Automatic classification of 10 blood cell subtypes using transfer learning via pre-trained convolutional neural networks

Informatics in Medicine Unlocked, Год журнала: 2024, Номер 49, С. 101542 - 101542

Опубликована: Янв. 1, 2024

Human blood is primarily composed of plasma, red cells, white and platelets. It plays a vital role in transporting oxygen nutrients to all organs, stores essential health-related data about the human body. Blood cells are utilized defend body against infections disease. Hence, analysis permits physicians assess an individual's physiological condition. sub-classified into eight groups: Neutrophils, eosinophils, basophils, lymphocytes, monocytes, immature granulocytes (promyelocytes, myelocytes, metamyelocytes), erythroblasts, platelets (thrombocytes) on basis their nucleus, shape cytoplasm. Traditionally, pathologists hematologists have identified examined these via microscopy prior manual classification, with this approach being slow prone error. Therefore, it automate process. In current study, transfer learning series pre-trained Convolutional Neural Network (CNN) models—VGG16, VGG19, ResNet-50, ResNet-101, ResNet-152, InceptionV3, MobileNetV2 DenseNet-201 was applied normal peripheral dataset (PBC). The overall accuracy achieved individual CNNs ranged from 91.4 % 94.7 %. Based architectures, CNN-based architecture has been developed automatically classify ten cell types. proposed CNN model tested images PBC, Kaggle LISC datasets. Achieved 99.91 %, 99.68 98.79 respectively, across three presented outperforms previous results reported scientific/medical literature high capacity for framework generalization future applications classification.

Язык: Английский

---

Two-stage CNN-based framework for leukocytes classification

Computers in Biology and Medicine, Год журнала: 2025, Номер 187, С. 109616 - 109616

Опубликована: Фев. 5, 2025

Язык: Английский

---

Enhanced White Blood Cell and Platelet Segmentation: A Particle Swarm Optimization-based Chromaticity approach

Pertanika journal of science & technology, Год журнала: 2025, Номер 33(3)

Опубликована: Апрель 22, 2025

Microscopic image examination is essential for medical diagnostics to identify anomalies using cell counts based on morphology. Sickle Cell Disease (SCD) an inherited blood condition characterized by defective hemoglobin, leading severe anemia and complications. Detecting sickle cells in smears essential, but the presence of White (WBCs) platelets often leads miscounting as they are classified incorrectly red (RBCs). This study proposed approach segmenting WBCs resembling human color recognition process differentiate regions accurate identification. First, RGB space converted RG chromaticity locate with high pixel chromatic variance. Parametric segmentation applied images appropriate channel probability distribution values. The optimal threshold values have been determined Particle Swarm Optimization (PSO) dynamically narrowing search obtained through manual experimentation ranging from 0.001 1. systematic effectively identifies segments WBCs, ensuring that overlapping accurately segmented. Compared state-of-the-art techniques, achieved accuracy 96.32 %, 96.97% sensitivity, 96.96 % precision 97.46% F- score pixel-wise platelets.

Язык: Английский

---

Advanced leukocyte classification using attention mechanisms and dual channel U-Net architecture

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Апрель 22, 2025

Язык: Английский

---