PRISM: Perinuclear Ring-based Image Segmentation Method for Acute Lymphoblastic Leukemia Classification

TL;DR

PRISM is a novel image segmentation method for leukemia classification that uses adaptive concentric zones around nuclei to improve robustness and accuracy without relying on explicit cell boundary detection.

Contribution

The paper introduces PRISM, a new segmentation approach that replaces explicit membrane detection with concentric zones, enhancing robustness across staining and acquisition variability.

Findings

Achieved 98.46% accuracy in leukemia classification.

Attained a precision-recall AUC of 0.9937.

Outperformed existing methods in robustness and generalization.

Abstract

Automated analysis of peripheral blood smears for Acute Lymphoblastic Leukemia (ALL) is hindered by low contrast and substantial variability in cytoplasmic appearance, which complicate conventional membrane-based segmentation. We found that many recent approaches rely on heavy neural architectures and extensive training, but still struggle to generalize across staining and acquisition variability. To address these limitations, we propose the Perinuclear Ring-based Image Segmentation Method (PRISM), which replaces explicit cytoplasmic delineation with adaptive concentric zones constructed around the nucleus. These perinuclear regions enable the extraction of robust cytoplasmic descriptors by integrating color information with texture statistics derived from grey-level co-occurrence patterns, without requiring accurate cell-boundary detection. A calibrated stacking ensemble of traditional classifiers leverages these descriptors to achieve a high performance, with an accuracy of 98.46% and a precision-recall AUC of 0.9937.