About half of all cancer patients receive radiation therapy throughout their illness, thus it continues to be a vital component of cancer treatment. However, a significant number of these patients suffer from radiation-induced skin damage or acute radiation dermatitis (ARD). Severe discomfort, difficulties with everyday tasks, a general decline in quality of life, and occasionally the need to forgo required radiation therapy are all side effects of ARD that have an adverse effect on survival rates. Unfortunately, research on the causes of ARD and prospective therapeutic methods has been hampered by the absence of biomarkers to quantitatively assess early changes related to ARD. In order to identify low-grade ARD, this study will use optical coherence tomography (OCT) images coupled with images from traditional image intensity and novel features. Twenty-two patients had imaging twice weekly during radiation therapy, producing a total of 1487 pictures. Each case's severity was assessed by an experienced oncologist. The preliminary results of the research show that a deep learning approach achieved an 88% accuracy in distinguishing between normal skin and early ARD. These findings provide a promising foundation for further studies aimed at creating a quantitative assessment tool to improve the management of ARD.
Radiation therapy remains an essential component of cancer treatment, with nearly 50% of cancer patients receiving radiation therapy at some point during the course of their illness. Of those, the vast majority experience some form of acute radiation dermatitis (ARD) or radiation-induced skin injury. ARD results in significant discomfort, restriction of daily activities, overall decrease in the quality of life and even cessation of the necessary radiation therapy with detrimental survival effects. Unfortunately, research into the causes and possible management strategies for ARD is hindered by the lack of biomarkers for the quantitative assessment of the early changes associated with the condition. This study provides the basis to identify low grade ARD, using Optical Coherence Tomography (OCT) images with the extraction of conventional image intensity and novel features. Twenty-two patients were imaged twice each week until the end of their radiation treatment resulting in 1487 images. The severity of the cases was graded by an expert oncologist. The preliminary results of various machine learning (ML) classifiers have shown that an LDA based approach provided the best performance, separating normal skin from very early ARD, with an accuracy of 85%.
Radiation therapy remains an essential component of cancer treatment, with nearly 50% of cancer patients receiving radiation therapy at some point during the course of their illness. Of those, as many as 90-95% may experience some form of acute radiation dermatitis (ARD) or radiation-induced skin injury. ARD results in significant discomfort, restriction of daily activities, overall decrease in the quality of life and even cessation of the necessary radiation therapy with detrimental survival effects. Unfortunately, research into the causes and possible management strategies for ARD is hindered by the lack of biomarkers for the quantitative assessment of the early changes associated with the condition. This study provides the basis to yield such novel biomarkers using Optical Coherence Tomography (OCT) images with the extraction of conventional image intensity and novel features. Patients were imaged twice each week over the six-week course of their radiation treatment. The severity of the cases was graded by an expert oncologist. Preliminary results, separating normal skin from early ARD, were very promising, yielding an accuracy of 88.3%.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.