What is it about?
The research was about developing an advanced deep learning model to improve the diagnosis of acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML), two major types of blood disorders. The study used two popular deep learning networks, ResNet-50 and VGG-19, but with modified weights and learning parameters. The researchers also introduced a new model, referred to as "i-Net", which has additional convolutional layers and fine-tuned hyperparameters for the classification of normal and cancerous white blood cells.
Featured Image
Photo by Steve Johnson on Unsplash
Why is it important?
This research is crucial for several reasons: 1) Improving Leukemia Diagnosis: Acute lymphoblastic leukaemia (ALL) and acute myeloid leukaemia (AML) are serious blood disorders, and their early and accurate diagnosis can greatly influence treatment effectiveness and patient survival. By leveraging deep learning models like the proposed i-Net, this study is advancing the potential for more accurate and efficient diagnosis. 2) Artificial Intelligence in Healthcare: This study further emphasizes the role of AI and machine learning in healthcare. It demonstrates that deep learning models can be successfully applied in clinical settings, potentially reducing the workload of clinicians and enabling more accurate diagnoses. 3) Innovation in AI Models: The research introduces a novel deep learning model, i-Net, which outperforms existing models in terms of accuracy. It signifies the possibility of improving existing models or creating new ones that could enhance various applications in the healthcare sector. 4) Potential for Integration into Clinical Decision Support Systems: The high accuracy of the proposed i-Net model indicates its potential to be incorporated into clinical decision support systems (CDSS), which could help clinicians make more informed decisions about diagnosis and treatment options for leukemia.
Perspectives
The work was conducted using datasets obtained from the cancer imaging archive (TCIA) repository, and various techniques like data augmentation, dropout regularization, and batch normalization were applied to reduce overfitting in the models. Further, it demonstrated that the newly proposed i-Net model performed better than the pre-existing models, achieving a validation accuracy of 99.18%. This high accuracy suggests the potential of using this model in clinical decision support systems for more effective leukemia detection. However, it was noted that the accuracy of the i-Net model decreased after the 30th training cycle (epoch) when trained without early stoppage, which implies room for further optimization.
Mr Victor Ikechukwu Agughasi
Maharaja Institute of Technology
Read the Original
This page is a summary of: i-Net: a deep CNN model for white blood cancer segmentation and classification, International Journal of Advanced Technology and Engineering Exploration, October 2022, Association of Computer, Communication and Education for National Triumph Social and Welfare Society (ACCENTS),
DOI: 10.19101/ijatee.2021.875564.
You can read the full text:
Contributors
The following have contributed to this page
