Abstract
White blood cells (WBCs) on Giemsa-stained thick blood smears look like early-stage Plasmodium falciparum ring-form trophozoites in three respects: they are small, round, and stain intensely purple. This makes them a plausible source of false positives in parasite-only detectors, but the link has not been tested. We trained two YOLOv12s models on the Lacuna Malaria Detection dataset (8,000 images from Uganda and Ghana), identical except that Model B adds WBC labels, and asked whether false positives (FPs) fall preferentially
near annotated WBCs. They do not: in both models 95% of FPs are background detections, and none overlap a WBC annotation at the class-confusion threshold. When the comparison holds the observed detection pattern fixed and contrasts FPs against the model’s own correct detections, no test places FPs closer to WBCs than true positives at any of three confidence thresholds. We also show that two natural analysis choices, a uniform spatial null over letterbox-padded frames and the inclusion of auxiliary-class detections, each produce large spurious effects in opposite directions. Model B’s overall advantage (mAP50 0.859 against 0.755) does not concentrate near WBCs, pointing to multi-task representation learning rather than WBC suppression. Efforts to reduce FPs should target staining variability rather than WBC annotation.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/MIRASOL_018.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: https://openreview.net/forum?id=DAv0RwrdIk
BibTex
@InProceedings{AdeSam_Investigating_MICCAISAT2026,
author = { Adeniji, Samuel A. AND Obasi, Goodness C. AND Ntwali, Chris-Victor AND Iorumbur, Aondona Moses AND Raymond, Confidence AND Uwimana, Lowami AND Issah, Ahmed Tahiru},
title = { { Investigating White Blood Cells as a Source of False-Positive Malaria Parasite Detection in African Blood-Smear Images } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17264},
month = {pending},
page = {pending}
}