Development and investigation of skin melanoma phantoms for ultrasonic examination

Correspondence to: Kristina Andrėkutė, Ultrasound Institute, Kaunas University of Technology, Studentų 50, LT-51368 Kaunas, Lithuania. E-mail: kristina.andrekute@stud.ktu.lt Background. The purpose of this study was to develop skin mimicking phantoms with melanoma-like insertion with acoustic properties similar to the human skin. In order to guarantee that these phantoms are suitable for mimicking of the skin, we measured their parameters with an ultrasonography tool and compared with the human skin properties measured by other authors. Materials and methods. Distilled water, gelatin and Intralipid® 20% IV fat emulsion were mixed in different proportions thus manufacturing four models of the skin tissue. For melanoma-mimicking insertion only gelatin and distilled water were used. Ultrasonic examinations of manufactured phantoms were performed with the ultrasound system DUB-USB equipped with a mechanically scanned single-element focused transducer. Results. The speed of sound and the attenuation coefficient of all the manufactured phantoms were evaluated. This study demonstrates that the speed of sound decreases increasing concentration of Intralipid and it is close to the human skin tissue. The ultrasound velocity in the phantom material varied from 1 533.9 m/s up to 1 565.8 m/s depending on Intralipid fat emulsion concentration. The ultrasound velocity in the melanomalike insertion was 1 602.4 ± 23.5 m/s (mean ± SD). It was also found that the concentration dependent magnitude of attenuation increment matched the theoretically defined tendency. The attenuation in the range of 0.15–0.4 dB/mm/MHz was estimated in the phantoms possessing different concentrations of Intralipid. The attenuation in the melanomalike insertion region was 0.16 ± 0.02 dB/mm/MHz (mean ± SD). The magnitude of the attenuation coefficient is close to the attenuation in the human tissue. Conclusions. The four skin tissue mimicking phantoms were developed and their acoustic properties were estimated during this study. The investigation showed that the estimated speed of sound and the attenuation coefficient were close to the values being estimated in the human skin tissue. Furthermore, it was noted that the acoustic proper ties could be controlled by changing the concentration of Intralipid and such a flexible phantom could be used for mimicking of the external tissue of the human body.


INTRODUCTION
Skin melanoma is a malignant tumor that arises from melanocytic cells and primarily involves the human skin (1).Even small tumors may have a tendency towards metastasis and thus a relatively unfavourable prognosis.Melanomas account for 90% of the deaths associated with cutaneous tumours (1,2).Diagnosis of primary melanoma is based on early observation of clinical features and thickness, which are an important biomarker for the prognosis of the disease (1,3,4).The noninvasive ultrasonic diagnostics of melanoma is not always easy and is related to the experience of the clinician, accuracy of the imaging systems or image processing algorithms.Due to this reason a lot of new devices have been introduced in experimental stages for non-invasive diagnosis of skin lesions (5).In medical imaging it is necessary to have suitable phantoms, which might mimic the characteristics of the real skin.Phantoms are commonly used in the development of imaging systems and evaluation of image processing algorithms used for recognition or extraction of suspicious regions.A tissue-mimicking phantom emulates important properties of biological tissue for the purpose of providing a more clinically realistic imaging environment (5).In ultrasound imaging the most important phantom properties are the material's speed of sound and the acoustic attenuation coefficient (5).It is assumed that in the soft tissue the average speed of sound is 1 540 m/s (5,6), the attenuation coefficient have been shown to be frequency-dependent (5,(7)(8)(9).It was demons trated that the average value of attenuation in healthy human dermis at the forearm region is 0.21 dB/mm/MHz (in the frequency range of 14-50 MHz) (9).Both ultrasound velocity and attenuation are related to collagen content in the skin tissue (4,9).Previous studies (5,7,10,11) have shown that hydrogels can be efficiently used for tissuemimicking phantom design.The hydrogel phantoms are attractive for biomedical applications due to their mechanical properties.Other authors developed the skin phantoms for investigation of other non-invasive techniques (12,13).Mazolli et al. manufactured optical skin phantoms with different thickness of melanoma-like insertion and tried to mimic the optical absorption and scattering (12).However, until now the ultrasonic skin phantom mimicking pigmented skin lesion has not been created and acoustic properties have not been investigated.
The purpose of this study was to develop the skin mimicking phantoms with melanoma-like insertion with acoustic properties similar to the human skin.In order to guarantee that these phantoms are suitable for the skin mimicking, we measured their parameters with an ultrasonography tool.

MATERIALS AND METHODS
In this study the skin phantoms for experimental ultrasonic investigation were manufactured by using three different ingredients listed in Table 1.Distilled water, gelatine and Intralipid® 20% IV fat emulsion (Fresenius Kabi, Austria) were mixed in different proportions thus manufacturing different four models of the skin tissue.
The mixtures were poured into Petri dishes and were placed into a cold chamber (T = 5 °C) for about 20 minutes.The horizontal position of each dish was followed precisely using the levels.The mixtures were taken out when the solid state was achieved.We used a scalpel to make cylindrical holes of approximately 5 mm diameter and 3.8 mm thickness (up to the bottom of the dish).In the next stage the melanoma mimicking material was prepared.It was known that melanomas appear The melanoma mimicking mixtures were poured into the cylindrical holes which were made in the centres of the skin tissue mimicking phantoms.The melanoma-mimicking insertion was coloured by using liquid Indian ink (negligible amount to compare with the concentrations of the main ingredients) in order to separate visually the insertion from the surrounding media.Finally, phantoms were placed horizontally for cooling again.Therefore, the uniform thickness of the phantom was achieved.
The ultrasound system DUB-USB (Taberna Pro Medicum, Luneburg, Germany) equipped with a mechanically scanned single element focu sed transducer (central frequency of 22 MHz, bandwidth approximately 12-28 MHz, focal depth of 11 mm) was used for ultrasonic examinations of manufactured skin melanoma mimicking phan toms.The pulse-echo experimental set-up is shown in Fig. 1.The ultrasound beam was focused at the surface of the phantom as shown in Fig. 1.
The ultrasonic transducer was in the perpen dicular position during scanning of all four phantoms.The lateral scanning width was 12.8 mm, the scanning step was 0.033 mm and 382 A-scans of each phantom were acquired.The scanning duration was less than 0.4 sec per single B-scan.The radio-frequency echo signals were digitized by a 8-bit A/D converter at 100 MHz sampling rate and stored on the hard disk of a personal computer.
The acoustic properties (speed of sound and attenuation) of the phantoms were investigated.For accurate determination of the speed of sound in the layers of the phantom, the reference ultrasound time of flight (TOF) was measured by using the ultrasound system and a tank containing distilled water.The speed of sound (c ph ) was determined by measuring the duration between the echoes reflected from the surface of the Petri dish (in the case of the phantom and in the case of the insertion filled with distilled water only) and the surface of the phantom could be expressed by the following formula (5,14): where c ph is the speed of sound in the phantom, c w = 1 490.2 m/s (measurements were performed at 22.8 °C ± 0.5 °C temperature) is the temperaturedependent speed of sound in the distilled water, L 2 is the thickness of the phantom, Δt w and Δt ph are TOF of ultrasound in the water and the phantom, respectively.The speed of sound in the insertion (c i ) was evaluated at the same thickness as in the phantom (L 2 ).A convex deformation of the insertion surface appeared due to the tension which occurs during the cooling (see Fig. 2D).Thickness of the insertion (L i = c i Δt i /2) was evaluated by summing the thickness of the phantom (L 2 ) and the thickness of deformation through which TOF is Δt def .The frequency-dependent ultrasound attenuation can be estimated by using Fourier analysis of the signals.The ultrasound attenuation coefficient was estimated according to the following equation (15,16): where α ph (f) is the frequency-dependent ultrasound attenuation of the phantom in dB/mm, R is the total reflection coefficient at the water-phantom and phantom-dish interfaces, A(f) is the magnitude of the spectrum of echo reflected from the water- phantom interface, A 0 (f) is the magnitude of the spectrum of echo reflected from the phantom-Petri dish interface, L 2 is the thickness of the phantom in mm.The echo-signals from the phantom surface and the phantom bottom were gated using the Hamming window before application of the fast Fourier transform.The calculated α ph (f) dependences were linearly approximated.The regression lines were obtained using the method of least squares fitting.The acoustic energy is reflected when the ultrasound waves penetrate through the layers having different acoustic impedances.The im pedance Z = ρc depends on the ultrasound velocity (c) and density (ρ) of the scanned media (7).The acoustic reflection coefficient of ultrasonic waves from the surface of the phantom R 12 (water-phantom interface) can be expressed as follows (7): where Z 1 and Z 2 are the acoustic impedances of water and the phantom, respectively.The transmission coefficient of ultrasonic waves through the waterphantom interface (T 12 ) can be expressed by the following formula: The total reflection coefficient in the three layer medium was expressed as follows: where R 23 is the reflection coefficient of ultrasonic waves reflecting from the interface between the phantom and the Petri dish (calculated in a similar way as (3)) and T 21 is the transmission coefficient of ultrasonic waves through the phantom-water interface.The density of the Petri dish is 1.18 g/ml and the speed of sound is 2 672 m/s (Petri dish material poly(methyl methacrylate)) (17).

RESULTS
The speed of sound and the attenuation coefficient of all the manufactured phantoms were evaluated.
The acquired ultrasound B-scans and A-scans signals which were used for the evaluation are presented in Fig. 2.
The ultrasound velocity dependence on Intralipid fat emulsion concentration is presented in Fig. 3.The ultrasound velocity in the phantom ma terial varies from 1 533.9 m/s up to 1 565.8 m/s depending on Intralipid fat emulsion concentration.The ultrasound velocity in the melanoma-like insertion was 1 602.4 ± 23.5 m/s (mean ± SD).As expected, it was observed that the speed of sound decreases increasing concentra tion of Intralipid.Accordingly, this acoustic property could be con trolled in the process of physical modeling of the human skin.It could be noticed that the speed of sound in the tissue mimicking phantom is close to that of the real skin tissue (Table 2).
The acoustic attenuation was determined in the range from 12 up to 28 MHz.Fig. 4 shows the dependence of the attenuation versus frequency in skin tissue phantoms with different concentrations of Intralipid and insertion.It could be summarized that the concentration dependent magnitude and the slope of attenuation increment match the theoretically defined tendency.Attenuation in the range of 0.18-0.4dB/mm/MHz was estimated in the phantoms with different concentrations of Intralipid.An attenuation in the melanoma-like insertion region was 0.16 ± 0.02 dB/mm/MHz  (mean ± SD).The magnitude of the attenuation coefficient is close to that of the human tissue measured by other authors (5,8,9).

DISCUSSION AND CONCLUSIONS
The four skin tissue mimicking phantoms were developed and their acoustic properties were estimated during this study.The investigation showed that the estimated speed of sound and the attenuation coefficient were close to the values being estimated in human skin tissues.Furthermore, it was noted that the acoustic properties could be controlled by changing concentration of Intralipid and such a flexible phantom could be used for mimicking of various areas of the human body.These phantoms will be used in development and evaluation of the measurement methods and image processing algorithms in further studies with the aim to perform the early stage melanoma detection and evaluation.
A limitation of this study is that only four phantoms with different concentrations were manufactured and investigated.The repeatability of phantom properties will be estimated in forthcoming studies when the acoustic properties of a few phantoms having the same content will be estimated.Also, long-term stability of the acoustic properties of the phantoms was not evaluated in this study.Usually, phantoms should be stored at low temperatures for maintaining their mechanical stability since gelatin dissolves in a warm environment.But it should be mentioned that our expectation was to manufacture the phantom to be used for acquisition of B-mode images which will be pressed further, therefore it is sufficient if the phantom will be stable as long as the scanning data of B-mode image will be recorded.

Fig. 1 .
Fig.1.Experimental set-up for the speed of sound measurements.The average thickness of the phantom (L 2 ) was 3.8 ± 0.56 mm, the distance up to the focus (L 1 ) 11 mm, all three materials had different acoustic impedances -Z 1 (acoustic impedance of water), Z 2 (acoustic impedance of phantom), Z 3 (acoustic impedance of Petri dish)

Fig. 2 .Fig. 3 .
Fig. 2. Ultrasound B-scan images and A-scan signals of the phantoms: A -the B-scan image of the region with a cylindrical hole containing distilled water; B -the reflected A-scan signal from the water-phantom interface; C -the reflected A-scan signal from the water-dish interface; D -the B-scan image of the region with the melanoma-like insertion; E -the reflected A-scan signal from interfaces of the phantom; F -the reflected A-scan signal from interfaces of the insertion region.The circles placed on B-scans and A-scans denote the boundaries of the phantom which were detected by the amplitude threshold of the reflected ultrasonic waves

Fig. 4 .
Fig. 4. The relationship between the ultrasound attenu ation in the phan tom and ultrasound frequency.The legend indicates different scanned mediums: the phan toms with different Intralipid fat emulsion concentration (Table1) and the me lanoma-like insertion

Table 1 .
Concentrations of the ingredients used for manufacture of

Table 2 .
Comparison of acoustic properties of the phantoms and properties of the skin

Table 1
) and the me lanoma-like insertion