{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"institution":[{"name":"bioRxiv"}],"indexed":{"date-parts":[[2026,4,8]],"date-time":"2026-04-08T13:00:17Z","timestamp":1775653217506,"version":"3.50.1"},"posted":{"date-parts":[[2026,4,6]]},"group-title":"Bioengineering","reference-count":62,"publisher":"openRxiv","license":[{"start":{"date-parts":[[2026,4,6]],"date-time":"2026-04-06T00:00:00Z","timestamp":1775433600000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/100000050","name":"nhlbi","doi-asserted-by":"crossref","award":["R01HL154150,"],"award-info":[{"award-number":["R01HL154150,"]}],"id":[{"id":"10.13039\/100000050","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"accepted":{"date-parts":[[2026,4,6]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                <jats:p>Microscopy-based analysis of red blood cell (RBC) morphology is widely used to study phenotypes in sickle cell disease (SCD). Although AI models have been developed to automate classification, most are trained on pre-cropped single-cell images and thus struggle with full-scope microscopic images containing densely packed cells and diverse morphologies, which require both accurate detection and fine-grained classification. We propose an end-to-end computational framework to identify individual RBCs in full-scope microscopy images and classify them into five morphological categories: discocytes (DO), echinocytes (E), elongated and sickle-shaped cells (ES), granular cells (G), and reticulocytes (R). We first evaluate advanced detection-classification models, including You Only Look Once (YOLO) and Detection Transformers (DETR), and demonstrate that while these models effectively detect cells, their classification performance falls short of specialized classifiers trained on single-cell images, particularly for minority phenotypes. To address this limitation, we introduce a two-step framework in which a YOLO-based detector localizes and crops individual cells from full-scope images, followed by a fine-tuned DenseNet121 ensemble classifier that assigns each cell to one of the five morphological categories. The proposed framework achieves a detection-level F1-score of 0.9661 and a weighted-average classification F1-score of 0.9708, with an overall classification accuracy of 97.06%. Compared with the single-step YOLO26n baseline, the two-step pipeline yields a macro-average F1-score improvement of +0.1675, with particularly substantial gains for minority classes (E: +0.1623; G: +0.2774; R: +0.2603). Overall, this hybrid framework demonstrates a practical strategy for adapting fast, general-purpose detection models to domain-specific biomedical tasks by combining them with specialized classifiers, delivering both efficiency and high accuracy for scientific and clinical image analysis.<\/jats:p>","DOI":"10.64898\/2026.03.31.715578","type":"posted-content","created":{"date-parts":[[2026,4,6]],"date-time":"2026-04-06T10:25:15Z","timestamp":1775471115000},"source":"Crossref","is-referenced-by-count":0,"title":["Decoupling Detection and Classification to Improve Morphological Phenotype Analysis of Sickle Red Blood Cells in Full-Scope Microscopy"],"prefix":"10.64898","author":[{"given":"Suqiang","family":"Ma","sequence":"first","affiliation":[{"name":"School of Electrical and Computer Engineering, University of Georgia, Athens, GA 30605"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Mengjia","family":"Xu","sequence":"additional","affiliation":[{"name":"Department of Data Science, Ying Wu College of Computing, New Jersey Institute of Technology, Newark, NJ 07102"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Ming","family":"Dao","sequence":"additional","affiliation":[{"name":"Department of Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, MA, 02139"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"He","family":"Li","sequence":"additional","affiliation":[{"name":"School of Chemical, Material and Biomedical Engineering, University of Georgia, Athens, GA 30605"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"54368","reference":[{"key":"2026040804551101000_2026.03.31.715578v1.1","first-page":"29","article-title":"On-chip biomedical imaging","volume":"6","year":"2012","journal-title":"IEEE reviews in biomedical engineering"},{"key":"2026040804551101000_2026.03.31.715578v1.2","doi-asserted-by":"crossref","unstructured":"Reiner Salzer . Biomedical imaging: principles and applications. John Wiley & Sons, 2012.","DOI":"10.1002\/9781118271933"},{"key":"2026040804551101000_2026.03.31.715578v1.3","doi-asserted-by":"publisher","DOI":"10.1007\/S11095-019-2703-7\/FIGURES\/8"},{"key":"2026040804551101000_2026.03.31.715578v1.4","unstructured":"Andrew Webb . Introduction to biomedical imaging. John Wiley & Sons, 2022."},{"key":"2026040804551101000_2026.03.31.715578v1.5","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.1508524112"},{"key":"2026040804551101000_2026.03.31.715578v1.6","doi-asserted-by":"publisher","DOI":"10.1001\/jama.285.5.562"},{"issue":"12","key":"2026040804551101000_2026.03.31.715578v1.7","doi-asserted-by":"crossref","first-page":"910","DOI":"10.1038\/nmeth817","article-title":"Fluorescence microscopy","volume":"2","year":"2005","journal-title":"Nature Methods"},{"issue":"10","key":"2026040804551101000_2026.03.31.715578v1.8","doi-asserted-by":"crossref","first-page":"811","DOI":"10.1038\/nmeth.1709","article-title":"Optical sectioning microscopy with planar or structured illumination","volume":"8","year":"2011","journal-title":"Nature Methods"},{"key":"2026040804551101000_2026.03.31.715578v1.9","unstructured":"Rafael C Gonzalez and Richard E Woods . Digital Image Processing. Pearson Education, Upper Saddle River, NJ, 3rd edition, 2008."},{"key":"2026040804551101000_2026.03.31.715578v1.10","doi-asserted-by":"publisher","DOI":"10.1109\/MSP.2012.2204190"},{"issue":"10","key":"2026040804551101000_2026.03.31.715578v1.11","doi-asserted-by":"crossref","first-page":"1034","DOI":"10.3390\/bioengineering11101034","article-title":"Advances in medical image segmentation: a comprehensive review of traditional, deep learning and hybrid approaches","volume":"11","year":"2024","journal-title":"Bioengineering"},{"key":"2026040804551101000_2026.03.31.715578v1.12","doi-asserted-by":"crossref","first-page":"48","DOI":"10.1038\/s41746-022-00592-y","article-title":"Machine learning for medical imaging: methodological failures and recommendations for the future","volume":"5","year":"2022","journal-title":"npj Digital Medicine"},{"key":"2026040804551101000_2026.03.31.715578v1.13","doi-asserted-by":"crossref","first-page":"16878","DOI":"10.1038\/s41598-017-17204-5","article-title":"Qupath: Open source software for digital pathology image analysis","volume":"7","year":"2017","journal-title":"Scientific Reports"},{"key":"2026040804551101000_2026.03.31.715578v1.14","first-page":"57877","article-title":"Fast segmentation and multiplexing imaging of organelles in live cells","volume":"16","year":"2025","journal-title":"Nature Communications"},{"issue":"7","key":"2026040804551101000_2026.03.31.715578v1.15","first-page":"1520","article-title":"Deep learning on medical image analysis: a review","volume":"15","year":"2022","journal-title":"IET Image Processing"},{"key":"2026040804551101000_2026.03.31.715578v1.16","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2017.07.005"},{"key":"2026040804551101000_2026.03.31.715578v1.17","doi-asserted-by":"crossref","first-page":"2312","DOI":"10.1016\/j.csbj.2020.08.003","article-title":"A bird\u2019s-eye view of deep learning in bioimage analysis","volume":"18","year":"2020","journal-title":"Computational and Structural Biotechnology Journal"},{"key":"2026040804551101000_2026.03.31.715578v1.18","first-page":"886","volume-title":"In 2005 IEEE computer society conference on computer vision and pattern recognition (CVPR\u201905)","volume":"1","year":"2005"},{"issue":"5","key":"2026040804551101000_2026.03.31.715578v1.19","doi-asserted-by":"crossref","first-page":"10491","DOI":"10.4249\/scholarpedia.10491","article-title":"Scale invariant feature transform","volume":"7","year":"2012","journal-title":"Scholarpedia"},{"key":"2026040804551101000_2026.03.31.715578v1.20","doi-asserted-by":"crossref","unstructured":"Ross Girshick . Rich feature hierarchies for accurate object detection and semantic segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 580\u2013587, 2014.","DOI":"10.1109\/CVPR.2014.81"},{"key":"2026040804551101000_2026.03.31.715578v1.21","unstructured":"Shaoqing Ren , Kaiming He , Ross Girshick , and Jian Sun . Faster r-cnn: Towards real-time object detection with region proposal networks. In Advances in neural information processing systems, pages 91\u201399, 2015."},{"key":"2026040804551101000_2026.03.31.715578v1.22","doi-asserted-by":"crossref","unstructured":"Joseph Redmon , Santosh Divvala , Ross Girshick , and Ali Farhadi . You only look once: Unified, real-time object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 779\u2013788, 2016.","DOI":"10.1109\/CVPR.2016.91"},{"key":"2026040804551101000_2026.03.31.715578v1.23","doi-asserted-by":"crossref","unstructured":"Joseph Redmon and Ali Farhadi . YOLO9000: Better, faster, stronger. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 7263\u20137271, 2017.","DOI":"10.1109\/CVPR.2017.690"},{"key":"2026040804551101000_2026.03.31.715578v1.24","article-title":"YOLOv3: An incremental improvement","year":"2018","journal-title":"arXiv preprint"},{"key":"2026040804551101000_2026.03.31.715578v1.25","article-title":"YOLOv4: Optimal speed and accuracy of object detection","year":"2020","journal-title":"arXiv preprint"},{"key":"2026040804551101000_2026.03.31.715578v1.26","unstructured":"Glenn Jocher et al. YOLOv5. https:\/\/github.com\/ultralytics\/YOLOv5, 2020."},{"key":"2026040804551101000_2026.03.31.715578v1.27","article-title":"YOLOv7: Train-able bag-of-freebies sets new state-of-the-art for real-time object detectors","year":"2022","journal-title":"arXiv preprint"},{"key":"2026040804551101000_2026.03.31.715578v1.28","unstructured":"Ultralytics. YOLOv8. https:\/\/github.com\/ultralytics\/ultralytics, 2023."},{"key":"2026040804551101000_2026.03.31.715578v1.29","unstructured":"Glenn Jocher et al. YOLO by ultralytics. https:\/\/github.com\/ultralytics\/ultralytics, 2023. Accessed: [your access date]."},{"key":"2026040804551101000_2026.03.31.715578v1.30","article-title":"YOLO26: key ar-chitectural enhancements and performance benchmarking for real-time object detection","year":"2025","journal-title":"arXiv preprint"},{"issue":"19","key":"2026040804551101000_2026.03.31.715578v1.31","doi-asserted-by":"crossref","first-page":"6025","DOI":"10.3390\/s25196025","article-title":"Object detection with transformers: A review","volume":"25","year":"2025","journal-title":"Sensors"},{"key":"2026040804551101000_2026.03.31.715578v1.32","unstructured":"Kendall Preston . Digital processing of biomedical images. Springer Science & Business Media, 2012."},{"key":"2026040804551101000_2026.03.31.715578v1.33","doi-asserted-by":"publisher","DOI":"10.1038\/nrm761"},{"issue":"3","key":"2026040804551101000_2026.03.31.715578v1.34","doi-asserted-by":"crossref","first-page":"036602","DOI":"10.1088\/0034-4885\/78\/3\/036602","article-title":"The unforeseen challenge: from genotype-to-phenotype in cell populations","volume":"78","year":"2015","journal-title":"Reports on Progress in Physics"},{"issue":"13","key":"2026040804551101000_2026.03.31.715578v1.35","doi-asserted-by":"crossref","first-page":"133001","DOI":"10.1088\/1751-8121\/aa5db4","article-title":"Stochastic switching in biology: from genotype to phenotype","volume":"50","year":"2017","journal-title":"Journal of Physics A: Mathematical and Theoretical"},{"key":"2026040804551101000_2026.03.31.715578v1.36","first-page":"517","article-title":"Peculiar elongated and sickle-shaped red blood corpuscles in a case of severe anemia","volume":"6","year":"1910","journal-title":"Arch. Intern. Med"},{"key":"2026040804551101000_2026.03.31.715578v1.37","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.71.12.4864"},{"key":"2026040804551101000_2026.03.31.715578v1.38","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.73.9.3035"},{"key":"2026040804551101000_2026.03.31.715578v1.39","doi-asserted-by":"crossref","first-page":"752","DOI":"10.1056\/NEJM197810052991405","article-title":"Sickle-cell anemia: Molecular and cellular bases of therapeutic approaches","volume":"299","year":"1978","journal-title":"N. Engl. J. Med"},{"key":"2026040804551101000_2026.03.31.715578v1.40","doi-asserted-by":"publisher","DOI":"10.1016\/S0006-3495(80)84962-9"},{"key":"2026040804551101000_2026.03.31.715578v1.41","doi-asserted-by":"publisher","DOI":"10.1182\/blood.V58.6.1057.1057"},{"key":"2026040804551101000_2026.03.31.715578v1.42","doi-asserted-by":"publisher","DOI":"10.1056\/NEJM199904013401307"},{"key":"2026040804551101000_2026.03.31.715578v1.43","doi-asserted-by":"publisher","DOI":"10.1097\/00062752-200203000-00003"},{"key":"2026040804551101000_2026.03.31.715578v1.44","doi-asserted-by":"publisher","DOI":"10.1182\/blood-2013-05-498311"},{"key":"2026040804551101000_2026.03.31.715578v1.45","doi-asserted-by":"publisher","DOI":"10.1182\/blood-2015-12-674606"},{"key":"2026040804551101000_2026.03.31.715578v1.46","doi-asserted-by":"publisher","DOI":"10.1126\/sciadv.aax3905"},{"key":"2026040804551101000_2026.03.31.715578v1.47","doi-asserted-by":"crossref","unstructured":"Ying Qian , Nazanin Ahmadi Daryakenari , Melissa Hallow , Lu Lu , Pierre A Buffet , Ming Dao , He Li , and George Em Karniadakis . In silico post-screening of anti-polymerization agents to treat sickle cell disease. bioRxiv, pages 2025\u201310, 2025.","DOI":"10.1101\/2025.10.25.684554"},{"issue":"10","key":"2026040804551101000_2026.03.31.715578v1.48","doi-asserted-by":"crossref","first-page":"e1005746","DOI":"10.1371\/journal.pcbi.1005746","article-title":"A deep convolutional neural network for classification of red blood cells in sickle cell anemia","volume":"13","year":"2017","journal-title":"PLoS computational biology"},{"key":"2026040804551101000_2026.03.31.715578v1.49","first-page":"309","article-title":"A review of automated methods for the detection of sickle cell disease","volume":"13","year":"2019","journal-title":"IEEE reviews in biomedical engineering"},{"issue":"3","key":"2026040804551101000_2026.03.31.715578v1.50","doi-asserted-by":"crossref","first-page":"427","DOI":"10.3390\/electronics9030427","article-title":"Deep learning models for classification of red blood cells in microscopy images to aid in sickle cell anemia diagnosis","volume":"9","year":"2020","journal-title":"Electronics"},{"key":"2026040804551101000_2026.03.31.715578v1.51","doi-asserted-by":"crossref","unstructured":"Laith Alzubaidi , Omran Al-Shamma , Mohammed A Fadhel , Laith Farhan , and Jinglan Zhang . Classification of red blood cells in sickle cell anemia using deep convolutional neural network. In International Conference on Intelligent Systems Design and Applications, pages 550\u2013559. Springer, 2018.","DOI":"10.1007\/978-3-030-16657-1_51"},{"issue":"2","key":"2026040804551101000_2026.03.31.715578v1.52","doi-asserted-by":"crossref","first-page":"221","DOI":"10.11591\/ijai.v9.i2.pp221-228","article-title":"A deep learning alexnet model for classification of red blood cells in sickle cell anemia","volume":"9","year":"2020","journal-title":"Int J Artif Intell"},{"issue":"11","key":"2026040804551101000_2026.03.31.715578v1.53","doi-asserted-by":"crossref","first-page":"3095","DOI":"10.1109\/JBHI.2020.3000484","article-title":"Automated semantic segmentation of red blood cells for sickle cell disease","volume":"24","year":"2020","journal-title":"IEEE journal of biomedical and health informatics"},{"key":"2026040804551101000_2026.03.31.715578v1.54","article-title":"Roboflow 100: A rich, multi-domain object detection benchmark","year":"2022","journal-title":"arXiv preprint"},{"key":"2026040804551101000_2026.03.31.715578v1.55","unstructured":"Tsung-Yi Lin , Priya Goyal , Ross Girshick , Kaiming He , and Piotr Doll\u00e1r . Focal loss for dense object detection. In Proceedings of the IEEE International Conference on Computer Vision (ICCV), pages 2980\u20132988, 2017."},{"key":"2026040804551101000_2026.03.31.715578v1.56","doi-asserted-by":"crossref","unstructured":"Golnaz Ghiasi , Yin Cui , Aravind Srinivas , Rui Qian , Tsung-Yi Lin , Ekin D Cubuk , Quoc V Le , and Barret Zoph . Simple copy-paste is a strong data augmentation method for instance segmentation. In Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 2918\u20132928, 2021.","DOI":"10.1109\/CVPR46437.2021.00294"},{"key":"2026040804551101000_2026.03.31.715578v1.57","doi-asserted-by":"crossref","unstructured":"Yin Cui , Menglin Jia , Tsung-Yi Lin , Yang Song , and Serge Belongie . Class-balanced loss based on effective number of samples. In Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 9268\u20139277, 2019.","DOI":"10.1109\/CVPR.2019.00949"},{"key":"2026040804551101000_2026.03.31.715578v1.58","doi-asserted-by":"crossref","unstructured":"Kaiming He , Xiangyu Zhang , Shaoqing Ren , and Jian Sun . Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 770\u2013778, 2016.","DOI":"10.1109\/CVPR.2016.90"},{"key":"2026040804551101000_2026.03.31.715578v1.59","doi-asserted-by":"crossref","unstructured":"Gao Huang , Zhuang Liu , Laurens Van Der Maaten , and Kilian Q Weinberger . Densely con-nected convolutional networks. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 4700\u20134708, 2017.","DOI":"10.1109\/CVPR.2017.243"},{"key":"2026040804551101000_2026.03.31.715578v1.60","unstructured":"Mingxing Tan and Quoc V. Le . Efficientnet: Rethinking model scaling for convolutional neural networks. In Proceedings of the 36th International Conference on Machine Learning (ICML), 2019."},{"key":"2026040804551101000_2026.03.31.715578v1.61","unstructured":"Alexey Dosovitskiy , Lucas Beyer , Alexander Kolesnikov , et al. An image is worth 16\u00d716 words: Transformers for image recognition at scale. In International Conference on Learning Representations (ICLR), 2021."},{"key":"2026040804551101000_2026.03.31.715578v1.62","first-page":"233","volume-title":"In Medical imaging 2019: imaging informatics for healthcare, research, and applications","volume":"10954","year":"2019"}],"container-title":[],"original-title":[],"link":[{"URL":"https:\/\/syndication.highwire.org\/content\/doi\/10.64898\/2026.03.31.715578","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,4,8]],"date-time":"2026-04-08T11:55:45Z","timestamp":1775649345000},"score":1,"resource":{"primary":{"URL":"http:\/\/biorxiv.org\/lookup\/doi\/10.64898\/2026.03.31.715578"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,4,6]]},"references-count":62,"URL":"https:\/\/doi.org\/10.64898\/2026.03.31.715578","relation":{},"subject":[],"published":{"date-parts":[[2026,4,6]]},"subtype":"preprint"}}