Early Detection of Acute Myeloid Leukemia (AML) Using YOLOv12 Deep Learning Model

2026-04-21 · Source: cs.AI updates on arXiv.org · Field: Science & Research — Health & Medical Research, Mathematics & Computational Sciences, Life Sciences & Biology · Depth: Advanced, long

Summary

A study utilized the YOLOv12 deep learning model for the multiclass classification of Acute Myeloid Leukemia (AML) cells, a challenging task due to visual similarities between cell types. The research applied two segmentation approaches, Hue channel and Otsu thresholding, based on cell and nucleus features, to preprocess images. Experiments demonstrated that YOLOv12, when combined with Otsu thresholding on cell-based segmentation, achieved the highest validation and test accuracy, both reaching 99.3%. The dataset comprised five key blood cell types relevant to AML: myeloblasts, segmented neutrophils, basophils, monocytes, and erythroblasts, sourced from Kaggle, The Cancer Imaging Archive (TCIA), and Hospital Clinic of Barcelona, with images captured at 1000x magnification and 1024x1024 pixel resolution.

Key takeaway

For Computer Vision Engineers developing diagnostic tools for hematological cancers, this research indicates that integrating YOLOv12 with Otsu thresholding for cell-based image segmentation offers a highly accurate approach for multiclass AML cell classification. You should consider this combination to enhance the reliability and performance of your early detection systems, especially when working with diverse myeloid cell types. This method could significantly improve diagnostic precision.

Key insights

YOLOv12 with Otsu thresholding achieved 99.3% accuracy in multiclass AML cell classification.

Principles

• Image segmentation improves deep learning classification.
• Area Attention enhances YOLOv12 speed and accuracy.

Method

The method involves segmenting blood cell images using Hue channel or Otsu thresholding, then training and validating YOLOv12, ResNet50, and InceptionResNet50 v2 models on the segmented datasets.

In practice

• Use Otsu thresholding for cell image segmentation.
• Prioritize YOLOv12 for high-accuracy AML detection.
• Balance datasets to improve model generalization.

Topics

• Acute Myeloid Leukemia
• YOLOv12
• Deep Learning
• Image Segmentation
• Otsu Thresholding

Best for: Computer Vision Engineer, AI Scientist, Machine Learning Engineer, Research Scientist

Related on AIssential

• Early Detection of Acute Myeloid Leukemia (AML) Using YOLOv12 Deep Learning Model — YOLOv12 with Otsu thresholding on cell-based segmentation achieves 99.3% accuracy for AML cell classification.
• DualGate-Net: A Prior-Gated Dual-Encoder Framework for Histopathology Cell Detection — Adaptive, gated fusion of local and global features with contextual priors improves histopathology cell detection.
• YOLOv26: An Object Detector Built for Real-Time Deployment — YOLOv26 prioritizes edge deployment through architectural simplification and hardware-friendly optimizations.
• Colour Extraction Pipeline for Odonates using Computer Vision — A computer vision pipeline segments Odonates and extracts color data to enable large-scale ecological correlation analysis.
• SegMix:Shuffle-based Feedback Learning for Semantic Segmentation of Pathology Images — SegMix uses shuffle-based feedback learning to generate high-quality pseudo-segmentation masks from image-level labels.
• DeepHistoViT: An Interpretable Vision Transformer Framework for Histopathological Cancer Classification — DeepHistoViT uses Vision Transformers with attention for interpretable, high-accuracy histopathological cancer classification.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.AI updates on arXiv.org.