Related Experiment Video
Updated: Dec 13, 2025
06:17
Analysis of Multidimensional Microscopy Data Using Cell-ACDC
Published on: November 7, 2025
335
DeepACEv2: Automated Chromosome Enumeration in Metaphase Cell Images Using Deep Convolutional Neural Networks.
IEEE Transactions on Medical Imaging
|August 4, 2020
Summary
DeepACEv2 automates chromosome enumeration in karyotyping using advanced object detection. This framework significantly improves accuracy in counting chromosomes, even with challenging occlusions, advancing automated cytogenetic analysis.
Area of Science:
- Computational Biology
- Genetics
- Medical Imaging
Background:
- Chromosome enumeration is critical for karyotyping but is labor-intensive.
- Automating this process can enhance efficiency and accuracy in cytogenetic analysis.
Purpose of the Study:
- To develop an automated chromosome enumeration framework, DeepACEv2.
- To improve the accuracy of chromosome counting, especially in cases of occlusion and partial chromosomes.
Main Methods:
- Utilized a ResNet-101 backbone with Feature Pyramid Network (FPN) for multi-level feature extraction.
- Incorporated Hard Negative Anchors Sampling to detect partial chromosomes.
- Introduced a Template Module and modified Non-Maximum Suppression (NMS) to handle chromosome occlusion.
- Developed a Truncated Normalized Repulsion Loss to refine localization.
Main Results:
- DeepACEv2 achieved a Whole Correct Ratio (WCR) of 71.39% on 1375 clinical metaphase images.
- The framework demonstrated an Average Error Ratio (AER) of approximately 1.17% for chromosome counting.
- Ablation studies confirmed the effectiveness of individual modules within the DeepACEv2 framework.
Conclusions:
- DeepACEv2 significantly outperforms previous methods for automated chromosome enumeration.
- The developed framework offers a robust solution for accurate chromosome counting in clinical karyotyping.
- This automated approach has the potential to streamline cytogenetic analysis workflows.
