Open Access
Add to BibSonomy
Abstract
Smartphones are now integral to many telehealth services that provide remote patients with an improved diagnostic standard of care. The ongoing management of burn wounds and scars is one area in which telehealth has been adopted, using video and photography to assess the repair process over time. However, a current limitation is the inability to evaluate scar stiffness objectively and repeatedly: an essential measurement for classifying the degree of inflammation and fibrosis. Optical elastography detects mechanical contrast on a micrometer- to millimeter-scale, however, typically requires expensive optics and bulky imaging systems, making it prohibitive for wide-spread adoption in telehealth. More recently, a new variant of optical elastography, camera-based optical palpation, has demonstrated the capability to perform elastography at low cost using a standard digital camera. In this paper, we propose smartphone-based optical palpation, adapting camera-based optical palpation by utilizing a commercially available smartphone camera to provide sub-millimeter resolution imaging of mechanical contrast in scar tissue in a form factor that is amenable to telehealth. We first validate this technique on a silicone phantom containing a 5 × 5 × 1 mm3 embedded inclusion, demonstrating comparative image quality between mounted and handheld implementations. We then demonstrate preliminary in vivo smartphone-based optical palpation by imaging a region of healthy skin and two scars on a burns patient, showing clear mechanical contrast between regions of scar tissue and healthy tissue. This study represents the first implementation of elastography on a smartphone device, extending the potential application of elastography to telehealth.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Burn injuries affect approximately 9 million people globally per year and result in over 100,000 deaths [1]. Persistent inflammation and increased mechanical strain on a healing wound leads to the formation of pathological scars, characterized by a palpable increase in scar stiffness and thickness [2,3]. In this process, excessive collagen is deposited (fibrosis) during the healing process, leading to permanent changes in skin structure and extensibility, which is associated with persistent pain and reduced quality of life [4–6]. The optimal treatment method for burn scars includes face-to-face assessment by a clinical expert. Assessments are typically performed several weeks apart in the first two years following an injury, and the burn is monitored over the patient’s lifetime, depending on the severity of the scarring. This is a costly, inconvenient and distressing process for patients living in rural and remote communities due to lengthy travel times and time spent away from home. In the USA, 75% of all residents live more than one hour’s drive from their nearest verified burns clinic, with coverage lowest in southern states [7]. In Australia, all 15 burns units are located in capital cities [8], despite one-third of the population living outside of these regions. As similar scenarios exist in many countries, telehealth services have been developed globally so that patients can be managed within their local community under the guidance of specialists [9]. Current telehealth services are typically performed over video, aided by a local healthcare practitioner who assists the patient and assesses scar stiffness and texture via manual palpation, relaying this information to a remote specialist [9]. Palpation is an important aspect of burn scar assessment, as the scar mechanical properties indicate the severity of scarring which is critical for prescribing treatments [10,11].
Currently, the most prominent clinical burn scar assessments are the Vancouver Scar Scale and the Patient and Observer Scar Assessment Scale (POSAS), the latter of which rates scars on a scale from 1-10 on several visual and palpable variables as well as pain and itch, where 1 indicates normal skin and 10 indicates severe scarring [10,12]. Despite being widely used, these scales are subjective, thus ratings can vary greatly between assessors with different levels of experience [13,14]. This presents a limitation for telehealth as the local practitioners who assist with assessments often lack the experience and expertise of burns specialists from dedicated clinics, leading to ambiguous measurements. A number of objective burn scar assessment methods have been developed to characterize scar appearance [15,16], evaluate scar geometry [17], measure blood flow [18] and measure mechanical properties [19]. While these objective measurement tools may potentially enable non-specialists to assess scars in remote settings, they are typically expensive, bulky and impractical to use, and as such are not feasible for many telehealth environments. Therefore, a method to characterize inflammatory biomarkers, objectively, for telehealth may help to identify early signs of pathological healing allowing timely interventions such as laser therapy [20] or massage therapy [21] to be implemented to minimize inflammation, maximize healing and reduce the degree of scar tissue formed.
Optical elastography is a family of techniques that utilize optical imaging to map the mechanical properties of tissue [22,23]. This approach has the potential to quantify burn scar inflammation and fibrosis due to the combination of relatively high spatial resolution and high sensitivity to subtle changes in mechanical properties. While much of the focus of optical elastography has been in oncology [23–25] and ophthalmology [26–29], it has also been applied to imaging of the mechanical properties of skin [30–35]. One such technique is optical palpation, where a pre-characterized compliant silicone layer is placed on the surface of the skin and compressed under an external load [36]. The deformation of the compliant layer is detected using optical coherence tomography (OCT) allowing for the measurement of layer strain. The stress in the sample is then determined from the measured layer strain through knowledge of the mechanical characterization curve of the silicone layer, enabling the generation of a two-dimensional (2-D) stress map, termed an optical palpogram, indicating the relative stiffness of features in the tissue. This technique provides a spatial resolution of ∼200 µm and has demonstrated strong mechanical contrast between skin lesions, scars and surrounding normal skin [32,36]. Despite these promising results, OCT-based optical palpation has similar limitations to existing scar evaluation methods, as it requires a relatively costly and bulky imaging system which would preclude its use in many telehealth applications. More recently, an optical palpation technique that utilizes a digital camera rather than an OCT system has been proposed. This technique, known as camera-based optical palpation (CBOP), detects the light transmitted through a porous silicone layer under compression. Of key importance, as compression increases, the pore size decreases and, therefore, refractive index differences between silicone and pores within the layer reduce, allowing more light to be transmitted through the layer. By pre-characterizing the relationship between the layer stress and optical transmission, digital photographs can be converted to optical palpograms. It has previously been demonstrated that CBOP can identify the boundaries between healthy adipose tissue, fibrous tissue and tumor in freshly excised human breast tissue specimens [37]. CBOP is potentially well-suited to burn scar assessment in telehealth scenarios, as it can be implemented using lightweight and cost-effective hardware components.
In this paper, we present the first demonstration of optical elastography on a smartphone by adapting CBOP to a Google Pixel 3 smartphone. This technique, termed smartphone-based optical palpation (SBOP), has the potential to aid in the assessment and diagnosis of burns scars by probing mechanical contrast in scars and represents a step towards developing optical elastography for telehealth applications. Here, we describe the components of SBOP and outline the imaging procedure. We then validate the accuracy of SBOP by imaging a 5 × 5 × 1 mm3 inclusion embedded in a silicone phantom and compare the image quality between both mounted and handheld systems. Finally, we present preliminary in vivo SBOP results from a healthy region of skin and two burn scars and, in each case, co-locate the stress in the optical palpograms to corresponding photographs of the skin.
2. Materials and methods
2.1 Smartphone-based optical palpation setup
SBOP can readily be implemented using any smartphone model with a back-facing camera. In our SBOP setup, we utilized a Google Pixel 3 containing a 12.2 megapixel (MP) in-built back-facing camera to acquire photographs of the optical transmission through the porous layer. This camera has a 1/2.55” dual-pixel sensor (Exmor IMX363 RS, Sony Corporation, Japan) and an in-built f/1.8 lens. To optimize the camera for burn scar applications, we increased its magnification by attaching a 10x magnification external macro lens (Moment Inc., USA) which has a 25 mm focal length and a working distance of 18.5 mm. The pixel size at the focus in the specimen (object) plane is 7.9 × 7.9 µm which provides high-resolution imaging of the porous layer compression over a maximum field of view of 32 × 24 mm2. The smartphone is housed in a three-dimensional (3-D) printed plastic case to which the macro lens is attached through a twist-and-lock mechanism, as shown in the schematic in Fig. 1(a). A glass window housed in a 3-D printed mount is in turn attached to the case over the lens so that the external face of the window is positioned at the focus of the camera. This window is used to apply compression to the sample. All parts are designed using SolidWorks 2019 (Dassault Systèmes, France) and 3-D printed using a Form 2 stereolithography printer (Formlabs, USA). A custom built 30 mm diameter light-emitting diode (LED) ring provides white light illumination to the sample through eight equally-spaced 0603 surface mount diode LEDs (Inolux Corporation, USA) which are fixed within the mount and powered by a USB-C connection to the smartphone. The undiffused LEDs create bright back-reflections at the edges of the glass window which produce imaging artifacts in the optical palpograms. To remove the effect of these artifacts, optical palpograms are generated over a smaller 14 × 14 mm2 field of view. Photographs of the assembled system are shown in Figs. 1(b) and 1(c).
Fig. 1. Diagram of the SBOP system. (a) An exploded diagram of the SBOP system showing the phone case, macro lens, inner cap and outer cap, LED ring and glass window. Photographs of the assembled SBOP system taken from (b) side-on and (c) front-on.
To generate optical palpograms, we place both a porous silicone layer and a green-dyed homogeneous silicone layer between the sample surface and the glass window (Fig. 2). In an uncompressed state, refractive index differences between the silicone and pores result in most of the white-light provided by the LEDs being reflected. In this case, the image of the layer captured by the camera is predominantly white. As compression increases, the pores collapse, reducing in both size and number. In addition, the total thickness of the layer reduces. These effects cause less light scattering from the pores, and in turn, more light to transmit through the layer and reach the green layer below where mostly green light is reflected, resulting in an image with a more intense green color [37]. This is illustrated in the inset of Fig. 2. For a heterogeneous sample, stiffer features will cause the layer to compress more than softer features, producing differences in the optical transmission of the captured images. The relationship between optical transmission and mechanical stress is established prior to imaging using a uniaxial compression test of the porous layers. This characterization curve is used to convert the acquired digital photographs into 2-D en face optical palpograms. Images are acquired using Open Camera (Open Camera, Mark Harman), an open-source camera application for Android devices. Open Camera allows for adjustments to be made to the white balance, ISO and shutter speed to enhance the contrast in transmission based on the illumination from the LEDs. A burst of three photographs are acquired per location at 60 frames per second, which are averaged together to remove the effect of camera noise.
Fig. 2. Working principle of SBOP. The imaging window on the outside of the outer cap compresses the porous and green silicone layers against a tissue sample. The inset shows that as compression is increased, the voids in the porous layer collapse, allowing more light to transmit and be detected by the smartphone camera.
2.2 Signal processing
Signal processing is performed using MATLAB 2016A (MathWorks, USA). Temporal averaging of the three photographs acquired at each location is performed, creating a 2-D photograph of the average red-green-blue (RGB) values. A median filter with a 250 × 250 µm kernel size is applied to the averaged photograph to further reduce the effect of noise in the image. The size of the median filter kernel was selected to be approximately half the resolution in the original optical palpogram. As the porous layer is compressed, the amount of light received by the sensor through each of the RGB channels is different due to the green layer placed below the porous layer. The light transmission through the porous layer is quantified by the relationship between the individual RGB color channels in the processed digital photograph. In previous work, this quantification was achieved by measuring the saturation in the image according to the following equation [37]:
Fig. 3. Histograms of the red, green and blue color channels captured from photographs of the layers under uniform increments of compressive strain for (a) CBOP and (b) SBOP. Both measurements were taken from 100 × 100 pixel central regions of the acquired images. Due to the lower cost hardware where the sensitivity of each of the RGB channels is fixed, SBOP does not present sufficient red channel intensity reduction and color channel separation with increasing strain as CBOP does, thereby requiring a tailored expression for color contrast (Eq. (2)).
Figure 4 illustrates the procedure for converting a color contrast map into an optical palpogram for the case of a silicone phantom containing a 5 × 5 × 1 mm3 inclusion that is stiffer than the bulk material. The color contrast image in Fig. 4(a) is generated by applying Eq. (2) to each pixel in the RBG photograph. The resulting optical palpogram shown in Fig. 4(b) is produced by using the stress-color contrast characterization curve in Fig. 4(c) as a look-up table to perform a pixel-by-pixel transformation of the optical transmission image. The detail of this layer characterization will be described in Section 2.4. To verify the accuracy of this technique, we used SBOP to measure the stress of a homogeneous silicone phantom with a 2 mm thickness and 20 mm diameter over 0-60% strain. We compared the resulting stress-strain curve to one generated using a uniaxial compression tester which measured stress via a load cell (LSB205, Futek Inc., USA) as shown in Fig. 4(d). In both cases, a motorized translation stage applied strain to the sample which was determined by measuring the initial thickness of the sample and the displacement applied by the stage. The overall trend of the stress-strain curves is similar, with a root mean square error of 0.9 kPa, indicating that SBOP has a high level of accuracy compared to the uniaxial compression tester.
Fig. 4. Optical palpogram and characterization of SBOP. (a) Color contrast image showing the varying optical transmission between inclusion and bulk regions of the phantom and (b) the corresponding optical palpogram, where the stress values are determined from (c) the stress-color contrast characterization curve, where the standard deviation of color contrast is shown in red. (d) A comparison between the measured stress-strain curves of a homogeneous silicone test target acquired using a uniaxial compression tester (UCT) and SBOP.
2.3 Layer fabrication
A detailed explanation of the porous layer fabrication process is provided in [37]. Here, we briefly describe the main steps. The porous layers are fabricated by mixing equal parts silicone resin, crosslinker (Elastosil P7676, Wacker, Germany) and polydimethylsiloxane (PDMS) oil with fine sugar crystals (particle size, 30-100 µm) at a concentration of 0.25 g/ml. The mixture is cured in an oven for 30 minutes before the sugar particles are leeched out in a water bath for up to 48 hrs. Finally, the layers are dehydrated to remove residual water before being cut into 20 mm diameter disks with a thickness of 1 mm. The green layers are made by combining one part of green silicone-based pigment (SP-Green, Barnes Products, Australia) with four parts of Elastosil P7676 silicone elastomer and are fabricated to a thickness of 0.5 mm and a diameter of 35 mm. The green layer has a larger diameter than the porous layer to prevent the porous layer from sliding off the edge of the green layer under compression. The porous and green layers exhibited a combined elasticity of 8.2 ± 1.5 kPa at 20% compressive strain.
The three key design parameters of layer material, porosity and thickness determine the elasticity range, sensitivity and spatial resolution that can be measured in an optical palpogram. In general, the elasticity of the silicone material should match that of the imaged tissue type. Furthermore, there exists a trade-off between spatial resolution and sensitivity as thinner layers will provide higher resolutions, however, small variations in color contrast are better detected in thicker layers. While the values specified above for these parameters are tailored towards imaging burns scars, other clinical applications are likely to require different specifications and, therefore, changes to the layer material, porosity and thickness.
2.4 Layer characterization
To convert the digital photographs to optical palpograms, the porous layers must first be optically and mechanically characterized. Both the porous layer and green layer are placed on a uniaxial compression tester with the smartphone camera positioned above to capture images of the layer transmission as it is compressed against the imaging window of the SBOP setup. Strain is applied by an axially orientated linear translation stage at a rate of 0.1% s-1 until 70% strain is reached. At each 10% strain interval, three photographs are acquired, which are later averaged to remove the effect of noise from the camera sensor. A load cell records the applied force which is converted to stress using prior knowledge of the layer surface area and a resultant stress-color contrast characterization curve of the layers is produced. As the layers are fabricated in batches, a characterization is performed for each batch to account for slight variations in the fabrication procedure. Repeatability of the porous layers is measured by comparing the mean standard deviation in optical and mechanical properties of three different layers from 0-60% applied strain. The mean standard deviation in mechanical stress is 0.17 kPa (corresponding to ∼1% of the maximum stress) and the mean standard deviation in color contrast is 0.03 across the applied strain range. These results suggest a high level of repeatability between different porous layers.
2.5 Inclusion phantom fabrication
A silicone inclusion phantom is used to validate SBOP in both mounted and handheld configurations. The phantom is a 2 mm-thick cylinder with a diameter of 15 mm. The bulk of the phantom is made from Elastosil P7676 (Wacker, Germany), a two-part silicone elastomer in a 1:1 mixing ratio between resin and crosslinker. A 5 × 5 × 1 mm3 stiff silicone inclusion, situated at the surface of the phantom, is made by mixing Elastosil RT601 (Wacker, Germany) with PDMS oil (Wacker, Germany) in a 10:1:10 ratio between the resin, crosslinker and oil. The elasticity of the bulk and inclusion are 18.7 ± 1.2 kPa and 277.0 ± 10.9 kPa at 20% strain, respectively. The inclusion and bulk materials were selected as their mechanical properties corresponded well with those of burn scar (∼200 kPa) and unscarred tissue (∼10 kPa) [30].
2.6 Clinical protocol
Two patients with flame burn scars were recruited to this study and imaged in the State Adult Burns Clinic at Fiona Stanley Hospital, Western Australia. Palpation was performed on the scar site by a clinician and the POSAS score was recorded for both scar stiffness and thickness. A photograph of the tissue was also acquired. To assist with co-location between the photograph and the optical palpogram, three fiduciary marks were drawn on the patient using a skin-marker pencil, which corresponded with the marks at the same locations on the outer cap. Following this, the green and porous silicone layers were placed onto the surface of the scar with two drops of silicone oil applied between each interface to allow the layers to freely expand under compression, reducing the effects of friction. The clinician, while holding the SBOP system in one hand, positions the imaging window against the layers and applies compression whilst acquiring a set of photographs used to generate optical palpograms of the scar and surrounding tissue. A single optical palpogram was acquired at each scar site due to time constraints during scanning.
This study was approved by South Metropolitan Health Services Ethics Committee (EC00265) in Western Australia and informed consent was obtained from the patients prior to imaging. All methods and procedures were performed in accordance with the relevant guidelines and regulations, including following good clinical practices described at the International Conference on Harmonisation [38].
3. Results
3.1 Inclusion phantom imaging
To validate our technique, we imaged an inclusion phantom described in Section 2.5. Firstly, we performed SBOP with the phone mounted to a rigid post and applied compression by driving the sample on an axial translation stage to remove any hand motion artifacts that may be caused while holding the smartphone. Strain was applied incrementally by controlling the position of the stage until the total compression of the layers and the sample was 60%. The resulting optical palpogram is displayed in Fig. 5(a), where the inclusion, which exhibited a stress of 23.7 ± 0.5 kPa, can clearly be distinguished from the surrounding region of bulk which displayed a stress of 12.5 ± 0.7 kPa. To determine the effectiveness of temporal averaging on reducing digital noise, ten photographs were acquired in the mounted regime and averaged together before being converted to optical palpograms. Contrast-to-noise ratio (CNR) and spatial resolution are reported as they provide a measure of how variations in mechanical properties between healthy and scar tissue can be detected above the noise (CNR) and the smallest spatial change in mechanical properties (resolution). CNR in each of these averaged optical palpograms was measured at three separate 400 × 400 µm locations in both the inclusion and the bulk and the average CNR from these locations is plotted in Fig. 5(b). CNR is defined as:
where µinc and µbulk are the mean stress values in the inclusion and bulk respectively, and σinc and σbulk are the standard deviations. The CNR increases with more photographs averaged, however, there is negligible improvement beyond averaging three photographs. Freehand results were taken by applying compression with the phone held in the user’s hand. Once again, the inclusion is clearly distinguishable from the surrounding bulk, as shown in Fig. 5(c), and average stresses are measured to be 22.5 ± 0.3 kPa and 13.9 ± 0.4 kPa, respectively. Furthermore, by analyzing the CNR of optical palpograms (Fig. 5(d)), generated following the same procedure as described for Fig. 5(b), it was found that the optimal number of photographs for temporal averaging was also three. Overall, the CNR reduces by ∼20% for handheld operation, compared to the mounted setup. In addition to CNR, the resolution degrades slightly from 430 ± 20 µm while mounted, to 540 ± 20 µm for handheld operation. Of note, the resolution of a single optical palpogram generated during handheld operation is 430 ± 100 µm, closely matching the resolution measured for mounted operation. This result strongly suggests that lateral motion resulting from the additional time required to perform averaging causes the resolution to reduce to 540 ± 20 µm. Despite this lower resolution, we performed averaging of three photographs for the handheld operation results presented in Figs. 6–8, as this results in an increased CNR, as observed in Fig. 5(d).Fig. 5. Comparison of mounted and freehand SBOP images taken on a 5 × 5 × 1 mm3 silicone inclusion phantom. Optical palpogram from (a) SBOP in a mounted configuration and (b) the effect of increased temporal averaging on the CNR while mounted. Optical palpogram from (c) SBOP for freehand operation and (d) the CNR for increased temporal averaging for freehand operation. Error bars in (b) and (d) represent one standard deviation in CNR.
3.2 In vivo skin imaging
The in vivo imaging capability of SBOP was demonstrated on an unscarred site which served as a control, in addition to two different burn scars, obtained from the same patient. The control image was taken from the dorsal side of the left forearm which was unscarred and provides a baseline optical palpogram for unscarred tissue. A photograph of the unscarred region can be seen in Fig. 6(a). A zoomed-in subsection of the photograph is shown in Fig. 6(b) which corresponds to the same field-of-view as in the optical palpogram, presented in Fig. 6(c). There is no sign of scarring in the photographs which is consistent with the lack of features detected in the optical palpogram. There is a slight gradient in stress from the top left corner to the bottom right corner, which was likely caused by holding the smartphone at a slight angle relative to the tissue surface while acquiring the image. The distinct lack of features in the optical palpogram in Fig. 6(c) suggests that imaged region was relatively homogeneous which is indicative of unscarred tissue [32].
Fig. 6. Photograph and optical palpogram of the unscarred site. (a) The full-size photograph of the unscarred site where there is no scar present. (b) The zoomed-in section of the photograph that corresponds to the location where (c) the optical palpogram was acquired.
The first burn scar selected for imaging was located on the dorsal side of the right hand. The scar was palpated prior to SBOP by a burns specialist to confirm the presence of mechanical contrast. A score of 3 for stiffness and 4 for thickness was given based on the POSAS, which indicates moderate scarring. The scar site is shown in Fig. 7(a) where a region of scarring is visible on the right-hand-side of the image and unscarred tissue is present on the left-hand-side. Figure 7(b) shows the zoomed-in region of the scar that corresponds to the same field-of-view as the optical palpogram (Fig. 7(c)). The optical palpogram highlights two clear regions of elevated stress on the upper right side and the lower right side of the image in Fig. 7(c). The mechanical heterogeneity in the optical palpogram corresponds with the boundaries seen in the photograph, suggesting that these features are stiffer than the surrounding regions of the scar.
Fig. 7. Photograph and optical palpogram of a burn scar located on the back of the right hand. (a) The full-size photograph of the scar site and surrounding tissue. (b) The zoomed in section of the photograph that corresponds to the location where (c) the optical palpogram was acquired.
A second scar site located on the dorsal side of the patient’s right forearm was also imaged with SBOP. This scar, shown in Fig. 8(a), was approximately 30 mm long and 10 mm wide and was also palpated by a burns specialist prior to imaging, receiving POSAS scores of 5 for stiffness and 2 for thickness, which are indicative of moderate scarring. SBOP was performed at a region where more significant scarring was located next to a region of lessor scarring to demonstrate the difference in mechanical contrast between the two regions. Figure 8(b) shows the photograph of the imaged region, where the scarred tissue at the top of the image presents as darker than the less scarred skin. Figure 8(c) shows a region of elevated stress towards the top of the optical palpogram, which corresponds well with the approximate region of scar tissue in Fig. 8(b). There is a small imaging artifact on the right-hand-side of the optical palpogram which appears as a region of low stress, marked by a black arrow in Fig. 8(c). This is likely caused by the presence of air bubbles between the green and porous layers. Note that while the POSAS score indicated that this scar is stiffer than the scar imaged in Fig. 7, the relative stress is lower. This is potentially due to different loading mechanisms between manual palpation and SBOP. In addition, manual palpation is a subjective method which can lead to ambiguities between assessments. Despite the lower stress value, contrast is still clearly distinguishable and corresponds well with the scar in the photograph.
Fig. 8. Photograph and optical palpogram of a burn scar located on the right forearm. (a) The full-size photograph of the scar site and surrounding tissue. (b) The zoomed in section of the photograph that corresponds to the location where (c) the optical palpogram was acquired. The black arrow in (c) identifies an imaging artifact due to air bubbles between the layers.
4. Discussion
In this paper, we have presented the first demonstration of optical elastography on a smartphone device. This technique builds on a previous imaging technique developed by our group, CBOP, and utilizes the in-built smartphone camera of a Google Pixel 3 to provide sub-millimeter resolution imaging of mechanical contrast between regions of in vivo burn scars and adjacent tissue. In utilizing smartphone devices, we have developed a tool that provides elastography in a widely accessible format. Coupled with the low cost of the additional components, SBOP is well positioned to provide mechanical contrast, in addition to the visual and audial inspection already utilized in a range of telehealth applications [9,39,40].
While the camera used in SBOP was less expensive than that used previously in CBOP, it was also of a lower quality, with a smaller 1/2.55” (5.6 × 4.2 mm2) sensor in the Google Pixel 3 compared to the 1/1.7” (7.4 × 5.6 mm2) sensor used previously [37]. Despite both cameras providing resolutions of 12.2 MP, the smaller camera sensors typically provide lower dynamic range and higher noise levels as the individual pixels detect less light, resulting in lower quality images [41]. When comparing the relative spatial resolution of the two techniques, both SBOP and CBOP reported spatial resolution of 430 µm when imaging a 5 × 5 × 1 mm3 stiff inclusion in a silicone phantom. While this is not a direct comparison as the axial position of the inclusion differed slightly between the two measurements [36], it confirms that the resolution of these camera-based techniques is not limited by the camera sensor but rather by the mechanical deformation of the porous layer and the sample. Therefore, using low-cost smartphone cameras and lenses for telehealth applications will not lead to lower image resolution than higher specification benchtop implementations. Furthermore, the spatial resolution of SBOP is significantly higher than that of manual palpation which has previously reported resolutions of 1-3 mm [42], suggesting SBOP is able to identify finer features within burn scars than the current clinical practice. Despite the resolution being unaffected, there is a drop in CNR from CBOP to SBOP, most likely caused by the higher noise in the smartphone camera sensor and hand motion artifacts when operated freehand. This indicates that the more sophisticated camera used in CBOP can generate higher quality images and mechanical contrast, highlighting the trade-off between CNR and the camera system cost.
In addition, the CNR in the handheld image also reduces by ∼20% when compared to mounted SBOP, likely due to a difference in friction conditions between the two scans, which is difficult to match exactly. The mounted set-up is uniaxial; however, it was observed that lateral motion in the handheld set-up (particularly when initially searching for the region containing the inclusion) can cause the lubricating oil to be squeezed out, resulting in increased friction. Increased friction restricts the lateral expansion of the silicone layers under compression, especially at higher preloads, resulting in reduced axial strain on the porous layer for the same stress and, consequently, an underestimation of stress at the inclusion [43]. While the reduction in image quality does not adversely affect the ability to detect features in the handheld case, there is scope to improve the image quality by incorporating an accurate estimation for the friction coefficient into the measurement of stress. This could be achieved by comparing experimental data to finite element analysis (FEA) simulations to match the friction conditions, from which a computational calculation of 3-D stress in the model layer could be acquired, providing improved spatial resolution and higher image quality [43].
Stress has previously been used as a contrast mechanism in OCT-based optical palpation to distinguish malignant disease [36], however, it is a qualitative measure and does not quantify elasticity. A previous study conducted on 34 freshly excised breast tissue specimens demonstrated that despite this, optical palpograms can localize tumor based on stress contrast [44]. This approach, however, requires mechanical heterogeneity within the field of view of a single image. For SBOP, and indeed all techniques utilizing stress as a contrast mechanism, this can lead to ambiguities in cases where there is no heterogeneity, as an entirely stiff sample could appear the same as an entirely soft sample. In this study, we have mitigated this by imaging scar sites at the boundary between soft tissue and stiff scar tissue. Additionally, we aimed to keep the applied loading on the tissue relatively constant, such that optical palpograms were generated at a similar point in the stress-strain curve of different tissues, facilitating direct comparison. In the future, methods could be developed to quantify the mechanical properties of the tissue using inverse methods [45,46] or by utilizing an indentation model to approximate elasticity [47,48]. Alternatively, the average stress level in the image could be computed [33] in real-time and used as a trigger for palpogram acquisition. These methods would allow SBOP to better distinguish healthy and diseased tissue as well as more accurately compare scars from different optical palpograms, expanding the clinical feasibility of the technique.
Both scar sites imaged in this study exhibited regions of raised and uneven tissue at the surface. While the thickness of the scar can be used as a diagnostic indicator and is indeed one of the criteria for the POSAS, it is important that our measurements are not incorrectly generating mechanical contrast based on thickness variations. Thickness, however, is unlikely to have produced the contrast in stress observed in the in vivo skin optical palpograms presented in the paper, as it has been shown previously that the raised topology of surface features does not lead to an overestimation of stress in OCT-based optical palpation [32].
Tilt is common source of artifacts in handheld imaging probes as it is often difficult to precisely control the orientation of the probe [49]. In SBOP, tilt causes a gradient of stress such as that shown in Fig. 6(c), and in extreme cases could lead to incorrect classification of a region of homogeneous non-scarred tissue as scarred. In this initial proof of concept, we utilized Open Camera, an open-source application to acquire images. This application is not designed to detect the tilt of the camera. In future studies, we seek to develop our own specialized camera application which will incorporate the phone’s in-built gyroscope to sense tilt in the camera. This would serve as a simple solution to address tilt as it would not require any additional components, while improving the overall image quality.
A limitation in the clinical application of the current SBOP method is in ensuring that the porous layer undergoes compression sufficient to generate optical palpograms. This is readily achieved when the sample is positioned between a rigid flat plate and the SBOP imaging window, such as in the case of the inclusion phantoms, as it allows for uniform compression to be applied. For in vivo scans, however, this may not always be possible depending on the location of the scar site on the body. For example, heightened stress may be observed in the regions above stiff features such as tendons or bones. Similarly, when imaging scars located above large regions of soft tissue, such as the abdomen, or regions with a lot of adipose tissue, the underlying tissue displaces considerably, making it difficult to generate compression in the layer. As a result, the contrast in the current iteration of SBOP is dependent on the anatomical location from which the scan was taken. This can be seen when comparing Figs. 7(c) and 8(c), where the stress is higher in Fig. 7(c) which was acquired on the back of the hand where there is less underlying soft tissue between the scar and skeletal structure. In contrast, Fig. 8(c) was taken from the arm where the underlying adipose tissue and muscle tissue was several centimeters thicker. As a result, there is less contrast between the scar and unscarred regions of the skin in this optical palpogram. One approach to mitigate these effects is to reduce the elasticity of the layer so that it is softer than the underlying tissue which would allow the layer to deform more easily. Another potential solution is to reduce the diameter of the imaging window and external cap which would enable the user to apply a more localized force, similar to an indenter, again making it easier to deform the layer and generate contrast while reducing the potential for erroneous measurements caused by spatial variations in the underlying features and tilt [32].
The results presented in this initial work demonstrate that SBOP is capable of distinguishing mechanical heterogeneity in burn scars. While this work has demonstrated strong potential towards telehealth applications, to fully understand the clinical feasibility of this technique, a longitudinal study is required. Such a study would seek to image multiple different scar sites and provide a measure of the inter-sample variability by taking repeat optical palpograms on the same scar site as well as determine the inter- and intra-operator variability. In addition, a detailed comparison between SBOP and POSAS on multiple scar sites would provide an indication of the diagnostic performance of SBOP relative to the current clinical standard.
5. Conclusions
In this paper, we have presented a proof of concept for SBOP, a novel optical palpation device designed to evaluate tissue stiffness on a commercially available smartphone. We have demonstrated this technique on a structured inclusion phantom in both a mounted and handheld setup and shown that there is minimal difference in image quality between the two. In addition, we have demonstrated mechanical contrast in two in vivo burn scars using SBOP, highlighting potential application in telehealth treatment of burns. SBOP uses low-cost and readily available components in conjunction with commercially available smartphones making it easily accessible to the public.
Funding
Australian Research Council; Cancer Council Western Australia; War Widows' Guild of Western Australia.
Author contributions
R.W.S., Q.F., A.C., H.M.D. and B.F.K. conceived and designed the experiments; R.W.S, Q.F., A.C., H.M.D. and B.F.K. were involved in the acquisition and interpretation of the data; F.M.W and H.M.D provided clinical guidance and expertise; A.T. provided technical assistance and expertise; all authors wrote, reviewed and/or and edited the manuscript; Q.F., A.C. and B.F.K. supervised the project.
Disclosures
B.F.K. and A.C. hold shares in OncoRes Medical, a startup company developing optical coherence elastography for surgical applications. The other authors declare no conflicts of interest related to this article.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. S. L. James, L. R. Lucchesi, C. Bisignano, C. D. Castle, Z. V. Dingels, J. T. Fox, E. B. Hamilton, N. J. Henry, D. McCracken, N. L. S. Roberts, D. O. Sylte, A. Ahmadi, M. B. Ahmed, F. Alahdab, V. Alipour, Z. Andualem, C. A. T. Antonio, J. Arabloo, A. D. Badiye, M. Bagherzadeh, A. Banstola, T. W. Bärnighausen, A. Barzegar, M. Bayati, S. Bhaumik, A. Bijani, G. Bukhman, F. Carvalho, C. S. Crowe, K. Dalal, A. Daryani, M. D. Nasab, H. T. Do, H. P. Do, A. Y. Endries, E. Fernandes, I. Filip, F. Fischer, T. Fukumoto, K. B. B. Gebremedhin, G. G. Gebremeskel, S. A. Gilani, J. A. Haagsma, S. Hamidi, S. Hostiuc, M. Househ, E. U. Igumbor, O. S. Ilesanmi, S. S. N. Irvani, A. U. Jayatilleke, A. Kahsay, N. Kapoor, A. Kasaeian, Y. S. Khader, I. A. Khalil, E. A. Khan, M. Khazaee-Pool, Y. Kokubo, A. D. Lopez, M. Madadin, M. Majdan, V. Maled, R. Malekzadeh, N. Manafi, A. Manafi, S. Mangalam, B. B. Massenburg, H. G. Meles, R. G. Menezes, T. J. Meretoja, B. Miazgowski, T. R. Miller, A. Mohammadian-Hafshejani, R. Mohammadpourhodki, S. D. Morrison, I. Negoi, T. H. Nguyen, S. H. Nguyen, C. T. Nguyen, M. R. Nixon, A. T. Olagunju, T. O. Olagunju, J. R. Padubidri, S. Polinder, N. Rabiee, M. Rabiee, A. Radfar, V. Rahimi-Movaghar, S. Rawaf, D. L. Rawaf, A. Rezapour, J. Rickard, E. M. Roro, N. Roy, R. Safari-Faramani, P. Salamati, A. M. Samy, M. Satpathy, M. Sawhney, D. C. Schwebel, S. Senthilkumaran, S. G. Sepanlou, M. Shigematsu, A. Soheili, M. A. Stokes, H. R. Tohidinik, B. X. Tran, P. R. Valdez, T. Wijeratne, E. Yisma, Z. Zaidi, M. Zamani, Z.-J. Zhang, S. I. Hay, and A. H. Mokdad, “Epidemiology of injuries from fire, heat and hot substances: global, regional and national morbidity and mortality estimates from the Global Burden of Disease 2017 study,” Inj. Prev. 26(Supp 1), i36–i45 (2020). [CrossRef]
2. I. A. Darby, B. Laverdet, F. Bonté, and A. Desmoulière, “Fibroblasts and myofibroblasts in wound healing,” Clin., Cosmet. Invest. Dermatol. 7, 301–311 (2014). [CrossRef]
3. J. A. Clark, J. C. Y. Cheng, and K. S. Leung, “Mechanical properties of normal skin and hypertrophic scars,” Burns 22(6), 443–446 (1996). [CrossRef]
4. A. L. Browne, R. Andrews, S. A. Schug, and F. Wood, “Persistent pain outcomes and patient satisfaction with pain management after burn injury,” Clin. J. Pain 27(2), 136–145 (2011). [CrossRef]
5. H. M. DeJong, M. Phillips, D. W. Edgar, and F. M. Wood, “Patient opinion of scarring is multidimensional: an investigation of the POSAS with confirmatory factor analysis,” Burns 43(1), 58–68 (2017). [CrossRef]
6. C. Öster, M. Kildal, and L. Ekselius, “Return to work after burn injury: burn-injured individuals’ perception of barriers and facilitators,” J. Burn Care Res. 31(4), 540–550 (2010). [CrossRef]
7. M. B. Klein, C. B. Kramer, J. Nelson, F. P. Rivara, N. S. Gibran, and T. Concannon, “Geographic Access to Burn Center Hospitals,” JAMA 302(16), 1774–1781 (2009). [CrossRef]
8. ANZBA, “Burn Units – ANZBA: Australian & New Zealand Burn Association,” https://anzba.org.au/resources/burn-units/.
9. T. McWilliams, J. Hendricks, D. Twigg, F. Wood, and M. Giles, “Telehealth for paediatric burn patients in rural areas: a retrospective audit of activity and cost savings,” Burns 42(7), 1487–1493 (2016). [CrossRef]
10. N. Brusselaers, A. Pirayesh, H. Hoeksema, J. Verbelen, S. Blot, and S. Monstrey, “Burn scar assessment: a systematic review of different scar scales,” J. Surg. Res. 164(1), e115–e123 (2010). [CrossRef]
11. H. Goei, C. H. van der Vlies, W. E. Tuinebreijer, P. P. M. van Zuijlen, E. Middelkoop, and M. E. van Baar, “Predictive validity of short term scar quality on final burn scar outcome using the Patient and Observer Scar Assessment Scale in patients with minor to moderate burn severity,” Burns 43(4), 715–723 (2017). [CrossRef]
12. P. D. H. M. Verhaegen, M. B. A. van der Wal, E. Middelkoop, and P. P. M. van Zuijlen, “Objective scar assessment tools: a clinimetric appraisal,” Plast. Reconstr. Surg. 127(4), 1561–1570 (2011). [CrossRef]
13. P. S. Powers, S. Sarkar, D. B. Goldgof, C. W. Cruse, and L. V. Tsap, “Scar assessment: current problems and future solutions,” J. Burn Care Rehabil. 20, 54–60 (1999). [CrossRef]
14. L. Forbes-Duchart, S. Marshall, A. Strock, and J. E. Cooper, “Determination of Inter-rater reliability in pediatric burn scar assessment using a modified version of the Vancouver Scar Scale,” J. Burn Care Res. 28(3), 460–467 (2007). [CrossRef]
15. A. J. Singer, H. C. Thode, and S. A. McClain, “Development of a histomorphologic scale to quantify cutaneous scars after burns,” Acad. Emerg. Med. 7(10), 1083–1088 (2000). [CrossRef]
16. J. M. Rawlins, W. L. Lam, R. O. Karoo, I. L. Naylor, and D. T. Sharpe, “Quantifying collagen type in mature burn scars: a novel approach using histology and digital image analysis,” J. Burn Care Res. 27(1), 60–65 (2006). [CrossRef]
17. M. C. T. Bloemen, M. S. van Gerven, M. B. A. van der Wal, P. D. H. M. Verhaegen, and E. Middelkoop, “An objective device for measuring surface roughness of skin and scars,” J. Am. Acad. Dermatol. 64(4), 706–715 (2011). [CrossRef]
18. Y. M. Liew, R. A. McLaughlin, F. M. Wood, and D. D. Sampson, “Motion correction of in vivo three-dimensional optical coherence tomography of human skin using a fiducial marker,” Biomed. Opt. Express 3(8), 1774–1786 (2012). [CrossRef]
19. S. S. L. Fong, L. K. Hung, and J. C. Y. Cheng, “The cutometer and ultrasonography in the assessment of postburn hypertrophic scar—a preliminary study,” Burns 23, S12–S18 (1997). [CrossRef]
20. B. M. Willows, M. Ilyas, and A. Sharma, “Laser in the management of burn scars,” Burns 43(7), 1379–1389 (2017). [CrossRef]
21. Y. S. Cho, J. H. Jeon, A. Hong, H. T. Yang, H. Yim, Y. S. Cho, D. H. Kim, J. Hur, J. H. Kim, W. Chun, B. C. Lee, and C. H. Seo, “The effect of burn rehabilitation massage therapy on hypertrophic scar after burn: A randomized controlled trial,” Burns 40(8), 1513–1520 (2014). [CrossRef]
22. B. F. Kennedy, P. Wijesinghe, and D. D. Sampson, “The emergence of optical elastography in biomedicine,” Nat. Photonics 11(4), 215–221 (2017). [CrossRef]
23. W. M. Allen, L. Chin, P. Wijesinghe, R. W. Kirk, B. Latham, D. D. Sampson, C. M. Saunders, and B. F. Kennedy, “Wide-field optical coherence micro-elastography for intraoperative assessment of human breast cancer margins,” Biomed. Opt. Express 7(10), 4139–4153 (2016). [CrossRef]
24. B. F. Kennedy, R. A. McLaughlin, K. M. Kennedy, L. Chin, P. Wijesinghe, A. Curatolo, A. Tien, M. Ronald, B. Latham, C. M. Saunders, and D. D. Sampson, “Investigation of optical coherence microelastography as a method to visualize cancers in human breast tissue,” Cancer Res. 75(16), 3236–3245 (2015). [CrossRef]
25. C. Li, G. Guan, Y. Ling, Y. T. Hsu, S. Song, J. T. J. Huang, S. Lang, R. K. Wang, Z. Huang, and G. Nabi, “Detection and characterisation of biopsy tissue using quantitative optical coherence elastography (OCE) in men with suspected prostate cancer,” Cancer Lett. 357(1), 121–128 (2015). [CrossRef]
26. M. A. Kirby, I. Pelivanov, S. Song, L. Ambrozinski, S. J. Yoon, L. Gao, D. Li, T. T. Shen, R. K. Wang, and M. O’Donnell, “Optical coherence elastography in ophthalmology,” J. Biomed. Opt. 22(12), 1 (2017). [CrossRef]
27. S. Wang and K. V. Larin, “Noncontact depth-resolved micro-scale optical coherence elastography of the cornea,” Biomed. Opt. Express 5(11), 3807–3821 (2014). [CrossRef]
28. M. R. Ford, W. J. Dupps Jr, A. M. Rollins, A. S. Roy, and Z. Hu, “Method for optical coherence elastography of the cornea,” J. Biomed. Opt. 16(1), 016005 (2011). [CrossRef]
29. V. Y. Zaitsev, A. L. Matveyev, L. A. Matveev, G. V. Gelikonov, A. I. Omelchenko, O. I. Baum, S. E. Avetisov, A. V. Bolshunov, V. I. Siplivy, D. V. Shabanov, A. Vitkin, and E. N. Sobol, “Optical coherence elastography for strain dynamics measurements in laser correction of cornea shape,” J. Biophotonics 10(11), 1450–1463 (2017). [CrossRef]
30. S. Es’haghian, K. M. Kennedy, P. Gong, Q. Li, L. Chin, P. Wijesinghe, D. D. Sampson, R. A. McLaughlin, and B. F. Kennedy, “In vivo volumetric quantitative micro-elastography of human skin,” Biomed. Opt. Express 8(5), 2458–2471 (2017). [CrossRef]
31. L. Bartolini, F. Feroldi, M. Slaman, J. J. A. Weda, J. F. de Boer, P. van Zuijlen, and D. Iannuzzi, “Toward clinical elastography of dermal tissues: A medical device to probe skin’s elasticity through suction, with subsurface imaging via optical coherence tomography,” Rev. Sci. Instrum. 91(7), 074101 (2020). [CrossRef]
32. S. Es’haghian, K. M. Kennedy, P. Gong, D. D. Sampson, R. A. McLaughlin, and B. F. Kennedy, “Optical palpation in vivo: imaging human skin lesions using mechanical contrast,” J. Biomed. Opt. 20(1), 016013 (2015). [CrossRef]
33. B. Krajancich, A. Curatolo, Q. Fang, R. Zilkens, B. F. Dessauvagie, C. M. Saunders, and B. F. Kennedy, “Handheld optical palpation of turbid tissue with motion-artifact correction,” Biomed. Opt. Express 10(1), 226–241 (2019). [CrossRef]
34. S. J. Kirkpatrick, R. K. Wang, D. D. Duncan, M. Kulesz-Martin, and K. Lee, “Imaging the mechanical stiffness of skin lesions by in vivo acousto-optical elastography,” Opt. Express 14(21), 9770–9779 (2006). [CrossRef]
35. X. Liang and S. A. Boppart, “Biomechanical properties of in vivo human skin from dynamic optical coherence elastography,” IEEE Trans. Biomed. Eng. 57(4), 953–959 (2010). [CrossRef]
36. K. M. Kennedy, S. Es’haghian, L. Chin, R. A. McLaughlin, D. D. Sampson, and B. F. Kennedy, “Optical palpation: optical coherence tomography-based tactile imaging using a compliant sensor,” Opt. Lett. 39(10), 3014–3017 (2014). [CrossRef]
37. R. W. Sanderson, Q. Fang, A. Curatolo, W. Adams, D. D. Lakhiani, H. M. Ismail, K. Y. Foo, B. F. Dessauvagie, B. Latham, C. Yeomans, C. M. Saunders, and B. F. Kennedy, “Camera-based optical palpation,” Sci. Rep. 10(1), 15951 (2020). [CrossRef]
38. J. R. Dixon, “The International Conference on Harmonization Good Clinical Practice Guideline,” Qual. Assur. 6(2), 65–74 (1999). [CrossRef]
39. S. Govender and M. Mars, “The use of telehealth services to facilitate audiological management for children: A scoping review and content analysis,” J. Telemed. Telecare 23(3), 392–401 (2017). [CrossRef]
40. G. Novara, E. Checcucci, A. Crestani, A. Abrate, F. Esperto, N. Pavan, C. De Nunzio, A. Galfano, G. Giannarini, A. Gregori, G. Liguori, R. Bartoletti, F. Porpiglia, R. M. Scarpa, A. Simonato, C. Trombetta, A. Tubaro, and V. Ficarra, “Telehealth in urology: a systematic review of the literature. how much can telemedicine be useful during and after the COVID-19 pandemic?” Eur. Urol. 78(6), 786–811 (2020). [CrossRef]
41. A. J. P. Theuwissen, “CMOS image sensors: state-of-the-art,” Solid-State Electron. 52(9), 1401–1406 (2008). [CrossRef]
42. H. Manning and F. Tremblay, “Age differences in tactile pattern recognition at the fingertip,” Somatosens. Mot. Res. 23(3-4), 147–155 (2006). [CrossRef]
43. P. Wijesinghe, D. D. Sampson, and B. F. Kennedy, “Computational optical palpation: a finite-element approach to micro-scale tactile imaging using a compliant sensor,” J. R. Soc., Interface 14(128), 20160878 (2017). [CrossRef]
44. W. M. Allen, P. Wijesinghe, B. F. Dessauvagie, B. Latham, C. M. Saunders, and B. F. Kennedy, “Optical palpation for the visualization of tumor in human breast tissuer,” J. Biophotonics 12(1), e201800180 (2019). [CrossRef]
45. P. S. Wellman, R. D. Howe, N. Dewagan, M. A. Cundari, E. Dalton, and K. A. Kern, “Tactile imaging: a method for documenting breast masses,” in Proceedings of the First Joint BMES/EMBS Conference (1999), Vol. 2, pp. 1131–1132.
46. S. M. Hosseini, M. Amiri, S. Najarian, and J. Dargahi, “Application of artificial neural networks for the estimation of tumour characteristics in biological tissues,” Int. J. Med. Robot. Comp. 3(3), 235–244 (2007). [CrossRef]
47. C. Pailler-Mattei, S. Bec, and H. Zahouani, “In vivo measurements of the elastic mechanical properties of human skin by indentation tests,” Med. Eng. Phys. 30(5), 599–606 (2008). [CrossRef]
48. V. L. Popov, M. Heß, and E. Willert, Handbook of Contact Mechanics: Exact Solutions of Axisymmetric Contact Problems (Springer Nature, 2019), Chap. 2.
49. Q. Fang, L. Frewer, R. Zilkens, B. Krajancich, A. Curatolo, L. Chin, K. Y. Foo, D. D. Lakhiani, R. W. Sanderson, P. Wijesinghe, J. D. Anstie, B. F. Dessauvagie, B. Latham, C. M. Saunders, and B. F. Kennedy, “Handheld volumetric manual compression-based quantitative microelastography,” J. Biophotonics 13(6), e201960196 (2020). [CrossRef]
