ABSTRACT
Although significant advancements in computer-aided diagnostics using artificial intelligence (AI) have been made, to date, no viable method for radiation-induced skin reaction (RISR) analysis and classification is available. The objective of this single-center study was to develop machine learning and deep learning approaches using deep convolutional neural networks (CNNs) for automatic classification of RISRs according to the Common Terminology Criteria for Adverse Events (CTCAE) grading system. Scarletred® Vision, a novel and state-of-the-art digital skin imaging method capable of remote monitoring and objective assessment of acute RISRs was used to convert 2D digital skin images using the CIELAB color space and conduct SEV* measurements. A set of different machine learning and deep convolutional neural network-based algorithms has been explored for the automatic classification of RISRs. A total of 2263 distinct images from 209 patients were analyzed for training and testing the machine learning and CNN algorithms. For a 2-class problem of healthy skin (grade 0) versus erythema (grade≥ 1), all machine learning models produced an accuracy of above 70%, and the sensitivity and specificity of erythema recognition were 67-72% and 72-83%, respectively. The CNN produced a test accuracy of 74%, sensitivity of 66%, and specificity of 83% for predicting healthy and erythema cases. For the severity grade prediction of a 3-class problem (grade 0 versus 1 versus 2), the test accuracy was 60-67%, and the sensitivity and specificity were 56-82%, 35-59%, and 65-72%, respectively. For estimating the severity grade of each class, the CNN obtained an accuracy of 73%, 66%, and 82%, respectively. Ensemble learning combines several individual predictions to obtain a better generalization performance. Furthermore, we exploited ensemble learning by deploying a CNN model as a meta-learner. The ensemble CNN based on bagging and majority voting shows an accuracy, sensitivity and specificity of 87%, 90%, and 82% for a 2-class problem, respectively. For a 3-class problem, the ensemble CNN shows an overall accuracy of 66%, while for each grade (0, 1, and 2) accuracies were 0.76%, 0.69%, and 0.87%, sensitivities were 0.70%, 0.57%, and 0.71%, and specificities were 0.78%, 0.75%, and 0.95%, respectively. This study is the first to focus on erythema in radiation-dermatitis and produces benchmark results using machine learning models. The outcome of this study validates that the proposed system can act as a pre-screening and decision support tool for oncologists or patients to provide fast, reliable, and efficient assessment of erythema grading.
Competing Interest Statement
The authors have declared no competing interest.
Clinical Trial
This is a medical device study carried out in Austria under the Ethic commission number: 30-023 ex 17/18
Funding Statement
This work was funded by SCARLETRED Holding GmbH
Author Declarations
I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained.
Yes
The details of the IRB/oversight body that provided approval or exemption for the research described are given below:
Our clinical study was carried out under the study title "Validating the method for objective remote assessment of radiation dermatitis by augmented digital skin imaging using Scarletred Vision" and approved under the Ethics Committee (EC) number 30-023 ex 17/18 by the EC board of the Medical University of Graz Augenbruggerplatz 2, A-8036, Graz Austria, on the date of 06. April 2018; Primary investigator and applicant: Dr. med. Richard Partl; Name of the chairman of the EC, who signed the EC approval letter: Univ. Prof. DI Dr. Josef Haas; Name of the deputy chairman of the EC, who co-signed the EC approval letter: Univ. Prof. Dr. Hermann Toplak; List of EC members (21 people) who were present in the meeting and who where entitle to vote: 1) Univ.Prof.Dl.Dr. Andrea Berghold, 2) Univ.Prof.Dr. Josef Donnerer, 3) Univ.Prof.Dl.Dr. Josef Haas, 4) Univ.Prof.Dr. Wolfgang Holzer, 5) Univ.Prof.Dr. Matthias Klatt, 6) OSr., Mag. Gabriele Moestl, 7) Univ.Prof.Dr. Leopold Neuhold, 8) Univ.Prof.Dr. Friedrich Reiterer, 9) Dl Dr. Regina Riedl, 10) Univ.Prof.Dr. Rudolf Stauber, 11) Univ.Prof.Dr. Hermann Toplak Ursula Vennemann, 12) Univ.Prof.Dr. Andreas Zimmer, 13) Univ.Prof.Dr. Werner Aberer, 14) Ing.Franz Deutschmann, 15) PD.Dr. Christian Fazekas, 16) Univ.Prof.Dr. Wolfgang Kroell, 17) Mag.Renate Skledar, 18) Univ.Prof.Dr. Michael Speicher; List of medical specialist who were present in the EC meeting and during the voting process: 19) Univ.Prof.Dr.Edgar Petru, 20) Univ.Prof.Dr.Heidi Stranzl-Lawatsch,
All necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived.
Yes
I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance).
Yes
I have followed all appropriate research reporting guidelines and uploaded the relevant EQUATOR Network research reporting checklist(s) and other pertinent material as supplementary files, if applicable.
Yes
Data Availability
Since the raw data (skin images) relates to sensitive health data from cancer patients, therefore the data cannot be made available under the public domain.
