Scholarlink

The need for accurate diagnosis requires blood smear image assessment yet this manual approach demands substantial human effort and can produce inaccurate results. Here, we explore the use of transfer learning with deep learning models to automate the classification of ALL cells into four distinct stages: benign, pre-B ALL, pro-B ALL, and early pre-B ALL. We fine-tuned three pre-trained models consisting of ResNet-50 and VGG-16 and EfficientNetB0 using microscopic ALL cell image data. The top-performing model proved to be ResNet-50 achieving 98% accuracy rate above VGG-16 and EfficientNetB0 with 96% accuracy. The analysis indicates transfer learning enhances classification performance to generate a dependable tool that facilitates ALL diagnosis. This research shows that deep learning algorithms enhance diagnostic precision and help diminish variability and strengthen medical choices for treating leukemia.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Leveraging Transfer Learning for Acute Lymphoblastic Leukemia Cell Classification: A Deep Learning Approach

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Akanksha Kochhar,
  • Preeti Kaur

摘要

The need for accurate diagnosis requires blood smear image assessment yet this manual approach demands substantial human effort and can produce inaccurate results. Here, we explore the use of transfer learning with deep learning models to automate the classification of ALL cells into four distinct stages: benign, pre-B ALL, pro-B ALL, and early pre-B ALL. We fine-tuned three pre-trained models consisting of ResNet-50 and VGG-16 and EfficientNetB0 using microscopic ALL cell image data. The top-performing model proved to be ResNet-50 achieving 98% accuracy rate above VGG-16 and EfficientNetB0 with 96% accuracy. The analysis indicates transfer learning enhances classification performance to generate a dependable tool that facilitates ALL diagnosis. This research shows that deep learning algorithms enhance diagnostic precision and help diminish variability and strengthen medical choices for treating leukemia.