Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Breaking Down the Hierarchy: A New Approach to Leukemia Classification

Ibraheem Hamdi, Hosam El-Gendy, Ahmed Sharshar, Mohamed Saeed, Muhammad Ridzuan, Shahrukh K. Hashmi, Naveed Syed, Imran Mirza, Shakir Hussain, Amira Mahmoud Abdalla, Mohammad Yaqub

TL;DR

The paper tackles leukemia subtype classification from blood smear images, addressing diagnostic errors stemming from manual morphology. It introduces a hierarchical multi-label deep-learning framework that leverages ConvNeXt-Tiny and ViT-Small, comparing a flat baseline to a hierarchical model with per-level losses and slide-level mode aggregation. Experiments show the hierarchical ViT-based approach achieving up to $90.97\%$ accuracy across seven leaf classes, with Grad-CAM visualizations providing interpretability of the model's decisions. This method has potential to reduce pathologist workload, lower testing costs, and enable scalable, non-invasive leukemia subtyping in clinical settings.

Abstract

The complexities inherent to leukemia, multifaceted cancer affecting white blood cells, pose considerable diagnostic and treatment challenges, primarily due to reliance on laborious morphological analyses and expert judgment that are susceptible to errors. Addressing these challenges, this study presents a refined, comprehensive strategy leveraging advanced deep-learning techniques for the classification of leukemia subtypes. We commence by developing a hierarchical label taxonomy, paving the way for differentiating between various subtypes of leukemia. The research further introduces a novel hierarchical approach inspired by clinical procedures capable of accurately classifying diverse types of leukemia alongside reactive and healthy cells. An integral part of this study involves a meticulous examination of the performance of Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs) as classifiers. The proposed method exhibits an impressive success rate, achieving approximately 90\% accuracy across all leukemia subtypes, as substantiated by our experimental results. A visual representation of the experimental findings is provided to enhance the model's explainability and aid in understanding the classification process.

Breaking Down the Hierarchy: A New Approach to Leukemia Classification

TL;DR

The paper tackles leukemia subtype classification from blood smear images, addressing diagnostic errors stemming from manual morphology. It introduces a hierarchical multi-label deep-learning framework that leverages ConvNeXt-Tiny and ViT-Small, comparing a flat baseline to a hierarchical model with per-level losses and slide-level mode aggregation. Experiments show the hierarchical ViT-based approach achieving up to accuracy across seven leaf classes, with Grad-CAM visualizations providing interpretability of the model's decisions. This method has potential to reduce pathologist workload, lower testing costs, and enable scalable, non-invasive leukemia subtyping in clinical settings.

Abstract

The complexities inherent to leukemia, multifaceted cancer affecting white blood cells, pose considerable diagnostic and treatment challenges, primarily due to reliance on laborious morphological analyses and expert judgment that are susceptible to errors. Addressing these challenges, this study presents a refined, comprehensive strategy leveraging advanced deep-learning techniques for the classification of leukemia subtypes. We commence by developing a hierarchical label taxonomy, paving the way for differentiating between various subtypes of leukemia. The research further introduces a novel hierarchical approach inspired by clinical procedures capable of accurately classifying diverse types of leukemia alongside reactive and healthy cells. An integral part of this study involves a meticulous examination of the performance of Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs) as classifiers. The proposed method exhibits an impressive success rate, achieving approximately 90\% accuracy across all leukemia subtypes, as substantiated by our experimental results. A visual representation of the experimental findings is provided to enhance the model's explainability and aid in understanding the classification process.

Paper Structure

This paper contains 18 sections, 1 equation, 3 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Dataset
  4. Pre-Processing
  5. Methodology Overview
  6. Flat/Leaf Classification (Baseline)
  7. Hierarchical Multi-Label Classification
  8. Experimental Setup
  9. Model Selection
  10. Simulating Pathological Evaluation
  11. Experimental Procedures
  12. Evaluation Metrics
  13. Results and Discussion
  14. Flat/Leaf Classification
  15. Base vs Proposed Hierarchical Classification
  16. ...and 3 more sections

Figures (3)

  • Figure 1: Flat/Leaf Classification Structure.
  • Figure 2: Hierarchical structure of leukemia subtypes, showcasing different levels.
  • Figure 3: Qualitative analysis of the ViT results using Grad-CAM. From left to right: CLL, CML, and Normal samples. From top to bottom: Input image, leaf classification output, and hierarchical classification output. The leaf model output has less discriminative distinction between WBCs and RBCs.