1
|
Baik M, Shin S, Kumar S, Seo D, Lee I, Jun HS, Kang KW, Kim BS, Nam MH, Seo S. Label-Free CD34+ Cell Identification Using Deep Learning and Lens-Free Shadow Imaging Technology. BIOSENSORS 2023; 13:993. [PMID: 38131753 PMCID: PMC10741567 DOI: 10.3390/bios13120993] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Subscribe] [Scholar Register] [Received: 09/30/2023] [Revised: 11/16/2023] [Accepted: 11/17/2023] [Indexed: 12/23/2023]
Abstract
Accurate and efficient classification and quantification of CD34+ cells are essential for the diagnosis and monitoring of leukemia. Current methods, such as flow cytometry, are complex, time-consuming, and require specialized expertise and equipment. This study proposes a novel approach for the label-free identification of CD34+ cells using a deep learning model and lens-free shadow imaging technology (LSIT). LSIT is a portable and user-friendly technique that eliminates the need for cell staining, enhances accessibility to nonexperts, and reduces the risk of sample degradation. The study involved three phases: sample preparation, dataset generation, and data analysis. Bone marrow and peripheral blood samples were collected from leukemia patients, and mononuclear cells were isolated using Ficoll density gradient centrifugation. The samples were then injected into a cell chip and analyzed using a proprietary LSIT-based device (Cellytics). A robust dataset was generated, and a custom AlexNet deep learning model was meticulously trained to distinguish CD34+ from non-CD34+ cells using the dataset. The model achieved a high accuracy in identifying CD34+ cells from 1929 bone marrow cell images, with training and validation accuracies of 97.3% and 96.2%, respectively. The customized AlexNet model outperformed the Vgg16 and ResNet50 models. It also demonstrated a strong correlation with the standard fluorescence-activated cell sorting (FACS) technique for quantifying CD34+ cells across 13 patient samples, yielding a coefficient of determination of 0.81. Bland-Altman analysis confirmed the model's reliability, with a mean bias of -2.29 and 95% limits of agreement between 18.49 and -23.07. This deep-learning-powered LSIT offers a groundbreaking approach to detecting CD34+ cells without the need for cell staining, facilitating rapid CD34+ cell classification, even by individuals without prior expertise.
Collapse
Affiliation(s)
- Minyoung Baik
- Department of Electronics and Information Engineering, Korea University, Sejong 30019, Republic of Korea; (M.B.); (S.S.); (S.K.)
| | - Sanghoon Shin
- Department of Electronics and Information Engineering, Korea University, Sejong 30019, Republic of Korea; (M.B.); (S.S.); (S.K.)
| | - Samir Kumar
- Department of Electronics and Information Engineering, Korea University, Sejong 30019, Republic of Korea; (M.B.); (S.S.); (S.K.)
| | - Dongmin Seo
- Department of Electrical Engineering, Semyung University, Jecheon 27136, Republic of Korea;
| | - Inha Lee
- Department of Biotechnology and Bioinformatics, Korea University, Sejong 30019, Republic of Korea; (I.L.); (H.S.J.)
| | - Hyun Sik Jun
- Department of Biotechnology and Bioinformatics, Korea University, Sejong 30019, Republic of Korea; (I.L.); (H.S.J.)
| | - Ka-Won Kang
- Department of Hematology, Anam Hospital, Korea University College of Medicine, Seoul 02841, Republic of Korea; (K.-W.K.); (B.S.K.)
| | - Byung Soo Kim
- Department of Hematology, Anam Hospital, Korea University College of Medicine, Seoul 02841, Republic of Korea; (K.-W.K.); (B.S.K.)
| | - Myung-Hyun Nam
- Department of Laboratory Medicine, Anam Hospital, Korea University College of Medicine, Seoul 02841, Republic of Korea
| | - Sungkyu Seo
- Department of Electronics and Information Engineering, Korea University, Sejong 30019, Republic of Korea; (M.B.); (S.S.); (S.K.)
| |
Collapse
|
2
|
Napolitano M, Gerardi C, Di Lucia A, Accardo PA, Rizzuto L, Ferraro M, Siragusa S, Buscemi F. Hematopoietic peripheral circulating blood stem cells as an independent marker of good transfusion management in patients with β-thalassemia: results from a preliminary study. Transfusion 2016; 56:827-30. [PMID: 26801519 DOI: 10.1111/trf.13452] [Citation(s) in RCA: 2] [Impact Index Per Article: 0.3] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 08/26/2015] [Revised: 11/08/2015] [Accepted: 11/13/2015] [Indexed: 12/14/2022]
Abstract
BACKGROUND Beyond hemoglobin (Hb) levels and performance status, further surrogate markers of appropriate transfusion management should improve the quality of thalassemia care. We investigated the levels of peripheral circulating CD34+ stem cells as an independent marker of appropriate hematopoietic balance in patients with thalassemia. STUDY DESIGN AND METHODS Peripheral circulating CD34+ stem cells, colony-forming unitgranulocyte, erythrocyte, macrophage, magakaryocyte (CF-GEMM), colony-forming unitgranulocyte/macrophage (CFU-GM), and erythroidburst-forming units (BFU-E) were assayed, according to standard procedures. Patients with thalassemia major (TM) and thalassemia intermedia (TI) were tested and compared to healthy controls. Demographic and clinical data were recorded. RESULTS Overall, 56 patients with TM (median age, 35 years; range, 13-52 years) and 13 with TI (median age, 44 years; range, 27-67 years) were evaluated. Annual red blood cell (RBC) transfusion requirements ranged from 10 to 65 units in all patients except four nontransfused cases. A significant increase in peripheral circulating stem cells was observed in patients, in comparison with healthy controls. Nontransfused patients showed the mean highest levels of stem cells (CD34, 32.5 ± 14.8/μL; BFU-E, 41.3 ± 22.8/mL; CFU-GM, 19.6 ± 5.6/mL; CFU-GEMM, 9.0 ± 6.1/mL). CD34+ cell count was 6.9 ± 4.5/μL in TM (p = 0.014) and 11.8 ± 14.8/μL (p = 0.051) in TI. Furthermore, only in patients with TI was a significant increase in CFU-GEMM (3.0 ± 4.8 vs. 0.75 ± 2.05/mL, p = 0.0001) observed. At multivariate analysis, peripheral circulating CD34+ stem cells did not correlate with age, sex, smoking habit, number of RBCs units transfused, Hb levels, iron chelation therapy, history of splenectomy, and hypothyroidism. CONCLUSION Circulating peripheral CD34 + stem cells are increased in β-thalassemia, in particular in nontransfused patients, compared to healthy controls.
Collapse
Affiliation(s)
| | - Calogera Gerardi
- Banca del Sangue da Cordone Ombelicale, UOC Medicina Trasfusionale, PO "Giovanni Paolo II,", Sciacca, Italy
| | - Anna Di Lucia
- Banca del Sangue da Cordone Ombelicale, UOC Medicina Trasfusionale, PO "Giovanni Paolo II,", Sciacca, Italy
| | | | - Luigi Rizzuto
- UOS Talassemia, UOC Medicina Trasfusionale, PO "Giovanni Paolo II,", Sciacca, Italy
| | - Maria Ferraro
- UOS Talassemia, UOC Medicina Trasfusionale, PO "Giovanni Paolo II,", Sciacca, Italy
| | - Sergio Siragusa
- UOC Ematologia con Trapianto, Università di Palermo, Palermo, Italy; and
| | - Filippo Buscemi
- UOS Talassemia, UOC Medicina Trasfusionale, PO "Giovanni Paolo II,", Sciacca, Italy
| |
Collapse
|
3
|
Rosenbaum ER, O'Connell B, Cottler-Fox M. Validation of a formula for predicting daily CD34(+) cell collection by leukapheresis. Cytotherapy 2012; 14:461-6. [PMID: 22277012 DOI: 10.3109/14653249.2011.652733] [Citation(s) in RCA: 40] [Impact Index Per Article: 3.3] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/13/2022]
Abstract
BACKGROUND AIMS The ability to predict how many CD34(+) cells a donor will collect on a given day is vital for efficient leukapheresis. METHODS We validated a formula to predict daily CD34(+) cell collections by leukapheresis, calculated as follows: (peripheral blood CD34(+) cells/L) × (adjusted collection efficiency of 30%)/body weight (kg), multiplied by the number of liters processed. This validation was performed from 234 donors undergoing 30 L large volume leukapheresis (LVL) and 162 donors undergoing smaller collections (non-LVL). The LVL group consisted of 811 collection events (625 multiple myeloma, 186 non-myeloma). The non-LVL group consisted of 224 collection events (196 multiple myeloma, 28 non-myeloma). All predicted and observed CD34(+) cell collection numbers were plotted (predicted versus observed) and assessed using linear regression analyses. Linear correlation coefficients (r-values), slopes and intercepts of the regression lines were evaluated. RESULTS Predicted versus observed data points across all quantities of CD34(+) cells/kg collected by both LVL and non-LVL had strong r-values of 0.947 and 0.913, respectively, demonstrating near perfect positive linear correlations. Data for LVL collections subgrouped by number of cells collected (poor, intermediate and good), mobilization regimen, collection day and diagnosis were analyzed the same way and showed consistent findings. CONCLUSIONS We have validated a formula with a strong ability to predict collection of CD34(+) cells/kg that would allow for individualization of collection for any donor once the peripheral blood CD34(+) cell count and optimal goal of collection were known; to date this has not been published by other groups.
Collapse
Affiliation(s)
- Eric R Rosenbaum
- Department of Pathology, University of Arkansas for Medical Sciences, Little Rock, Arkansas 72205, USA
| | | | | |
Collapse
|
4
|
Piedras-Ross J, León-Rodríguez E, Sánchez-Guerrero S, López-Karpovitch X. Comparison of single- and dual-platform approaches to enumerate CD34(+) cells in bone marrow and mobilized peripheral blood stem cells. Arch Med Res 2003; 34:16-9. [PMID: 12604369 DOI: 10.1016/s0188-4409(02)00451-4] [Citation(s) in RCA: 8] [Impact Index Per Article: 0.4] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 12/13/2022]
Abstract
BACKGROUND Different flow cytometric methods have been developed to derive absolute CD34(+) cells in predicting transplant outcome. Two techniques for preparing cells for quantification of CD34(+) cells were compared. METHODS Enumeration of CD34(+) cells in 16 samples of bone marrow (BM) and in 29 samples of mobilized peripheral blood stem cells (PBSC) obtained by leukapheresis was assessed simultaneously by single-platform (ProCOUNT kit) and dual-platform (Milan protocol) approaches within the first 3 h of collection. RESULTS Absolute number of CD34(+) cells obtained in PBSC and BM using single- and dual-platform methods showed high determination coefficients as follows: for PBSC, slope = 1.0515 +/- 0.048, y-intercept = 88.638 +/- 52.45, and r(2) = 0.941, and for BM, slope = 1.0203 +/- 0.093, y-intercept = 122.25 +/- 20.65, and r(2) = 0.878. There were no statistically significant differences in absolute number of CD34(+) cells from PBSC between single-platform (mean 575/microL, range 70-3683/microL) and dual-platform (786/microL, range 51-3804/microL) assays. In contrast, absolute number of CD34(+) cells from BM was significantly lower (p = 0.0002) when enumerated by ProCount kit (135/microL, 14-758/microL) than with dual-platform method (260/microL, 74-889/microL). CONCLUSIONS Both approaches can be used indistinctly to estimate absolute number of CD34(+) cells in PBSC but not in BM.
Collapse
Affiliation(s)
- Josefa Piedras-Ross
- Departamento de Hematología y Oncología, Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán, Mexico City, Mexico.
| | | | | | | |
Collapse
|
5
|
Potential Role for Hyaluronan and the Hyaluronan Receptor RHAMM in Mobilization and Trafficking of Hematopoietic Progenitor Cells. Blood 1999. [DOI: 10.1182/blood.v93.9.2918] [Citation(s) in RCA: 68] [Impact Index Per Article: 2.7] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/20/2022] Open
Abstract
Abstract
Although the mechanism(s) underlying mobilization of hematopoietic progenitor cells (HPCs) is unknown, detachment from the bone marrow (BM) microenvironment and motility are likely to play a role. This work analyzes the motile behavior of HPCs and the receptors involved. CD34+45lo/medScatterlo/med HPCs from granulocyte colony-stimulating factor (G-CSF)–mobilized blood and mobilized BM were compared with steady-state BM for their ability to bind hyaluronan (HA), their expression of the HA receptors RHAMM and CD44, and their motogenic behavior. Although RHAMM and CD44 are expressed by mobilized blood HPCs, function blocking monoclonal antibodies (MoAbs) identified RHAMM as a major HA binding receptor, with a less consistent participation by CD44. Permeabilization of mobilized blood HPCs showed a pool of intracellular (ic) RHAMM and a smaller pool of icCD44. In contrast, steady-state BM HPCs have significantly larger pools of icRHAMM and icCD44. Also, in contrast to mobilized blood HPCs, for steady-state BM HPCs, MoAbs to RHAMM and CD44 act as agonists to upregulate HA binding. The comparison between mobilized and steady-state BM HPCs suggests that G-CSF mobilization is associated with depletion of intracellular stores of HA receptors and modulates HA receptor usage. To confirm that mobilization alters the HA receptor distribution and usage by HPCs, samples of BM were collected at the peak of G-CSF mobilization in parallel with mobilized blood samples. HA receptor distribution of mobilized BM HPCs was closely matched with mobilized blood HPCs and different from steady-state BM HPCs. Mobilized BM HPCs had lower pools of icHA receptors, similar to those of mobilized blood HPCs. Treatment of mobilized BM HPCs with anti-RHAMM MoAb decreased HA binding, in contrast to steady-state BM HPCs. Thus, G-CSF mobilization may stimulate an autocrine stimulatory loop for HPCs in which HA interacts with basal levels of RHAMM and/or CD44 to stimulate receptor recycling. Consistent with this, treatment of HPCs with azide, nystatin, or cytochalasin B increased HA binding, implicating an energy-dependent process involving lipid rafts and the cytoskeleton. Of the sorted HPCs, 66% were adherent and 27% were motile on fibronectin plus HA. HPC adherence was inhibited by MoAbs to β1 integrin and CD44, but not to RHAMM, whereas HPC motility was inhibited by MoAb to RHAMM and β1 integrin, but not to CD44. This finding suggests that RHAMM and CD44 play reciprocal roles in adhesion and motility by HPCs. The G-CSF–associated alterations in RHAMM distribution and the RHAMM-dependent motility of HPCs suggest a potential role for HA and RHAMM in trafficking of HPCs and the possible use of HA as a mobilizing agent in vivo.
Collapse
|
6
|
Potential Role for Hyaluronan and the Hyaluronan Receptor RHAMM in Mobilization and Trafficking of Hematopoietic Progenitor Cells. Blood 1999. [DOI: 10.1182/blood.v93.9.2918.409k19_2918_2927] [Citation(s) in RCA: 4] [Impact Index Per Article: 0.2] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/20/2022] Open
Abstract
Although the mechanism(s) underlying mobilization of hematopoietic progenitor cells (HPCs) is unknown, detachment from the bone marrow (BM) microenvironment and motility are likely to play a role. This work analyzes the motile behavior of HPCs and the receptors involved. CD34+45lo/medScatterlo/med HPCs from granulocyte colony-stimulating factor (G-CSF)–mobilized blood and mobilized BM were compared with steady-state BM for their ability to bind hyaluronan (HA), their expression of the HA receptors RHAMM and CD44, and their motogenic behavior. Although RHAMM and CD44 are expressed by mobilized blood HPCs, function blocking monoclonal antibodies (MoAbs) identified RHAMM as a major HA binding receptor, with a less consistent participation by CD44. Permeabilization of mobilized blood HPCs showed a pool of intracellular (ic) RHAMM and a smaller pool of icCD44. In contrast, steady-state BM HPCs have significantly larger pools of icRHAMM and icCD44. Also, in contrast to mobilized blood HPCs, for steady-state BM HPCs, MoAbs to RHAMM and CD44 act as agonists to upregulate HA binding. The comparison between mobilized and steady-state BM HPCs suggests that G-CSF mobilization is associated with depletion of intracellular stores of HA receptors and modulates HA receptor usage. To confirm that mobilization alters the HA receptor distribution and usage by HPCs, samples of BM were collected at the peak of G-CSF mobilization in parallel with mobilized blood samples. HA receptor distribution of mobilized BM HPCs was closely matched with mobilized blood HPCs and different from steady-state BM HPCs. Mobilized BM HPCs had lower pools of icHA receptors, similar to those of mobilized blood HPCs. Treatment of mobilized BM HPCs with anti-RHAMM MoAb decreased HA binding, in contrast to steady-state BM HPCs. Thus, G-CSF mobilization may stimulate an autocrine stimulatory loop for HPCs in which HA interacts with basal levels of RHAMM and/or CD44 to stimulate receptor recycling. Consistent with this, treatment of HPCs with azide, nystatin, or cytochalasin B increased HA binding, implicating an energy-dependent process involving lipid rafts and the cytoskeleton. Of the sorted HPCs, 66% were adherent and 27% were motile on fibronectin plus HA. HPC adherence was inhibited by MoAbs to β1 integrin and CD44, but not to RHAMM, whereas HPC motility was inhibited by MoAb to RHAMM and β1 integrin, but not to CD44. This finding suggests that RHAMM and CD44 play reciprocal roles in adhesion and motility by HPCs. The G-CSF–associated alterations in RHAMM distribution and the RHAMM-dependent motility of HPCs suggest a potential role for HA and RHAMM in trafficking of HPCs and the possible use of HA as a mobilizing agent in vivo.
Collapse
|
7
|
Barnett D, Granger V, Storie I, Peel J, Pollitt R, Smart T, Reilly JT. Quality assessment of CD34+ stem cell enumeration: experience of the United Kingdom National External Quality Assessment Scheme (UK NEQAS) using a unique stable whole blood preparation. Br J Haematol 1998; 102:553-65. [PMID: 9695973 DOI: 10.1046/j.1365-2141.1998.00784.x] [Citation(s) in RCA: 41] [Impact Index Per Article: 1.6] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/20/2022]
Abstract
CD34+ peripheral blood stem cell (PBSC) mobilization and harvesting has rapidly replaced autologous bone marrow as a source of stem cells for transplantation. Timing and adequacy of harvests rely upon the accurate enumeration of circulating CD34+ cells. However, previous EQA programmes have reported interlaboratory CVs as high as 284%, suggesting the need for greater standardization. In addition the routine use of fresh and/or frozen cells as analytes also introduces antigen instability as a variable factor. To circumvent this problem and achieve a true reflection of interlaboratory variation, we have used a novel whole blood preparation in which the antigenic profiles of PBSCs, as determined by flow cytometry, are retained for > 200 d. This international scheme, currently the largest in the world, distributes aliquots of stabilized whole blood bi-monthly to 91 laboratories in 20 countries (44 U.K., 47 overseas). Participants are required to determine the percentage and absolute values for CD34+ PBSCs using in-house techniques. Adopting such a preparation, a more accurate determination of interlaboratory variation has been possible when compared to previous EQA studies, with CVs as low as 22% and 24% for percentage and absolute counts. In addition the programme has established that a wide range of methods are in routine use, emphasizing the urgent requirement for national/international consensus guidelines.
Collapse
Affiliation(s)
- D Barnett
- UK NEQAS for Leucocyte Immunophenotyping, Department of Haematology, Royal Hallamshire Hospital, Sheffield
| | | | | | | | | | | | | |
Collapse
|
8
|
Marti G, Johnsen H, Sutherland R, Serke S. A convergence of methods for a worldwide standard for CD34+ cell enumeration. JOURNAL OF HEMATOTHERAPY 1998; 7:105-9. [PMID: 9597567 DOI: 10.1089/scd.1.1998.7.105] [Citation(s) in RCA: 16] [Impact Index Per Article: 0.6] [Reference Citation Analysis] [MESH Headings] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 11/13/2022]
|