Ultrasonic assessment of post-radiotherapy fibrosis

Abstract

Tissue fibrosis is a common late effect of radiotherapy. Clinical assessment of radiation-induced fibrosis is still limited to hand palpation and other subjective grading scales to date, thus making it necessary to develop quantitative and objective methods for this purpose. In the current study, three different ultrasonic approaches were investigated for their potential uses in the assessment of tissue fibrosis. Firstly, a quasi-linear viscoelastic (QLV) model was adopted to extract the nonlinear mechanical properties of the fibrotic neck tissues from the force-deformation responses obtained with indentation tests using a curve-fitting process. The viscoelastic properties of the patients (n = 105) were compared with those of the normal subjects (n = 8) and also correlated to other measurements of fibrosis. The second part of this study was focused on the assessment of ultrasonic properties in skin fibrosis. Ultrasonic parameters of the dermal tissue in vivo were obtained from the echographic radiofrequency (RF) signals. These parameters included the attenuation slope b, the integrated attenuation (IA) and the integrated backscatter (IBS) in the frequency range of 10 to 25 MHz as well as the statistical characteristics of the signal envelope and the skin thickness. Testing sites were selected on the skin of the forearm and the neck. The intra-rater and inter-rater reliability of the measurements of the ultrasonic properties was assessed in 10 normal subjects using tests conducted by two investigators. 20 normal subjects (27+-3 yrs of age) and 38 patients (54+-11 yrs of age) with a history of radiotherapy in the neck region were recruited for testing. The results were compared between the normal subjects and the patients, and among different patient subgroups which had different palpation scores of fibrosis. Lastly, a high frequency (20-MHz) ultrasonic imaging system was modified to specifically study the skin elasticity. Preliminary tests on phantoms, porcine skins in vitro and the human skins in vivo were conducted to demonstrate the feasibility of this elasticity measurement system. Results of the QLV analysis showed that the initial modulus and the nonlinear factor were significantly larger in the patients than in the normal subjects (P < 0.001). Soft tissues with a higher degree of fibrosis were associated with higher elastic parameters. For the viscous properties, significant difference was demonstrated between the patients and the normal subjects (P < 0.007), but not among the patient subgroups (P > 0.05). It was demonstrated that the measurement of the ultrasonic properties in skin was reliable with an intraclass correlation coefficient (ICC) generally larger than 0.80. In the normal and fibrotic neck skins, b was 0.328+-0.029 and 0.303+-0.034 dB/mm/MHz, respectively (P = 0.005), and IA was 1.93+-1.81 and 2.85+-1.93 dB/mm, respectively (P = 0.04). It was found that b and IA were not age-dependent (P > 0.05), while IBS was age-dependent in the palmar forearm skin (P = 0.01). Its relative form of representation, IBSd, defined as the individual difference between the neck and the palmar forearm was -1.98+-2.13 and -4.23+-2.59 dB in normal subjects and patients, respectively. In terms of IBSd, a significant decrease was found in the fibrotic neck skin (P < 0.001). Age-matched comparison among four patient subgroups showed no significant difference for all the ultrasonic parameters (P > 0.05). No significant correlation between the ultrasonic properties and the effective Young's modulus (YM) of the neck soft tissue (P > 0.05) was demonstrated. Furthermore, it was found that the three ultrasonic parameters were correlated with each other (P < 0.01). The statistical parameters of the envelope signal including the mean/standard deviation ratio (MSR), skewness and kurtosis were demonstrated to be age-dependent (P < 0.007). A weakly significant difference was found between the irradiation-affected neck skin and the non-affected forearm skin in the patients (P close to 0.05). When the patients were divided into a total of 2 subgroups (with equivocal and severe fibrosis), the difference was demonstrated to be significant (P < 0.004). The skin thickness of the fibrotic neck skin was significantly larger than that of the normal neck skin (P < 0.01). However, no significant correlation between the skin thickness and the palpation score or the effective YM of the neck tissue was demonstrated. Preliminary results by using the modified elasticity measurement system showed that the difference of the stiffness of the two layers in phantoms or the layers in porcine skins in vitro could be discriminated. The deformation of the human skin in vivo could also be successfully obtained. It was concluded that the elastic properties extracted from the QLV model were good representations of severity of fibrosis. For the neck skin, b and IBS were decreased, while IA was increased by the fibrotic process. Ultrasonic and statistical parameters, and skin thickness, were potentially helpful for assessing the skin fibrosis. Our ultrasonic elasticity measurement system has the potential to measure the elasticity of the fibrotic skin with further improvement. However, future study is necessary to confirm some of the current findings.