An Efficient Acute Lymphoblastic Leukemia Screen Framework Based on Multi-Modal Deep Neural Network

Area of Science:

• Medical Diagnostics
• Artificial Intelligence in Healthcare
• Hematology

Background:

• Acute lymphoblastic leukemia (ALL) is a primary cause of mortality in pediatric cancers.
• Timely and accurate ALL diagnosis is vital for effective treatment and improved survival rates.

Purpose of the Study:

• To develop and validate a multi-modal deep neural network for early and efficient screening of ALL.
• To leverage both white blood cell (WBC) scattergrams and complete blood count (CBC) data for enhanced ALL detection.

Main Methods:

• A deep learning framework integrating WBC scattergrams and CBC data was proposed.
• The model was trained and validated on a dataset including patients with ALL, infectious mononucleosis (IM), and healthy controls (HCs).

Main Results:

• The combined approach achieved 98.43% accuracy in cross-validation and 96.67% sensitivity in external validation.
• The model's area under the curve (AUC) exceeded 0.99, surpassing human expert performance.

Conclusions:

• This framework represents a novel integration of WBC scattergrams and CBC data for ALL screening.
• The method demonstrates high sensitivity and specificity, promising improved early ALL diagnosis and reduced workload for medical staff.