Deep Texture Analysis—Enhancing CT Radiomics Features for Prediction of Head and Neck Cancer Treatment Outcomes: A Machine Learning Approach
(1) Background: Some cancer patients do not experience tumour shrinkage but are still at risk of experiencing unwanted treatment side effects. Radiomics refers to mining biomedical images to quantify textural characterization. When radiomics features are labelled with treatment response, retrospecti...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2024-02-01
|
Series: | Radiation |
Subjects: | |
Online Access: | https://www.mdpi.com/2673-592X/4/1/5 |
_version_ | 1797239522072723456 |
---|---|
author | Aryan Safakish Lakshmanan Sannachi Amir Moslemi Ana Pejović-Milić Gregory J. Czarnota |
author_facet | Aryan Safakish Lakshmanan Sannachi Amir Moslemi Ana Pejović-Milić Gregory J. Czarnota |
author_sort | Aryan Safakish |
collection | DOAJ |
description | (1) Background: Some cancer patients do not experience tumour shrinkage but are still at risk of experiencing unwanted treatment side effects. Radiomics refers to mining biomedical images to quantify textural characterization. When radiomics features are labelled with treatment response, retrospectively, they can train predictive machine learning (ML) models. (2) Methods: Radiomics features were determined from lymph node (LN) segmentations from treatment-planning CT scans of head and neck (H&N) cancer patients. Binary treatment outcomes (complete response versus partial or no response) and radiomics features for <i>n</i> = 71 patients were used to train support vector machine (SVM) and <i>k</i>-nearest neighbour (<i>k</i>-NN) classifier models with 1–7 features. A deep texture analysis (DTA) methodology was proposed and evaluated for second- and third-layer radiomics features, and models were evaluated based on common metrics (sensitivity (%S<sub>n</sub>), specificity (%S<sub>p</sub>), accuracy (%Acc), precision (%Prec), and balanced accuracy (%Bal Acc)). (3) Results: Models created with both classifiers were found to be able to predict treatment response, and the results suggest that the inclusion of deeper layer features enhanced model performance. The best model was a seven-feature multivariable <i>k</i>-NN model trained using features from three layers deep of texture features with %S<sub>n</sub> = 74%, %S<sub>p</sub> = 68%, %Acc = 72%, %Prec = 81%, %Bal Acc = 71% and with an area under the curve (AUC) the receiver operating characteristic (ROC) of 0.700. (4) Conclusions: H&N Cancer patient treatment-planning CT scans and LN segmentations contain phenotypic information regarding treatment response, and the proposed DTA methodology can improve model performance by enhancing feature sets and is worth consideration in future radiomics studies. |
first_indexed | 2024-04-24T17:52:52Z |
format | Article |
id | doaj.art-7bd17c9777c74771ae85fb68d9ac7611 |
institution | Directory Open Access Journal |
issn | 2673-592X |
language | English |
last_indexed | 2024-04-24T17:52:52Z |
publishDate | 2024-02-01 |
publisher | MDPI AG |
record_format | Article |
series | Radiation |
spelling | doaj.art-7bd17c9777c74771ae85fb68d9ac76112024-03-27T14:01:59ZengMDPI AGRadiation2673-592X2024-02-0141506810.3390/radiation4010005Deep Texture Analysis—Enhancing CT Radiomics Features for Prediction of Head and Neck Cancer Treatment Outcomes: A Machine Learning ApproachAryan Safakish0Lakshmanan Sannachi1Amir Moslemi2Ana Pejović-Milić3Gregory J. Czarnota4Physical Sciences, Sunnybrook Research Institute, Toronto, ON M4N 3M5, CanadaPhysical Sciences, Sunnybrook Research Institute, Toronto, ON M4N 3M5, CanadaPhysical Sciences, Sunnybrook Research Institute, Toronto, ON M4N 3M5, CanadaDepartment of Physics, Toronto Metropolitan University, Toronto, ON M5B 2K3, CanadaPhysical Sciences, Sunnybrook Research Institute, Toronto, ON M4N 3M5, Canada(1) Background: Some cancer patients do not experience tumour shrinkage but are still at risk of experiencing unwanted treatment side effects. Radiomics refers to mining biomedical images to quantify textural characterization. When radiomics features are labelled with treatment response, retrospectively, they can train predictive machine learning (ML) models. (2) Methods: Radiomics features were determined from lymph node (LN) segmentations from treatment-planning CT scans of head and neck (H&N) cancer patients. Binary treatment outcomes (complete response versus partial or no response) and radiomics features for <i>n</i> = 71 patients were used to train support vector machine (SVM) and <i>k</i>-nearest neighbour (<i>k</i>-NN) classifier models with 1–7 features. A deep texture analysis (DTA) methodology was proposed and evaluated for second- and third-layer radiomics features, and models were evaluated based on common metrics (sensitivity (%S<sub>n</sub>), specificity (%S<sub>p</sub>), accuracy (%Acc), precision (%Prec), and balanced accuracy (%Bal Acc)). (3) Results: Models created with both classifiers were found to be able to predict treatment response, and the results suggest that the inclusion of deeper layer features enhanced model performance. The best model was a seven-feature multivariable <i>k</i>-NN model trained using features from three layers deep of texture features with %S<sub>n</sub> = 74%, %S<sub>p</sub> = 68%, %Acc = 72%, %Prec = 81%, %Bal Acc = 71% and with an area under the curve (AUC) the receiver operating characteristic (ROC) of 0.700. (4) Conclusions: H&N Cancer patient treatment-planning CT scans and LN segmentations contain phenotypic information regarding treatment response, and the proposed DTA methodology can improve model performance by enhancing feature sets and is worth consideration in future radiomics studies.https://www.mdpi.com/2673-592X/4/1/5radiomicshead and neck cancerdeep texture analysistexture of textureresponse predictiontexture features |
spellingShingle | Aryan Safakish Lakshmanan Sannachi Amir Moslemi Ana Pejović-Milić Gregory J. Czarnota Deep Texture Analysis—Enhancing CT Radiomics Features for Prediction of Head and Neck Cancer Treatment Outcomes: A Machine Learning Approach Radiation radiomics head and neck cancer deep texture analysis texture of texture response prediction texture features |
title | Deep Texture Analysis—Enhancing CT Radiomics Features for Prediction of Head and Neck Cancer Treatment Outcomes: A Machine Learning Approach |
title_full | Deep Texture Analysis—Enhancing CT Radiomics Features for Prediction of Head and Neck Cancer Treatment Outcomes: A Machine Learning Approach |
title_fullStr | Deep Texture Analysis—Enhancing CT Radiomics Features for Prediction of Head and Neck Cancer Treatment Outcomes: A Machine Learning Approach |
title_full_unstemmed | Deep Texture Analysis—Enhancing CT Radiomics Features for Prediction of Head and Neck Cancer Treatment Outcomes: A Machine Learning Approach |
title_short | Deep Texture Analysis—Enhancing CT Radiomics Features for Prediction of Head and Neck Cancer Treatment Outcomes: A Machine Learning Approach |
title_sort | deep texture analysis enhancing ct radiomics features for prediction of head and neck cancer treatment outcomes a machine learning approach |
topic | radiomics head and neck cancer deep texture analysis texture of texture response prediction texture features |
url | https://www.mdpi.com/2673-592X/4/1/5 |
work_keys_str_mv | AT aryansafakish deeptextureanalysisenhancingctradiomicsfeaturesforpredictionofheadandneckcancertreatmentoutcomesamachinelearningapproach AT lakshmanansannachi deeptextureanalysisenhancingctradiomicsfeaturesforpredictionofheadandneckcancertreatmentoutcomesamachinelearningapproach AT amirmoslemi deeptextureanalysisenhancingctradiomicsfeaturesforpredictionofheadandneckcancertreatmentoutcomesamachinelearningapproach AT anapejovicmilic deeptextureanalysisenhancingctradiomicsfeaturesforpredictionofheadandneckcancertreatmentoutcomesamachinelearningapproach AT gregoryjczarnota deeptextureanalysisenhancingctradiomicsfeaturesforpredictionofheadandneckcancertreatmentoutcomesamachinelearningapproach |