biostudies-literatureSeesawad NNSRF via the Program Management Unit for Human Resources &amp;amp; Institutional Development, Research and InnovationNew Discovery and Frontier Research514-523https://www.ebi.ac.uk/biostudies/studies/S-EPMC11268940biostudies-literatureUnknown5<i>Background:</i> Deep learning models for patch classification in whole-slide images (WSIs) have shown promise in assisting follicular lymphoma grading. However, these models often require pathologists to identify centroblasts and manually provide refined labels for model optimization. <i>Objective:</i> To address this limitation, we propose <i>PseudoCell</i>, an object detection framework for automated centroblast detection in WSI, eliminating the need for extensive pathologist's refined labels. <i>Methods:</i> <i>PseudoCell</i> leverages a combination of pathologist-provided centroblast labels and pseudo-negative labels generated from undersampled false-positive predictions based on cell morphology features. This approach reduces the reliance on time-consuming manual annotations. <i>ReIEEE open journal of engineering in medicine and biologyPseudoCell: Hard Negative Mining as Pseudo Labeling for Deep Learning-Based Centroblast Cell Detection.PMC11268940B38G670007R016420005Hnoohom NSudhawiyangkul TThuwajit CIttichaiwong PBoonsakan PVeerakanjana KSeesawad NCharngkaew KWilaiprasitporn TSripodok SPongpaibul AThuwajit PAngkathunyakul NYuenyong SSawangjai PfalsePseudoCell: Hard Negative Mining as Pseudo Labeling for Deep Learning-Based Centroblast Cell Detection.<i>Background:</i> Deep learning models for patch classification in whole-slide images (WSIs) have shown promise in assisting follicular lymphoma grading. However, these models often require pathologists to identify centroblasts and manually provide refined labels for model optimization. <i>Objective:</i> To address this limitation, we propose <i>PseudoCell</i>, an object detection framework for automated centroblast detection in WSI, eliminating the need for extensive pathologist's refined labels. <i>Methods:</i> <i>PseudoCell</i> leverages a combination of pathologist-provided centroblast labels and pseudo-negative labels generated from undersampled false-positive predictions based on cell morphology features. This approach reduces the reliance on time-consuming manual annotations. <i>Re2024-01-01T00:00:00Z20242026-04-08T18:48:33.434Z2025-04-05T13:57:21.913ZS-EPMC112689403905097110.1109/OJEMB.2024.3407351