The Effectiveness of Image Augmentation in Deep Learning Networks for Detecting COVID-19: A Geometric Transformation Perspective
Chest X-ray imaging technology used for the early detection and screening of COVID-19 pneumonia is both accessible worldwide and affordable compared to other non-invasive technologies. Additionally, deep learning methods have recently shown remarkable results in detecting COVID-19 on chest X-rays, m...
| Main Authors: | , , , , , , , , , , , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Published: |
Frontiers Media SA
2021
|
| Online Access: | https://hdl.handle.net/1721.1/130233 |
| _version_ | 1826203255538974720 |
|---|---|
| author | Elgendi, Mohamed Nasir, Muhammad Umer Tang, Qunfeng Smith, David Grenier, John-Paul Batte, Catherine Spieler, Bradley Leslie, William Donald Menon, Carlo Fletcher, Richard R Howard, Newton Ward, Rabab Parker, William Nicolaou, Savvas |
| author2 | Massachusetts Institute of Technology. Department of Urban Studies and Planning |
| author_facet | Massachusetts Institute of Technology. Department of Urban Studies and Planning Elgendi, Mohamed Nasir, Muhammad Umer Tang, Qunfeng Smith, David Grenier, John-Paul Batte, Catherine Spieler, Bradley Leslie, William Donald Menon, Carlo Fletcher, Richard R Howard, Newton Ward, Rabab Parker, William Nicolaou, Savvas |
| author_sort | Elgendi, Mohamed |
| collection | MIT |
| description | Chest X-ray imaging technology used for the early detection and screening of COVID-19 pneumonia is both accessible worldwide and affordable compared to other non-invasive technologies. Additionally, deep learning methods have recently shown remarkable results in detecting COVID-19 on chest X-rays, making it a promising screening technology for COVID-19. Deep learning relies on a large amount of data to avoid overfitting. While overfitting can result in perfect modeling on the original training dataset, on a new testing dataset it can fail to achieve high accuracy. In the image processing field, an image augmentation step (i.e., adding more training data) is often used to reduce overfitting on the training dataset, and improve prediction accuracy on the testing dataset. In this paper, we examined the impact of geometric augmentations as implemented in several recent publications for detecting COVID-19. We compared the performance of 17 deep learning algorithms with and without different geometric augmentations. We empirically examined the influence of augmentation with respect to detection accuracy, dataset diversity, augmentation methodology, and network size. Contrary to expectation, our results show that the removal of recently used geometrical augmentation steps actually improved the Matthews correlation coefficient (MCC) of 17 models. The MCC without augmentation (MCC = 0.51) outperformed four recent geometrical augmentations (MCC = 0.47 for Data Augmentation 1, MCC = 0.44 for Data Augmentation 2, MCC = 0.48 for Data Augmentation 3, and MCC = 0.49 for Data Augmentation 4). When we retrained a recently published deep learning without augmentation on the same dataset, the detection accuracy significantly increased, with a χ[superscript 2]McNemar's statistic = 163.2 and a p-value of 2.23 × 10[superscript −37. This is an interesting finding that may improve current deep learning algorithms using geometrical augmentations for detecting COVID-19. We also provide clinical perspectives on geometric augmentation to consider regarding the development of a robust COVID-19 X-ray-based detector. |
| first_indexed | 2024-09-23T12:33:45Z |
| format | Article |
| id | mit-1721.1/130233 |
| institution | Massachusetts Institute of Technology |
| last_indexed | 2024-09-23T12:33:45Z |
| publishDate | 2021 |
| publisher | Frontiers Media SA |
| record_format | dspace |
| spelling | mit-1721.1/1302332022-10-01T09:48:01Z The Effectiveness of Image Augmentation in Deep Learning Networks for Detecting COVID-19: A Geometric Transformation Perspective Elgendi, Mohamed Nasir, Muhammad Umer Tang, Qunfeng Smith, David Grenier, John-Paul Batte, Catherine Spieler, Bradley Leslie, William Donald Menon, Carlo Fletcher, Richard R Howard, Newton Ward, Rabab Parker, William Nicolaou, Savvas Massachusetts Institute of Technology. Department of Urban Studies and Planning Massachusetts Institute of Technology. Device Research Laboratory Massachusetts Institute of Technology. Department of Mechanical Engineering Chest X-ray imaging technology used for the early detection and screening of COVID-19 pneumonia is both accessible worldwide and affordable compared to other non-invasive technologies. Additionally, deep learning methods have recently shown remarkable results in detecting COVID-19 on chest X-rays, making it a promising screening technology for COVID-19. Deep learning relies on a large amount of data to avoid overfitting. While overfitting can result in perfect modeling on the original training dataset, on a new testing dataset it can fail to achieve high accuracy. In the image processing field, an image augmentation step (i.e., adding more training data) is often used to reduce overfitting on the training dataset, and improve prediction accuracy on the testing dataset. In this paper, we examined the impact of geometric augmentations as implemented in several recent publications for detecting COVID-19. We compared the performance of 17 deep learning algorithms with and without different geometric augmentations. We empirically examined the influence of augmentation with respect to detection accuracy, dataset diversity, augmentation methodology, and network size. Contrary to expectation, our results show that the removal of recently used geometrical augmentation steps actually improved the Matthews correlation coefficient (MCC) of 17 models. The MCC without augmentation (MCC = 0.51) outperformed four recent geometrical augmentations (MCC = 0.47 for Data Augmentation 1, MCC = 0.44 for Data Augmentation 2, MCC = 0.48 for Data Augmentation 3, and MCC = 0.49 for Data Augmentation 4). When we retrained a recently published deep learning without augmentation on the same dataset, the detection accuracy significantly increased, with a χ[superscript 2]McNemar's statistic = 163.2 and a p-value of 2.23 × 10[superscript −37. This is an interesting finding that may improve current deep learning algorithms using geometrical augmentations for detecting COVID-19. We also provide clinical perspectives on geometric augmentation to consider regarding the development of a robust COVID-19 X-ray-based detector. NSERC (Grant RGPIN-2014-04462) 2021-03-22T22:39:31Z 2021-03-22T22:39:31Z 2021-03 2020-09 Article http://purl.org/eprint/type/JournalArticle 2296-858X https://hdl.handle.net/1721.1/130233 Elgendi, Mohamed et al. "The Effectiveness of Image Augmentation in Deep Learning Networks for Detecting COVID-19: A Geometric Transformation Perspective." Frontiers in Medicine 8 (March 2021): 629134. © 2021 Elgendi et al. https://doi.org/10.3389/fmed.2021.629134 Frontiers in Medicine Creative Commons Attribution 4.0 International license https://creativecommons.org/licenses/by/4.0/ application/pdf Frontiers Media SA Frontiers |
| spellingShingle | Elgendi, Mohamed Nasir, Muhammad Umer Tang, Qunfeng Smith, David Grenier, John-Paul Batte, Catherine Spieler, Bradley Leslie, William Donald Menon, Carlo Fletcher, Richard R Howard, Newton Ward, Rabab Parker, William Nicolaou, Savvas The Effectiveness of Image Augmentation in Deep Learning Networks for Detecting COVID-19: A Geometric Transformation Perspective |
| title | The Effectiveness of Image Augmentation in Deep Learning Networks for Detecting COVID-19: A Geometric Transformation Perspective |
| title_full | The Effectiveness of Image Augmentation in Deep Learning Networks for Detecting COVID-19: A Geometric Transformation Perspective |
| title_fullStr | The Effectiveness of Image Augmentation in Deep Learning Networks for Detecting COVID-19: A Geometric Transformation Perspective |
| title_full_unstemmed | The Effectiveness of Image Augmentation in Deep Learning Networks for Detecting COVID-19: A Geometric Transformation Perspective |
| title_short | The Effectiveness of Image Augmentation in Deep Learning Networks for Detecting COVID-19: A Geometric Transformation Perspective |
| title_sort | effectiveness of image augmentation in deep learning networks for detecting covid 19 a geometric transformation perspective |
| url | https://hdl.handle.net/1721.1/130233 |
| work_keys_str_mv | AT elgendimohamed theeffectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT nasirmuhammadumer theeffectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT tangqunfeng theeffectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT smithdavid theeffectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT grenierjohnpaul theeffectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT battecatherine theeffectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT spielerbradley theeffectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT lesliewilliamdonald theeffectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT menoncarlo theeffectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT fletcherrichardr theeffectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT howardnewton theeffectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT wardrabab theeffectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT parkerwilliam theeffectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT nicolaousavvas theeffectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT elgendimohamed effectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT nasirmuhammadumer effectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT tangqunfeng effectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT smithdavid effectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT grenierjohnpaul effectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT battecatherine effectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT spielerbradley effectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT lesliewilliamdonald effectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT menoncarlo effectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT fletcherrichardr effectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT howardnewton effectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT wardrabab effectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT parkerwilliam effectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective AT nicolaousavvas effectivenessofimageaugmentationindeeplearningnetworksfordetectingcovid19ageometrictransformationperspective |