Quantitative Radiomics Approach to Assess Acute Radiation Dermatitis in Breast Cancer Patients

doi:10.21203/rs.3.rs-1144295/v1

Download PDF

Research Article

Quantitative Radiomics Approach to Assess Acute Radiation Dermatitis in Breast Cancer Patients

https://doi.org/10.21203/rs.3.rs-1144295/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

We applied a radiomics approach to skin surface images to objectively assess radiodermatitis in patients undergoing radiotherapy for breast cancer. A prospective cohort study of 20 patients was conducted. Skin surface images in normal, polarised, and UV modes were acquired using a skin analysis device before starting radiotherapy (‘before RT’), 7 days after the first treatment ('RT_D7'), on ‘RT_D14’, and 10 days after radiotherapy ended (‘after RT_D10’). Eighteen types of radiomic feature ratios were calculated. We measured skin doses in ipsilateral breasts using OSLDs on the first day of radiotherapy. Clinical evaluation of acute radiodermatitis was performed using the RTOG scoring criteria on ‘RT_D14’ and ‘after RT_D10’. Several statistical analysis methods were used to test the performance of radiomic features as indicators of radiodermatitis evaluation. As the skin was damaged by radiation, the energy for normal mode and sum variance for polarised and UV modes decreased significantly for ipsilateral breasts. The radiomic feature ratios at ‘RT_D7’ had strong correlations to skin doses and those at ‘RT_D14’ and ‘after RT_D10’ with statistical significance. The energy for normal mode and sum variance for polarised and UV modes demonstrated the potential to evaluate and predict acute radiation, which assists in its appropriate management.

radiotherapy

breast cancer

OSLDs

UV modes

RTOG.

Acute radiation dermatitis is a common side effect in patients undergoing radiotherapy for breast cancer. The physical manifestations of radiodermatitis can range from faint erythema and desquamation to skin necrosis and ulceration, which impacts a patient's physical condition and quality of life^1–5. Risk factors for radiodermatitis include both treatment- and patient-related factors; however, they primarily depend on skin doses, which are classified as the treatment-related factor^6–8. Depending on its severity, acute radiation dermatitis may cause an interruption in or cessation of radiotherapy, which may lead to poor clinical outcomes^9–14.

Acute radiation reactions of the skin occur due to damage to the basal cell layer of the skin, which leads to an imbalance between the normal production of cells in this layer and the destruction of cells at the skin surface¹⁵. It appears 10–14 days from the first fraction of radiotherapy, corresponding with the time it takes for the damaged basal cells to migrate to the skin surface^{16, 17}. Because the skin is further damaged by radiation during radiotherapy, its reactions increase in severity and peak approximately 7–10 days after treatment has completed². Subsequently, the acute radiation dermatitis will start to subside and gradually heal^{18, 19}.

Traditionally, the severity of acute radiation dermatitis is still subjectively evaluated and graded by the clinician using scoring criteria, such as those provided by the Radiation Therapy Oncology Group (RTOG)^{17, 20, 21}, the Common Terminology Criteria for Adverse Events^{22, 23}, and the World Health Organization criteria^24–26. The inherent subjectivity of these criteria can cause high inter-observer variability, and they do not provide quantitative information for some clinical biomarkers. Another limitation is the diversity in the types of radiation damage that present the same physical manifestation on the skin (e.g. faint erythema); all of these types may be simply categorised under the same severity grade. Consequently, several ‘objective’ methods to quantify changes in the acute radiation reaction of the skin have been introduced by utilising a spectrophotometer, corneometer, and laser Doppler flowmetry^{19, 27–31}. Schmeel et al. provided a reliable and precise measure of skin severity, based on a three-dimensional colour system, which consists of L* (lightness), a* (colour ranging from red to green), and b* (colour ranging from blue to yellow)³¹. Because radiation makes the skin darker and redder, the L* values decreased and the a* values increased, whereas changes in b* values were negligible. Kawamura et al. generated an objective scoring system to predict acute radiation dermatitis in patients treated with radiotherapy for head and neck cancer using dose-volumetric parameters of the skin⁷.

Radiomics is a new concept that can be used to provide quantitative measures of tumours or normal tissues by analysing medical images from various modalities; it has attracted considerable interest in the field of radiology. Radiomic features from several non-invasive medical images can be utilised by integrating clinical biomarkers, genetic and biochemical indices, and treatment outcomes into a sophisticated model to improve diagnostic assessment and prognostic determinants of patients^{32, 33}. Recently, the clinical application of radiomics has expanded from internal image modalities, which image the internal body parts, such as computed tomography (CT), magnetic resonance imaging (MRI), and ultrasonography (US), to external image modalities, which scan the skin surface, such as a thermal imaging device and digital cameras. Several previous studies have shown that the radiomic features of skin surface images acquired by these modalities have the potential to monitor or predict the skin severity^34–36. Although the feasibility of thermal and digital skin imaging for the whole chest area to evaluate the occurrence of acute radiation dermatitis has been demonstrated in previous studies, there were limitations in analysing the large area image of the whole breast, resulting in a low resolution bias. Moreover, previous studies have investigated the performance of these objective methods by correlating the ‘subjective’ scoring criteria.

In our study, we evaluated the severity of acute radiation dermatitis for patients with breast cancer by applying a radiomics approach to topical images measured by a skin analysis device. The skin analysis device acquires images of the skin surface with the highest resolution, which can provide qualitatively and quantitatively sufficient information for radiomics analysis. We analysed the correlation between the radiomic features calculated from the images and measured skin doses. We propose an alternative objective method to predict the skin severity at early treatment time intervals with longitudinal assessment.

Patients

This is a prospective cohort study conducted in a single centre between October 2018 and September 2020. Female patients with newly diagnosed ductal carcinoma in situ were enrolled. All patients underwent breast-conserving surgery followed by whole-breast radiotherapy. The exclusion criteria were as follows: age > 70 years; prior history of radiotherapy for breast; recurrent breast cancer; bilateral breast cancer; patients with distant metastasis from breast cancer; previous breast reconstruction or implants; and patients receiving previous or simultaneous chemotherapy. This study was approved by the Institutional Review Board of Seoul National University hospital and was conducted according to the Declaration of Helsinki. Informed consent was obtained from all patients prior to their participation in this study.

Radiotherapy

All patients underwent CT scans using the Brilliance CT Big Bore^TM (Philips, Amsterdam, Netherlands) with a slice thickness of 3 mm. The patients were immobilised on a breast board (CIVCO Radiotherapy, Coraville, IA, USA) in supine position with both arms raised above their heads. The target volume was defined as the whole ipsilateral breast. The patients received 40.5 Gy in 15 fractions using 6 MV, or a combination of 6 MV and 10 MV, photons. Depending on the side of the breast cancer treatment, tangential photon beams as intensity modulated radiotherapy (IMRT) or field-in-field techniques were generated for left- or right-sided breast cancers, respectively. At least 95% of the target volume should be covered by 100% of the prescription dose. The maximum dose to the target volume did not typically exceed 115% of the prescription dose. For all patients, boost irradiation was not administered.

Patient evaluation

Clinical assessment of acute radiation dermatitis using scoring criteria

Clinical evaluation of acute radiation dermatitis using RTOG scoring criteria [20] was performed approximately 14 days after the first day of radiotherapy (‘RT D14’). The subsequent assessment was performed approximately 10 days after the end of the radiotherapy (‘after RT D10’). All patients were assessed independently by an experienced radiation oncologist who was blind to the results of both the radiomics analysis and skin doses. When the evaluation session overlapped with holidays or weekend, we evaluated weekday sessions within a 3-day window.

Skin dose measurement

The nanoDot optically stimulated luminescent dosimeter (OSLD) system (Landauer Inc., Glenwood, IL, USA) was used to measure skin doses to ipsilateral breasts of patients on the first day of radiotherapy. To reduce the dose measurement uncertainty, we used OSLDs prepared with pre-irradiation with dose values greater than 5 kGy [37]. Four measurement points were placed on the upper, lower, inner, and outer sides of the ipsilateral breast. Each point was located at a distance of 3 cm from the nipple of the ipsilateral breast. One set of three OSLDs was utilised for each measurement point and the dose values of those OSLDs were then averaged. Surgical scars were avoided when selecting the points.

Acquisition of skin surface image with skin analysis device and radiomics analysis

Skin surface imaging was performed at room temperature (25 ± 1°C) and under ambient room fluorescent lights using a portable skin analysis device (API-100; Aram Huvis, Gyeonggi-do, Korea). This device has a measurement area of 1 × 1 cm² with a resolution of 1624 × 1212 pixels and it provides three different modes of skin surface imaging: normal, polarised, and ultraviolet (UV) modes. For the acquisition of skin surface images, all patients were positioned, while lying on their back in a couch, with their arms raised above their heads. All measurements were performed by the same investigator. No topical products, including cosmetics, emollient, and cleansers, were permitted for at least 8 h before the skin measurement [29]. Skin surface images were acquired for the ipsilateral breasts in three modes at the corresponding points measured by OSLDs at four different times. The measurement times were before starting radiotherapy (‘before RT’), 'RT D7,' ‘RT D14,’ and ‘after RT D10’. For comparison, the corresponding points of the contralateral breast were measured in the same manner as the controls. The total measurement time was less 10 min. A timeline of our experimental design is depicted in Fig. 1.

The acquired image was in a red-green-blue (RGB) colour format with 24 bits per pixel, and it was separated into three channels for analysis. For the red channel, each image was cropped into a circular region-of-interest (ROI) in the centre of the image with a radius of 300 pixels to avoid the analysis of unnecessary portions of the image. To extract radiomic features from the image, the cropped images were analysed using second-order statistics based on a grey level co-occurrence matrix (GLCM) [38]. To generate the GLCM, the number of quantised grey levels (N_g), the value of the displacement vector (d), and the values of the offset angles (θ) were set to N_g = 64, d = 1, and θ = 0°, 45°, 90°, and 135°, respectively. A total of 18 radiomic features were calculated based on the acquired GLCMs using open-source code in MATLAB software (Mathworks, Natick, MA, USA) [39]. For each image, a total of four radiomic features were calculated according to the four offset angles, and those features were then averaged for each image. For the green and blue channels, we calculated the radiomic features in the same manner described above. The three radiomic features from each channel were averaged for each original image in the three channels. Thus, for each patient, a total of 1,728 radiomic features (18 radiomic features × four measurement points × both breasts × four time points × three imaging modes) were calculated in this study.

For each breast side, the ratio of the values of the radiomic features for all time points (‘before RT,’ 'RT D7,' ‘RT D14,’ and ‘after RT D10’) to those for the first time point (‘before RT’) were calculated to consider the diversity of colour, tone, and texture of the skin among individuals.

Statistical analysis

A two-way repeated measures analysis of variance (ANOVA) was used to test the significant changes in the ratio of the radiomic features within each group (ipsilateral or contralateral breast) across time, and the significant differences in those for the interaction of time and group. If the Mauchly's test of sphericity assumption was violated, the Greenhous–Geisser correction was applied. P-values less than 0.05 were considered as statistically significant. Depending on the results of the Shapiro–Wilk test for the normality of the two corresponding datasets, a paired-t test or Wilcoxon signed rank test was used for pairwise comparisons of the ratio of radiomic features between ipsilateral and contralateral breasts for each time point. Bonferroni's correction, which considers the adjusted significance level of 0.0167, was applied for post hoc comparisons. Correlations between skin doses and the ratio of radiomic features for ipsilateral breasts for all time points were determined using Pearson's or Spearman’s correlation test for parametric or non-parametric data, respectively. For ipsilateral breasts, Pearson's or Spearman's correlation coefficients were calculated to assess the correlations of the ratio of radiomic features for 'RT D7’ to those for 'RT D14’ and ‘after RT D10,’ depending on the normality of the data. For these statistical tests, p-values less than 0.05 were considered as statistically significant. All statistical analyses were performed using PASW statistic 18.0 software (SPSS, Chicago, IL, USA).

Patients and RTOG scoring

A total of 20 patients were enrolled in this prospective study. Table 1 shows the clinical characteristics of the patients, including age, tumour location, plan techniques and photon energies, and different treatments for breast cancer (previous chemotherapy and hormone therapy). For all patients, the pathologic stage of breast cancer was pTisN0M0.

Table 1

Patient characteristics (n = 20)
Characteristics	Case number (%)
Median age (range, years)	49.5 years (40 – 68)
Tumour location
Left	8 (40%)
Right	12 (60%)
Plan technique
FIF	12 (60%)
IMRT	8 (40%)
Energy
6 MV	17 (85%)
10 MV	1 (5%)
6 + 10 MV	2 (10%)
Chemotherapy
Yes	0 (0%)
No	20 (100%)
Hormone therapy
Yes	15 (75%)
No	5 (25%)

Abbreviations: FIF = field-in-field; IMRT = intensity modulated radiotherapy.

Based on the RTOG scoring criteria, all patients were classified as grade 1 at the ‘RT D14’ and ‘after RT D10' times.

Skin dose measurement

The values measured using the OSLD dosimeters are summarised in Supplementary Table A1. The mean values of the skin doses for the upper, lower, inner, and outer sides of the ipsilateral breast were 253.1 ± 14.7, 247.8 ± 18.1, 212.4 ± 19.6, and 226.8 ± 23.5 cGy, respectively. Most of the mean dose values for the inner ipsilateral breast (14 cases; 70%) were the lowest. The mean dose values for the upper side of the ipsilateral breast were comparable to those for the lower side.

Longitudinal assessment of changes in skin colour and texture by radiotherapy

We examined the changes in skin colour and texture according to the radiomic feature ratio of the skin surface image measured using a skin analysis device. Figure 2 shows the skin colour and texture changes and the corresponding GLCMs over time in the three imaging modes. There were noticeable changes in the skin surface images, owing to radiation damage. The calculated radiomic feature ratios for ipsilateral and contralateral breasts in three imaging modes (normal, polarised, and UV lights) are presented in Table 2, 3, and 4, respectively. P-values obtained using the two-way repeated measures ANOVA and paired-t tests (or Wilcoxon signed rank tests) are shown in Supplementary Table A2 and A3, respectively.

Table 2

The mean and standard deviation ratio of radiomic features for ipsilateral and contralateral breasts in the normal imaging mode
	Ipsilateral breast				Contralateral breast
	Before RT	RT D7	RT D14	After RT D10	Before RT	RT D7	RT D14	After RT D10
Autocorrelation^*	1.000 ± 0.000	0.966 ± 0.054	0.961 ± 0.036	1.008 ± 0.060	1.000 ± 0.000	0.990 ± 0.045	0.990 ± 0.046	1.002 ± 0.048
Contrast^*	1.000 ± 0.000	1.562 ± 0.624	1.979 ± 0.965	1.207 ± 0.583	1.000 ± 0.000	1.042 ± 0.358	1.078 ± 0.522	1.065 ± 0.498
Correlation^*	1.000 ± 0.000	0.992 ± 0.012	0.987 ± 0.015	1.002 ± 0.013	1.000 ± 0.000	1.002 ± 0.011	1.001 ± 0.015	1.002 ± 0.016
Cluster prominence^*	1.000 ± 0.000	1.789 ± 0.670	2.254 ± 1.173	1.334 ± 0.667	1.000 ± 0.000	1.230 ± 0.497	1.169 ± 0.497	1.078 ± 0.560
Cluster shade^*	1.000 ± 0.000	2.108 ± 1.083	2.508 ± 1.443	1.111 ± 1.235	1.000 ± 0.000	1.658 ± 1.622	1.540 ± 1.375	1.487 ± 1.902
Dissimilarity^*	1.000 ± 0.000	1.221 ± 0.260	1.399 ± 0.356	1.121 ± 0.301	1.000 ± 0.000	0.988 ± 0.185	0.984 ± 0.241	0.995 ± 0.261
Energy^{*, †}	1.000 ± 0.000	0.848 ± 0.109	0.775 ± 0.140	0.672 ± 0.155	1.000 ± 0.000	1.072 ± 0.164	1.107 ± 0.191	1.072 ± 0.235
Entropy^{*, †}	1.000 ± 0.000	1.045 ± 0.052	1.084 ± 0.066	1.039 ± 0.066	1.000 ± 0.000	0.994 ± 0.048	0.988 ± 0.057	0.989 ± 0.065
Homogeneity^{*, †}	1.000 ± 0.000	0.966 ± 0.049	0.933 ± 0.060	0.973 ± 0.065	1.000 ± 0.000	1.011 ± 0.042	1.017 ± 0.055	1.012 ± 0.068
Maximum probability^*	1.000 ± 0.000	0.971 ± 0.183	0.886 ± 0.166	0.844 ± 0.161	1.000 ± 0.000	1.073 ± 0.138	1.101 ± 0.146	1.045 ± 0.173
Variance^*	1.000 ± 0.000	0.967 ± 0.054	0.961 ± 0.036	1.008 ± 0.060	1.000 ± 0.000	0.990 ± 0.044	0.990 ± 0.046	1.002 ± 0.048
Sum average^*	1.000 ± 0.000	0.982 ± 0.027	0.979 ± 0.019	1.003 ± 0.030	1.000 ± 0.000	0.994 ± 0.023	0.995 ± 0.023	1.001 ± 0.024
Sum variance^*	1.000 ± 0.000	0.964 ± 0.056	0.957 ± 0.037	1.006 ± 0.063	1.000 ± 0.000	0.990 ± 0.047	0.990 ± 0.049	1.003 ± 0.051
Sum entropy^{*, †}	1.000 ± 0.000	1.020 ± 0.026	1.037 ± 0.029	1.028 ± 0.027	1.000 ± 0.000	0.999 ± 0.026	0.994 ± 0.025	0.994 ± 0.029
Difference variance^*	1.000 ± 0.000	1.562 ± 0.624	1.979 ± 0.965	1.207 ± 0.583	1.000 ± 0.000	1.042 ± 0.358	1.078 ± 0.522	1.065 ± 0.498
Difference entropy^*	1.000 ± 0.000	1.121 ± 0.143	1.210 ± 0.170	1.046 ± 0.171	1.000 ± 0.000	0.993 ± 0.120	0.988 ± 0.147	0.982 ± 0.154
IMC 1^*	1.000 ± 0.000	0.945 ± 0.091	0.893 ± 0.100	0.984 ± 0.116	1.000 ± 0.000	1.024 ± 0.086	1.030 ± 0.113	1.028 ± 0.134
IMC 2^*	1.000 ± 0.000	0.996 ± 0.009	0.992 ± 0.011	1.000 ± 0.010	1.000 ± 0.000	1.002 ± 0.007	1.002 ± 0.010	1.001 ± 0.012

Abbreviations: Before RT = before starting the radiotherapy; RT Dn = approximately n days after the first day of the radiotherapy; After RT Dn = approximately n days after the end of the radiotherapy; IMC = inverse measure of correlation.

^*indicates statistical significances (p < 0.05) of both changes in the radiomic feature ratio within each group across time and differences in those for the interaction of the time and group with a two-way repeated measures analysis of variance (ANOVA).

^†indicates statistical significances (p-value < 0.05/3 = 0.0167) of differences in the radiomic feature ratio between ipsilateral and contralateral breasts for all time points with a paired-t test or Wilcoxon signed rank test, depending on the normality of data.

Table 3

The mean and standard deviation ratio of radiomic features for ipsilateral and contralateral breasts in the polarised imaging mode
	Ipsilateral breast				Contralateral breast
	Before RT	RT D7	RT D14	After RT D10	Before RT	RT D7	RT D14	After RT D10
Autocorrelation^{*, †}	1.000 ± 0.000	0.936 ± 0.062	0.879 ± 0.098	0.904 ± 0.115	1.000 ± 0.000	0.973 ± 0.043	0.961 ± 0.060	0.954 ± 0.072
Contrast^*	1.000 ± 0.000	1.458 ± 0.602	1.931 ± 0.737	1.858 ± 1.358	1.000 ± 0.000	1.139 ± 0.330	1.165 ± 0.348	1.309 ± 0.469
Correlation	1.000 ± 0.000	0.997 ± 0.009	0.998 ± 0.009	1.003 ± 0.013	1.000 ± 0.000	1.000 ± 0.008	1.003 ± 0.008	1.003 ± 0.008
Cluster prominence	1.000 ± 0.000	2.779 ± 3.898	4.773 ± 7.285	18.476 ± 56.921	1.000 ± 0.000	1.962 ± 2.764	2.076 ± 2.317	3.451 ± 5.296
Cluster shade	1.000 ± 0.000	1.100 ± 3.978	4.062 ± 8.047	18.125 ± 56.285	1.000 ± 0.000	1.414 ± 3.313	1.168 ± 2.299	2.042 ± 5.708
Dissimilarity^{*, †}	1.000 ± 0.000	1.274 ± 0.325	1.590 ± 0.333	1.638 ± 0.402	1.000 ± 0.000	1.079 ± 0.246	1.151 ± 0.249	1.251 ± 0.306
Energy^*	1.000 ± 0.000	0.902 ± 0.205	0.713 ± 0.169	0.621 ± 0.165	1.000 ± 0.000	0.984 ± 0.252	0.883 ± 0.284	0.831 ± 0.254
Entropy^{*, †}	1.000 ± 0.000	1.074 ± 0.133	1.175 ± 0.130	1.228 ± 0.137	1.000 ± 0.000	1.031 ± 0.134	1.074 ± 0.142	1.103 ± 0.140
Homogeneity^{*, †}	1.000 ± 0.000	0.983 ± 0.017	0.962 ± 0.015	0.952 ± 0.030	1.000 ± 0.000	0.996 ± 0.014	0.989 ± 0.017	0.982 ± 0.024
Maximum probability	1.000 ± 0.000	0.931 ± 0.205	0.778 ± 0.170	0.703 ± 0.166	1.000 ± 0.000	1.011 ± 0.254	0.915 ± 0.266	0.869 ± 0.246
Variance^{*, †}	1.000 ± 0.000	0.936 ± 0.062	0.879 ± 0.098	0.904 ± 0.115	1.000 ± 0.000	0.973 ± 0.042	0.961 ± 0.060	0.954 ± 0.072
Sum average^{*, †}	1.000 ± 0.000	0.967 ± 0.033	0.935 ± 0.055	0.946 ± 0.067	1.000 ± 0.000	0.986 ± 0.022	0.980 ± 0.032	0.975 ± 0.039
Sum variance^{*, †}	1.000 ± 0.000	0.920 ± 0.059	0.859 ± 0.093	0.795 ± 0.105	1.000 ± 0.000	0.971 ± 0.047	0.958 ± 0.065	0.949 ± 0.077
Sum entropy^*	1.000 ± 0.000	1.061 ± 0.126	1.149 ± 0.122	1.203 ± 0.131	1.000 ± 0.000	1.028 ± 0.130	1.070 ± 0.137	1.094 ± 0.131
Difference variance^*	1.000 ± 0.000	1.458 ± 0.602	1.931 ± 0.737	1.858 ± 1.358	1.000 ± 0.000	1.139 ± 0.330	1.165 ± 0.348	1.309 ± 0.469
Difference entropy^{*, †}	1.000 ± 0.000	1.162 ± 0.179	1.341 ± 0.180	1.347 ± 0.196	1.000 ± 0.000	1.050 ± 0.148	1.092 ± 0.144	1.146 ± 0.169
IMC 1^{*, †}	1.000 ± 0.000	0.962 ± 0.030	0.930 ± 0.025	0.937 ± 0.056	1.000 ± 0.000	0.994 ± 0.018	0.989 ± 0.021	0.978 ± 0.032
IMC 2	1.000 ± 0.000	1.000 ± 0.008	1.003 ± 0.007	1.006 ± 0.010	1.000 ± 0.000	1.001 ± 0.009	1.003 ± 0.010	1.004 ± 0.009

^*indicates statistical significances (p < 0.05) of both changes in the radiomic feature ratios within each group across time and differences in those for the interaction of the time and group with a two-way repeated measures analysis of variance (ANOVA).

^†indicates statistical significances (p-value < 0.05/3 = 0.0167) of differences in the radiomic feature ratios between ipsilateral and contralateral breasts for all time points with a paired-t test or Wilcoxon signed rank test, depending on the normality of data.

Table 4

The mean and standard deviation ratio of radiomic features for ipsilateral and contralateral breasts in the UV imaging mode
	Ipsilateral breast				Contralateral breast
	Before RT	RT D7	RT D14	After RT D10	Before RT	RT D7	RT D14	After RT D10
Autocorrelation^{*, †}	1.000 ± 0.000	0.874 ± 0.113	0.716 ± 0.190	0.579 ± 0.255	1.000 ± 0.000	0.995 ± 0.049	0.951 ± 0.084	0.908 ± 0.142
Contrast^*	1.000 ± 0.000	2.026 ± 1.261	3.380 ± 2.425	4.468 ± 4.430	1.000 ± 0.000	1.116 ± 0.570	1.565 ± 0.784	1.792 ± 0.920
Correlation	1.000 ± 0.000	1.001 ± 0.005	1.003 ± 0.008	1.009 ± 0.008	1.000 ± 0.000	0.998 ± 0.007	1.000 ± 0.009	1.004 ± 0.008
Cluster prominence	1.000 ± 0.000	9.390 ± 27.491	10.022 ± 16.845	71.194 ± 187.39	1.000 ± 0.000	1.767 ± 2.244	2.028 ± 1.737	5.985 ± 6.946
Cluster shade	1.000 ± 0.000	3.540 ± 11.673	0.797 ± 16.016	13.359 ± 59.846	1.000 ± 0.000	1.392 ± 1.325	1.348 ± 1.070	2.478 ± 4.292
Dissimilarity^*	1.000 ± 0.000	1.864 ± 0.956	2.800 ± 1.767	3.492 ± 3.035	1.000 ± 0.000	1.112 ± 0.545	1.499 ± 0.730	1.718 ± 0.839
Energy	1.000 ± 0.000	0.497 ± 0.279	0.330 ± 0.242	0.303 ± 0.319	1.000 ± 0.000	0.935 ± 0.362	0.747 ± 0.445	0.662 ± 0.461
Entropy^*	1.000 ± 0.000	1.451 ± 0.457	1.746 ± 0.710	1.973 ± 1.099	1.000 ± 0.000	1.039 ± 0.321	1.233 ± 0.381	1.361 ± 0.417
Homogeneity^*	1.000 ± 0.000	0.934 ± 0.055	0.889 ± 0.091	0.862 ± 0.134	1.000 ± 0.000	0.993 ± 0.033	0.966 ± 0.053	0.953 ± 0.055
Maximum probability	1.000 ± 0.000	0.584 ± 0.329	0.565 ± 0.864	0.670 ± 1.600	1.000 ± 0.000	0.931 ± 0.305	0.773 ± 0.405	0.721 ± 0.424
Variance^{*, †}	1.000 ± 0.000	0.874 ± 0.112	0.716 ± 0.190	0.580 ± 0.255	1.000 ± 0.000	0.995 ± 0.049	0.951 ± 0.084	0.908 ± 0.142
Sum average^{*, †}	1.000 ± 0.000	0.930 ± 0.065	0.831 ± 0.132	0.727 ± 0.199	1.000 ± 0.000	0.997 ± 0.026	0.974 ± 0.045	0.947 ± 0.088
Sum variance^{*, †}	1.000 ± 0.000	0.856 ± 0.109	0.697 ± 0.191	0.540 ± 0.234	1.000 ± 0.000	0.994 ± 0.056	0.945 ± 0.092	0.899 ± 0.149
Sum entropy^*	1.000 ± 0.000	1.412 ± 0.429	1.654 ± 0.640	1.836 ± 0.960	1.000 ± 0.000	1.035 ± 0.308	1.214 ± 0.362	1.335 ± 0.398
Difference variance^*	1.000 ± 0.000	2.026 ± 1.261	3.380 ± 2.425	4.468 ± 4.430	1.000 ± 0.000	1.116 ± 0.570	1.565 ± 0.784	1.792 ± 0.920
Difference entropy^*	1.000 ± 0.000	1.507 ± 0.511	1.915 ± 0.830	2.148 ± 1.258	1.000 ± 0.000	1.056 ± 0.333	1.281 ± 0.423	1.405 ± 0.443
IMC 1^*	1.000 ± 0.000	0.921 ± 0.067	0.878 ± 0.100	0.882 ± 0.139	1.000 ± 0.000	0.986 ± 0.040	0.954 ± 0.061	0.951 ± 0.066
IMC 2	1.000 ± 0.000	1.067 ± 0.090	1.089 ± 0.109	1.100 ± 0.131	1.000 ± 0.000	0.998 ± 0.061	1.031 ± 0.060	1.056 ± 0.074

In the normal imaging mode, all radiomic features exhibited statistically significant changes within and between groups (ipsilateral and contralateral breasts) over time. Among all features, the energy, entropy, homogeneity, and sum entropy exhibited differences in the radiomic feature ratios between ipsilateral and contralateral breasts for all time points with statistical significance. Every ratio of radiomic features for ipsilateral breasts tended to increase or decrease until 'RT D14.' However, they exhibited the opposite trend 'after RT D10,' except for the energy and maximum probability, which decreased significantly. While all radiomic feature ratios for ipsilateral breasts showed noticeable tendencies over time, those for contralateral breasts exhibited unpredictable tendencies with smaller variations over time.

In the polarised imaging mode, every ratio of radiomic features exhibited statistically significant changes within and between groups across time, except for the correlation, cluster prominence, cluster shade, maximum probability, and inverse measure of correlation 2 (IMC 2). Among them, most of the features also showed statistically significant differences in the ratio values between the groups for all time points, except for the contrast, energy, sum entropy, and difference variance. For the statistically significant radiomic feature ratios of ipsilateral breasts, the dissimilarity, entropy, sum entropy, and difference entropy increased, while the energy, homogeneity, and sum variance decreased during the time course. The corresponding radiomic feature ratio for contralateral breasts exhibited the same trends as those for ipsilateral breasts; however, its change width was smaller than that for ipsilateral breasts.

In the UV imaging mode, of the 12 radiomic features that showed significant differences for the interaction of time and group, the autocorrelation, variance, sum average, and sum variance showed statistically significant differences between groups for all time points. Every radiomic feature for the ipsilateral breasts exhibited increasing or decreasing tendencies over time, except the cluster shade, maximum probability, and IMC 1. Similar to the polarised imaging mode, the corresponding ratio of radiomic features for contralateral breasts showed the same trends as those for the ipsilateral breast; however, its change width was smaller than that for ipsilateral breasts. Overall, changes between time points in the UV imaging mode were larger than those in the normal and polarised imaging modes.

No significant relationship between the ratio of radiomic features for ipsilateral breasts and RTOG scoring was observed.

Correlations of radiomic feature ratio with skin doses

For this analysis, the ratios of the radiomic features for the ipsilateral breasts were used. The values of the correlation coefficient (r), the corresponding p-values of the skin doses, and the ratio of the radiomic features at 'RT D7,' 'RT D14,' and 'after RT D10’ were calculated. No significant correlations between the radiomic feature ratio at 'RT D14' and 'after RT D10’ and the skin doses were identified. The values of r and corresponding p-values of skin doses and radiomic feature ratios at 'RT D7’ are shown in Table 5. Only the r values with p-values less than 0.05 are listed in Table 5.

Table 5

Correlation coefficients (r) with corresponding p-values of Pearson's or Spearman's correlation test between skin dose values and radiomic feature ratios approximately 7 days after the first day of the radiotherapy (‘RT D7')
	Normal		Polarised		UV
	r	p	r	p	r	p
Autocorrelation	-	-	-0.540	0.014	-0.485	0.043
Contrast	-	-	0.465	0.039	-	-
Correlation	-	-	-	-	-	-
Cluster prominence	-	-	-	-	-	-
Cluster shade	-	-	-	-	-	-
Dissimilarity	-	-	-	-	-	-
Energy	-0.625	0.003	-	-	-	-
Entropy	0.515	0.020	-	-	-	-
Homogeneity	-0.527	0.017	-0.479	0.033	-	-
Maximum probability	-0.545	0.013	-	-	-	-
Variance	-	-	-0.540	0.014	-0.485	0.043
Sum average	-	-	-0.542	0.013	-0.510	0.038
Sum variance	-	-	-0.569	0.014	-0.590	0.042
Sum entropy	0.505	0.023	-	-	-	-
Difference variance	-	-	0.465	0.039	-	-
Difference entropy	-	-	-	-	-	-
IMC 1	-0.446	0.049	-0.491	0.028	-	-
IMC 2	-	-	-	-	-	-

Abbreviations: UV = ultraviolet; IMC = inverse measure of correlation

In the normal imaging mode, radiomic features with p-values less than 0.05 were the energy, entropy, homogeneity, maximum probability, sum entropy, and IMC 1. The energy had a significantly higher value of r compared with other radiomic features to skin doses. In the polarised imaging mode, eight radiomic features, which were the autocorrelation, contrast, homogeneity, variance, sum average, sum variance, difference variance, and IMC 1, showed statistically significant correlations (p < 0.05) to skin doses. The sum variance exhibited a significantly higher value of r compared with other radiomic features to skin doses. In the UV imaging mode, radiomic features with p-values less than 0.05 included the autocorrelation, variance, sum average, and sum variance. Similar to the polarised imaging mode, the sum variance exhibited a significantly higher value of r compared with other radiomic features to skin doses. In the three imaging modes, the r values of radiomic features with p-values less than 0.05 were higher than 0.44.

Correlations between radiomic feature ratios for all time points

The values of r and the corresponding p-values of the radiomic feature ratios of 'RT D7’ to 'RT D14’ and 'after RT D10’ are shown in Table 6. Only the r values with p-values less than 0.05 are listed in Table 6. For this analysis, the ratios of the radiomic features for ipsilateral breasts were used.

Table 6

Correlation coefficients (r) with corresponding p-values of Pearson's or Spearman's correlation test of radiomic feature ratios approximately 7 days after the first day of the radiotherapy (‘RT D7') to those on ‘RT D14’ and approximately 10 days after the end of the radiotherapy ('After RT D10')
	Normal				Polarised				UV
	RT D14		After RT D10		RT D14		After RT D10		RT D14		After RT D10
	r	p	r	p	r	p	r	p	r	p		r	p
Autocorrelation	0.709	<0.001	0.631	0.003	-	-	0.462	0.040	0.765	<0.001		0.723	<0.001
Contrast	-	-	0.629	0.003	-	-	-	-	0.728	<0.001		0.472	0.036
Correlation	-	-	-	-	0.537	0.015	0.830	<0.001	0.652	0.002		0.567	0.009
Cluster prominence	0.470	0.036	0.485	0.030	0.942	<0.001	0.794	<0.001	0.944	<0.001		0.943	<0.001
Cluster shade	0.479	0.033	0.518	0.019	0.574	0.008	0.457	0.043	0.618	0.004		0.933	<0.001
Dissimilarity	-	-	0.535	0.015	0.616	0.004	-	-	0.726	<0.001		0.459	0.042
Energy	0.615	0.020	0.693	0.003	0.775	<0.001	0.609	0.004	0.584	0.007		0.449	0.047
Entropy	-	-	-	-	0.839	<0.001	0.770	<0.001	0.871	<0.001		0.667	0.001
Homogeneity	-	-	-	-	0.561	0.010	-	-	0.743	<0.001		0.509	0.022
Maximum probability	0.625	0.003	0.513	0.021	0.592	0.006	0.561	0.010	0.636	0.003		0.509	0.022
Variance	0.710	<0.001	0.631	0.003	-	-	0.462	0.040	0.764	<0.001		0.723	<0.001
Sum average	0.706	0.001	0.621	0.003	0.789	0.010	0.762	0.040	0.743	<0.001		0.698	0.001
Sum variance	0.700	0.001	0.632	0.003	-	-	0.463	0.040	0.761	<0.001		0.716	<0.001
Sum entropy	0.591	0.006	-	-	0.869	<0.001	0.828	<0.001	0.889	<0.001		0.682	0.001
Difference variance	-	-	0.629	0.003	-	-	-	-	0.728	<0.001		0.472	0.036
Difference entropy	-	-	0.486	0.030	0.618	0.004	0.502	0.024	0.826	<0.001		0.594	0.006
IMC 1	-	-	-	-	-	-	0.582	0.007	0.712	<0.001		-	-
IMC 2	-	-	-	-	0.823	<0.001	0.803	<0.001	0.864	<0.001		0.780	<0.001

Abbreviations: UV = ultraviolet; RT Dn = approximately n days after the first day of the radiotherapy; After RT Dn = approximately n days after the end of the radiotherapy; IMC = inverse measure of correlation

In the normal imaging mode, eight radiomic features for 'RT D7,’ which included the autocorrelation, cluster prominence, cluster shade, energy, maximum probability, variance, sum average, and sum variance, exhibited statistically significant correlations with those at both 'RT D14’ and 'after RT D10.' Among these, the variance indicated the strongest correlations at both 'RT D14’ and 'after RT D10.' In the polarised imaging mode, all radiomic features with p-values less than 0.05 exhibited statistically significant correlations with both 'RT D14’ and 'after RT D10,' except for the autocorrelation, contrast, dissimilarity, homogeneity, variance, sum variance, difference variance, and IMC 1. The cluster prominence and sum entropy had significantly higher r values than other values. In the UV imaging mode, every value of r, except for the IMC 1, was statistically significant and exhibited p-values less 0.05. Among these, the cluster prominence showed statistically significant correlations to those at both 'RT D14’ and 'after RT D10.' Statistically significant correlations were more frequently observed in the UV imaging mode than other modes.

In this study, we applied a radiomics approach to skin surface images acquired by a skin analysis device to objectively assess the acute radiation dermatitis in patients undergoing radiotherapy for breast cancer. The 18 radiomic features were calculated from images in the normal, polarised, and UV modes before, during, and after radiotherapy. To test the performance of radiomic features as indicators of radiodermatitis evaluation, we assessed the statistically significant changes and differences within and between groups (ipsilateral and contralateral breasts) over time and correlations between the radiomic features and the RTOG score or skin dose. Furthermore, correlations between the radiomic features at ‘RT D7’ to those at 'RT D14’ and 'after RT D10’ for the ipsilateral breast were analysed to evaluate whether the degree of acute radiation dermatitis can be predicted in advance using the feature.

Figure 2 shows that the skin surface images and the patterns of the corresponding GLCMs in the three imaging modes were changed as the skin was damaged by radiation. Accordingly, most of the radiomic feature ratios for the ipsilateral breasts showed increasing or decreasing tendencies, exhibiting better performance or correlations in all statistical analyses. Of these, the performance of the energy in the normal mode and the sum variance in the polarised and UV modes were better than others in general.

The normal mode in the skin analysis device could analyse skin pores. As shown in Fig. 2(a), it was observed that the skin surface became redder and darker over time on the side of the breast cancer treatment. The results of this study correspond well with those obtained in earlier studies^{19, 27, 28, 40}. Previous studies reported that the skin reaction for irradiated breasts exhibited higher a* values (reddish) and lower L* values (darker); these values were measured by a spectrophotometer during radiotherapy. Momm et al. showed strong correlations to the radiation dose using spectrophotometry. By analysing the spatial distribution of the GLCM intensities in Fig. 2(a), it can be observed that as the skin was damaged by radiation, the peak of the high intensities of GLCMs tended to gradually disappear. These intensities were subsequently shifted toward the lowest elements of the GLCMs. It was demonstrated that intensity homogeneity of the skin surface images deteriorated, and the colour tone of the skin became darker as the radiodermatitis progressed. The energy for the normal imaging mode, which exhibited the best performance for assessing acute radiation dermatitis, had decreasing tendencies over time because it is a measure of homogeneity of an image. A higher value for this feature indicates that the intensity in an image varies less⁴¹.

The polarised mode in the skin analysis device could evaluate the melanin pigmentation of the basal layer of the epidermis. Figure 2(b) shows that the skin pigmentation in ipsilateral breasts increased and peaked 'after RT D10.' The results of this study are similar to those of Hu et al.²⁹. Their study reported that the skin pigmentation of irradiated breasts increased, owing to radiation damage, compared with that of unirradiated breasts. The intensities in the GLCM distributions generated from skin surface images in the polarised mode were gradually spread out, and these intensities then shifted toward the lowest elements of the GLCMs, demonstrating that there was inhomogeneity and darkness in the images. The sum variance typically exhibits high values when the intensities of the GLCM distributions are gathered in the lowest and highest elements³⁸. For this reason, the values of the sum variance decreased over time in this study.

The UV mode in the skin analysis device was typically utilised to evaluate skin sebum; however, it also showed melanin accumulations below the skin surface. In this study, the skin surface images in the UV mode continued to darken from the periphery inward, toward the centre of the images, over time. This demonstrated that the melanin accumulated, owing to radiation damage, as shown in Fig. 2(c). It was difficult to evaluate radiodermatitis using skin sebum. Hu et al. also showed that there was no significant change in skin sebum in both ipsilateral and contralateral breasts before and after radiotherapy²⁹. Unlike other imaging modes, the periphery of the images in the UV mode appeared optically darker than the centre, owing to non-uniformity in UV intensity. For this reason, we cropped the circular ROI of the original images obtained in UV mode. Other images in the normal and polarised mode were also cropped in the same manner to calculate the radiomic features. The intensities in the GLCM distributions generated from skin surface images in the UV mode exhibited a pattern similar to those in the polarised mode. Therefore, the sum variance also had decreasing tendencies, providing the best performance for both polarised and UV imaging modes.

Figure 3 presents the changes in the ratios of the representative radiomic features for ipsilateral and contralateral breasts over time in the three imaging modes. As mentioned above, these radiomic features showed the statistical significance of the two-way repeated measures ANOVA test. In the unirradiated breasts, there were unpredictable tendencies in the energy in the normal mode. There were decreasing tendencies in the sum variance in both the polarised and UV modes; however, these showed a much smaller decrease compared to the irradiated breasts. This may be explained by the fact that radiation that was used to treat breast cancer was scattered or penetrated to the contralateral breast, thereby causing damage. Hu et al. reported similar results, showing an increase in pigmentation for the contralateral breast after radiotherapy²⁹. According to the results of our study, radiation-induced skin damage for contralateral breasts ‘after RT D10’ was comparable to that for ipsilateral breasts on ‘RT D7’. Thus, the degree of radiation-induced skin damage to the unirradiated breast could be quantified and evaluated using radiomics analysis of the skin surface image proposed in this study.

Several studies have addressed the limitation of these subjective methods, which create several uncertainties. They demonstrated that these methods could not represent patient-reported breast symptoms, including pain, itching, tightness, and local heat⁴². The subjective assessments were affected by intra- and inter-evaluator variations, and were considered less suitable for close scrutiny^{15, 22, 31}. Although these various studies highlighted the limitations of the clinician-assessed scoring criteria, the performances of various skin biophysical parameters as indicators for assessing radiodermatitis were evaluated by analysing the correlations between the scoring criteria and the parameters. In this study, we used the RTOG scoring criteria, and all enrolled patients were evaluated as grade 1 on ‘RT D14’ and ‘after RT D10.’ For this reason, our study could not show the correlations between the RTOG scoring and the radiomic feature ratios for different time points. However, we presented strong correlations between the skin doses and radiomic feature ratios on ‘RT D7.’ Thus, the severity of acute radiation dermatitis can vary within the same grade, showing the limitations of clinician-assessed scoring criteria. Because there is a significant relationship between the severity of radiodermatitis and the absolute doses delivered to the patients’ skin^{6, 7}, we could evaluate the severity of acute radiation dermatitis objectively using representative radiomic features (on 'RT D7'), regardless of the clinical-assessed scoring criteria. Furthermore, the representative radiomic features exhibited strong correlations to those on 'RT D14' and 'after RT D10,’ which means that radiation-induced skin damage during and after radiotherapy can be predicted in advance. Using a radiomics approach to skin surface images, it is possible to subdivide the severity of acute radiation dermatitis even within the same grade and then prepare appropriate countermeasures.

In future work, this radiomic approach for the evaluation of radiodermatitis will be further applied to patients undergoing radiotherapy for head and neck cancer, where acute radiation dermatitis is predominantly observed. Because patients over grade 2+ were not included, we were unable to evaluate the comprehensive severity of radiodermatitis, which is a limitation of this study. Further research using a large number of samples and various severities of radiodermatitis, various treatment sites, and various races will be performed in the future. In addition to the hand-crafted radiomic features utilised in our study, quantitative radiomic features extracted using a convolutional neural network can be used in supervised machine learning to predict the radiodermatitis²⁵.

A radiomics approach to skin surface images acquired by a skin analysis device was applied to objectively assess acute radiation dermatitis in patients undergoing radiotherapy for breast cancer. In general, the energy for the normal mode and sum variance for polarised and UV modes exhibited better performance than others in evaluating the radiation-induced skin damage. Using the ratios of the energy and sum variance on 'RT D7,’ radiodermatitis severity during and after radiotherapy can be predicted in advance, which assists in its appropriate management.

Acknowledgements

This study was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (NRF-2017R1D1A1B03036093) and by a VHS Medical Center Research Grant, Republic of Korea (grant number: VHSMC 20056).

Author contributions statement

S.P., and J.H.K. designed experiments. J.M.P., J.K., C.H.C, and M.C. carried out experiments. S.P., and J.H.K. performed statistical analysis, prepared figures, and wrote the manuscript. J.H.J. provided critical feedback. All authors critically reviewed the manuscript and approved the final draft to be published.

Competing interests

The authors declare no competing interests.

Ryan, J.L. Ionizing radiation: the good, the bad, and the ugly. J Invest Dermatol. 132, 985-993 (2012).
Hymes, S.R., Strom, E.A. & Fife, C. Radiation dermatitis: clinical presentation, pathophysiology, and treatment 2006. J Am Acad Dermatol. 54, 28-46 (2006).
Schnur, J.B., Ouellette, S.C., Dilorenzo, T.A., Green, S. & Montgomery, G.H. A qualitative analysis of acute skin toxicity among breast cancer radiotherapy patients. Psycho-Oncol. 20, 260-268 (2011).
Huh, D.H. & Kim, S. History of radiation therapy technology. Prog Med Phys. 31, 124-133 (2020).
Sakyanun, P., Saksornchai, K., Nantavithya, C., Chakkabat, C. & Chotelersuk, K. The effect of deep inspiration breath-hold technique on left anterior descending coronary artery and heart dose in left breast irradiation. Radiat Oncol J. 38, 181-188 (2020).
Mendelsohn, F.A., Divino, C.M., Reis, E.D. & Kerstein, M.D. Wound care after radiation therapy. Adv Skin Wound Care. 15, 216-224 (2002).
Kawamura, M., et al. A scoring system predicting acute radiation dermatitis in patients with head and neck cancer treated with intensity-modulated radiotherapy. Radiat Oncol.14, 14 (2019).
Lee, C. et al. Organ dose conversion coefficients calculated for Korean pediatric and adult voxel phantoms exposed to external photon fields. J Radiat Prot Res. 45, 69-75 (2020).
Salvo, N. et al. Prophylaxis and management of acute radiation-induced skin reactions: a systematic review of the literature. Curr Oncol. 17, 94-112 (2010).
Bese, N.S., Sut, P.A., Sut, N. & Ober, A. The impact of treatment interruptions on locoregional control during postoperative breast irradiation. Journal of BUON : J BUON. 12, 353-359 (2007).
Yadav, B.S. et al. Late-term effects of hypofractionated chest wall and regional nodal radiotherapy with two-dimensional technique in patients with breast cancer. Radiat Oncol J. 38,109-118 (2020).
Park, S. & Kim, J. Regional nodal irradiation in pT1-2N1 breast cancer patients treated with breast-conserving surgery and whole breast irradiation. Radiat Oncol J. 38, 44-51 (2020).
Sayan, M. et al. Daily localization of partial breast irradiation patients with three-dimensional ultrasound imaging. Radiat Oncol J. 37, 259-264 (2019).
Kim, J. et al. Radiotherapy for initial clinically positive internal mammary nodes in breast cancer. Radiat Oncol J. 37, 91-100 (2019).
Sitton, E. Early and late radiation-induced skin alterations. Part II: Nursing care of irradiated skin. Oncol Nurs Forum. 19, 907-912 (1992).
Cleary, J.F., Anderson, B.M., Eickhoff, J.C., Khuntia, D. & Fahl, W.E. Significant suppression of radiation dermatitis in breast cancer patients using a topically applied adrenergic vasoconstrictor. Radiat Oncol. 12, 201 (2017).
Porock, D. & Kristjanson, L. Skin reactions during radiotherapy for breast cancer: the use and impact of topical agents and dressings. Eur J Cancer Care. 8, 143-153 (1999).
Aistars, J. The validity of skin care protocols followed by women with breast cancer receiving external radiation. Clin J Oncol Nurs. 10, 487-492 (2006).
Yamazaki, H. et al. Longitudinal practical measurement of skin color and moisture during and after breast-conserving therapy: influence of neoadjuvant systemic therapy. Jpn J Radiol. 27, 309-315 (2009).
Cox, J.D., Stetz, J. & Pajak, T.F. Toxicity criteria of the Radiation Therapy Oncology Group (RTOG) and the European Organization for Research and Treatment of Cancer (EORTC). Int J Radiat Oncol Biol Phys. 31, 1341-1346 (1995).
Dische, S. The Uniform Reporting of Treatment-Related Morbidity. Semin Radiat Oncol. 4, 112-118 (1994).
Trotti, A. et al. CTCAE v3.0: development of a comprehensive grading system for the adverse effects of cancer treatment. Semin Radiat Oncol. 13, 176-181 (2003).
Neben-Wittich, M.A. et al. Comparison of provider-assessed and patient-reported outcome measures of acute skin toxicity during a Phase III trial of mometasone cream versus placebo during breast radiotherapy: the North Central Cancer Treatment Group (N06C4). Int J Radiat Oncol Biol Phys. 81, 397-402 (2011).
Miller, A.B., Hoogstraten, B., Staquet, M. & Winkler, A. Reporting results of cancer treatment. Cancer. 47, 207-214 (1981).
Sharp, L. et al. Randomized trial on two types of thermoplastic masks for patient immobilization during radiation therapy for head-and-neck cancer. Int J Radiat Oncol Biol Phys. 61, 250-256 (2005).
Sharp, L., Johansson, H., Landin, Y., Moegelin, I.M. & Bergenmar, M. Frequency and severity of skin reactions in patients with breast cancer undergoing adjuvant radiotherapy, the usefulness of two assessment instruments - a pilot study. Eur J Cancer. 47, 2665-2672 (2014).
Momm, F., Bartelt, S., Haigis, K., Grosse-Sender, A. & Witucki, G. Spectrophotometric skin measurements correlate with EORTC/RTOG-common toxicity criteria. Strahlenther Onkol. 181, 392-395 (2005).
Yoshida, K. et al. Objective assessment of dermatitis following post-operative radiotherapy in patients with breast cancer treated with breast-conserving treatment. Strahlenther Onkol. 186, 621-629 (2010).
Hu, SC. et al. Changes in biophysical properties of the skin following radiotherapy for breast cancer. J Dermatol. 41, 1087-1094 (2014).
Gonzalez-Sanchis, A. et al. Evaluation of acute skin toxicity in breast radiotherapy with a new quantitative approach. Radiother Oncol. 122, 54-59 (2017).
Schmeel, L.C. et al. Acute radiation-induced skin toxicity in hypofractionated vs. conventional whole-breast irradiation: An objective, randomized multicenter assessment using spectrophotometry. Radiother Oncol. 146, 172-179 (2020).
Itakura, H. et al. Magnetic resonance image features identify glioblastoma phenotypic subtypes with distinct molecular pathway activities. Sci Transl Med. 7, 303-338 (2015).
Song, J. et al. A review of original articles published in the emerging field of radiomics. Eur J Radiol. 127, 108991 (2020).
Saednia, K. et al. Quantitative Thermal Imaging Biomarkers to Detect Acute Skin Toxicity From Breast Radiation Therapy Using Supervised Machine Learning. Int J Radiat Oncol Biol Phys. 106, 1071-1083 (2020).
Milosevic, M., Jankovic, D. & Peulic, A. Thermography based breast cancer detection using texture features and minimum variance quantization. EXCLI J. 13, 1204-1215 (2014).
Partl, R., Lehner, J., Winkler, P. & Kapp, K.S. Testing the feasibility of augmented digital skin imaging to objectively compare the efficacy of topical treatments for radiodermatitis. PloS one. 14, e0218018 (2019).
Park, S.Y. et al. Sensitivity and stability of optically stimulated luminescence dosimeters with filled deep electron/hole traps under pre-irradiation and bleaching conditions. Phys Med. 38, 81-87 (2017).
Haralick, R.M., Shanmugam, K. & Dinstein, I. Textural Features for Image Classification. IEEE Trans Syst Man Cybern. 3, 610–621 (1973).
MathWorks. File Exchange. https://kr.mathworks.com/matlabcentral/fileexchange/22354-glcm_features4-m-vectorized-version-of-glcm_features1-m-with-code-changes?s_tid=srchtitle (2010). [accessed 8 December 2008].
Yamazaki, H. et al. Comparison of radiation dermatitis between hypofractionated and conventionally fractionated postoperative radiotherapy: objective, longitudinal assessment of skin color. Sci Rep. 8, 12306 (2018).
Park, B & Chen, Y.R. Co-occurrence matrix texture features of multi-spectral images on poultry Carcasses. Engng Res. 78, 127-139 (2001).

Huang, C.J. et al. RTOG, CTCAE and WHO criteria for acute radiation dermatitis correlate with cutaneous blood flow measurements. Breast. 24, 230-236 (2015).

No competing interests reported.

Supplementarymaterial20201216.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Quantitative Radiomics Approach to Assess Acute Radiation Dermatitis in Breast Cancer Patients

Status:

Version 1

Abstract

Figures

Introduction

Methods

Patients

Radiotherapy

Patient evaluation

Clinical assessment of acute radiation dermatitis using scoring criteria

Skin dose measurement

Acquisition of skin surface image with skin analysis device and radiomics analysis

Statistical analysis

Results

Patients and RTOG scoring

Skin dose measurement

Longitudinal assessment of changes in skin colour and texture by radiotherapy

Correlations of radiomic feature ratio with skin doses

Correlations between radiomic feature ratios for all time points

Discussion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1