An XAI Integrated Identification System of White Blood Cell Type Using Variants of Vision Transformer
Paper
Dipto, Shakib Mahmud, Reza, Md Tanzim, Rahman, Md Nowroz Junaed, Parvez, Mohammad Zavid, Barua, Prabal Datta and Chakraborty, Subrata. 2023. "An XAI Integrated Identification System of White Blood Cell Type Using Variants of Vision Transformer." Second International Conference on Innovations in Computing Research (ICR’23). Madrid, Spain 04 - 06 Sep 2023 Spain. https://doi.org/10.1007/978-3-031-35308-6_26
| Paper/Presentation Title | An XAI Integrated Identification System of White Blood Cell Type Using Variants of Vision Transformer |
|---|---|
| Presentation Type | Paper |
| Authors | Dipto, Shakib Mahmud, Reza, Md Tanzim, Rahman, Md Nowroz Junaed, Parvez, Mohammad Zavid, Barua, Prabal Datta and Chakraborty, Subrata |
| Journal or Proceedings Title | Proceedings of the Second International Conference on Innovations in Computing Research (ICR’23) |
| Journal Citation | 721, pp. 303-315 |
| Year | 2023 |
| Place of Publication | Spain |
| ISBN | 9783031353079 |
| 9783031353086 | |
| Digital Object Identifier (DOI) | https://doi.org/10.1007/978-3-031-35308-6_26 |
| Web Address (URL) of Paper | https://link.springer.com/chapter/10.1007/978-3-031-35308-6_26 |
| Web Address (URL) of Conference Proceedings | https://link.springer.com/book/10.1007/978-3-031-35308-6 |
| Conference/Event | Second International Conference on Innovations in Computing Research (ICR’23) |
| Event Details | Second International Conference on Innovations in Computing Research (ICR’23) Delivery In person Event Date 04 to end of 06 Sep 2023 Event Location Madrid, Spain |
| Abstract | White Blood Cells (WBCs) serve as one of the primary defense mechanisms against various diseases. Therefore, in order to detect blood cancer as well as many other disorders, routine WBC monitoring may be necessary. Numerous studies have proposed automated 4 types of WBC detection through Machine Learning and Deep Learning based solutions. However, transformers based applications, which primarily originated from the field of Natural Language Processing, are very scarce. Our proposed study showcases the applications of Vision Transformers (VTs) for WBC type identification. Firstly, a pre-augmented dataset of nearly 12,500 images was taken. Afterward, two variants of VTs were trained and evaluated on the dataset. Our analysis revealed that the accuracy for all the models ranged from 83% to 85%, making the performance of the VTs equivalent to that of the standard Deep Learning models. Meanwhile, VTs have demonstrated significantly faster learning symptoms during the training phase, which can be useful when one wants to maximize learning through fewer epochs, for example, in a federated learning environment. Finally, the application of Explainable AI (XAI) was visualized on the VTs using Gradient-weighted Class Activation Mapping (GradCam). |
| Keywords | Blood Cell; Vision Transformer; GradCam; White Blood Cell; Monocyte; Lymphocyte; Neutrophil; Eosinophil; Transformer |
| ANZSRC Field of Research 2020 | 461299. Software engineering not elsewhere classified |
| Public Notes | The accessible file is the accepted version of the paper. Please refer to the URL for the published version. |
| Series | Lecture Notes in Networks and Systems |
| Byline Affiliations | BRAC University, Bangladesh |
| Asia Pacific International College (APIC), Australia | |
| Torrens University | |
| Australian Catholic University | |
| Charles Sturt University | |
| School of Business | |
| University of New England | |
| Cogninet Australia, Australia |
Permalink -
https://research.usq.edu.au/item/z275x/an-xai-integrated-identification-system-of-white-blood-cell-type-using-variants-of-vision-transformer
Download files
Accepted Version
79
total views52
total downloads3
views this month4
downloads this month