter in the receiver circuit producing a distinct spectral peak in the pulse frequency band only if the received signal signature matches the coded sequence of the match filter.

(61) 2.3 Volume Estimation

(62) 2.3.1 Cell Echo Intensity-Based Method (1D or 2D)

(63) An estimation of the sampled volume can be obtained by putting into relation the average intensity level of cell echoes in the focal region with respect to an effective focal volume previously measured at different attenuations because the larger the attenuation the more reduced the effective sampled volume. To this aim, the intensity frequency distribution of cell echoes is binned and distribution parameters such as intensity levels span over a probability threshold, e.g. 10%, are calculated. The larger the cell echoes intensity levels span the larger the focal volume that is being sampled. This relationship between intensity levels span and effective sampled volume can be calibrated for different attenuations. Alternatively, the intensity levels shift can be mapped to the effective sampled volume. The lower the cell echoes intensity levels the larger the attenuation and the smaller the effective sampled volume. Similarly, the intensity shift of cell echoes in relation to the effective sampled volume can be previously calibrated.

(64) 2.3.2 2D Array, Linearly Scanned 1D Array, 2D-Scanned Single Element

(65) If a 3D volume is obtained by means of a 2D array, a linearly scanned linear array or a 2D-scanned single element, the volume can be estimated with minimal error from the image voxel dimensions.

(66) 2.4 Cell Properties and Viability

(67) Cell concentration, size and viability can be determined from the spectral energy, the spectral bandwidth or the spectral slope of the returned cell backscattering signal.

(68) Cell backscatter energy is linearly related to cell concentration in the range 0-100 cells/L. At higher concentrations the cell backscatter is attenuated by cells located between the scattering cell and the transducer in the direction of the backscatter signal.

(69) Size is determined by cell echo width if a 3D image is obtained by relating the echo intensity with the size of the cell.

(70) Cell viability can be determined by combination of backscatter energy (measure of cell nucleus hardness) and spectral slope (measure of scatterer's size) in the frequency range of interest. There are different ways of cell death involving different structural changes of the nucleus. In apoptosis, the nucleus is condensed and fragmented reducing the size of the cell, increasing the backscattered energy, and preserving the spectral slope. In mitotic arrest/catastrophe the cell and nucleus size increase, the backscatter energy increases but the spectral slope decreases.