WGAN Latent Space Embeddings for Blast Identification in Childhood Acute Myeloid Leukaemia

Licandro, Roxane; Schlegl, Thomas; Reiter, Michael J.; Dworzak, Michael; Schumich, Angela; Kampel, Martin

doi:10.1109/icpr.2018.8546177

Cited by 11 publications

(7 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [ 38 , 39 ] recent developments in Convolutional Neural Networks (CNNs) are applied to imaging FCM applications. For non-imaging FCM data initial works processed the samples cell-wise [ 40 , 41 , 42 ]. CellCNN [ 43 ] uses a 1D-convolution layer to project the measurements of each cell to an embedding space then applies a pooling layer to aggregate information in order to learn the associated phenotype from multi-cell input.…”

Section: Appendix A1 Fcm Analysis With Statistical Methodsmentioning

confidence: 99%

“…CellCNN [ 43 ] uses a 1D-convolution layer to project the measurements of each cell to an embedding space then applies a pooling layer to aggregate information in order to learn the associated phenotype from multi-cell input. In [ 40 , 41 ], the problem is cast as a binary classification problem. In [ 42 ], Li et al check the similarity of a given sample with a set of reference samples and then train a four layer network on the best match.…”

Section: Appendix A1 Fcm Analysis With Statistical Methodsmentioning

confidence: 99%

See 1 more Smart Citation

UMAP Based Anomaly Detection for Minimal Residual Disease Quantification within Acute Myeloid Leukemia

Weijler

Kowarsch

Wödlinger

et al. 2022

Cancers

Self Cite

View full text Add to dashboard Cite

Leukemia is the most frequent malignancy in children and adolescents, with acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML) as the most common subtypes. Minimal residual disease (MRD) measured by flow cytometry (FCM) has proven to be a strong prognostic factor in ALL as well as in AML. Machine learning techniques have been emerging in the field of automated MRD quantification with the objective of superseding subjective and time-consuming manual analysis of FCM-MRD data. In contrast to ALL, where supervised multi-class classification methods have been successfully deployed for MRD detection, AML poses new challenges: AML is rarer (with fewer available training data) than ALL and much more heterogeneous in its immunophenotypic appearance, where one-class classification (anomaly detection) methods seem more suitable. In this work, a new semi-supervised approach based on the UMAP algorithm for MRD detection utilizing only labels of blast free FCM samples is presented. The method is tested on a newly gathered set of AML FCM samples and results are compared to state-of-the-art methods. We reach a median F1-score of 0.794, while providing a transparent classification pipeline with explainable results that facilitates inter-disciplinary work between medical and technical experts. This work shows that despite several issues yet to overcome, the merits of automated MRD quantification can be fully exploited also in AML.

show abstract

Section: Appendix A1 Fcm Analysis With Statistical Methodsmentioning

confidence: 99%

Section: Appendix A1 Fcm Analysis With Statistical Methodsmentioning

confidence: 99%

UMAP Based Anomaly Detection for Minimal Residual Disease Quantification within Acute Myeloid Leukemia

Weijler

Kowarsch

Wödlinger

et al. 2022

Cancers

Self Cite

View full text Add to dashboard Cite

show abstract

“…Apart from imaging FCM applications [18,8], few examples of successful application of deep neural networks to FCM data exist. In [15,23,14] neural networks based on fully connected layers are presented that work on single events. These methods can only learn fixed decision boundaries to separate biologically meaningful sub-populations.…”

Section: Related Workmentioning

confidence: 99%

“…convolutional neural networks for a 2d grid or recurrent neural networks for sequences). Some methods [23,15] circumvent this problem by applying neural networks on single cells instead of samples, however, these approaches can only learn static decision boundaries and are not able to capture global sample information. In this work, we present a novel method for the detection of blast cells and MRD quantification that is capable of capturing long-range information in the full data space by attending to all events in a sample at once.…”

Section: Introductionmentioning

confidence: 99%

Automated Identification of Cell Populations in Flow Cytometry Data with Transformers

Wödlinger,

Reiter,

Weijler

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Acute Lymphoblastic Leukemia (ALL) is the most frequent hematologic malignancy in children and adolescents. A strong prognostic factor in ALL is given by the Minimal Residual Disease (MRD), which is a measure for the number of leukemic cells persistent in a patient. Manual MRD assessment from Multiparameter Flow Cytometry (FCM) data after treatment is time-consuming and subjective. In this work, we present an automated method to compute the MRD value directly from FCM data. We present a novel neural network approach based on the transformer architecture that learns to directly identify blast cells in a sample. We train our method in a supervised manner and evaluate it on publicly available ALL FCM data from three different clinical centers. Our method reaches a median f1 score of ≈ 0.93 when tested on 200 B-ALL samples.

show abstract

“…Recently, several national reference laboratories of the iBFM-FLOW network and the AIEOP-BFM AML FLOW-MRD study group from across Europe [Austria, Germany, Italy, Poland, Russia (Moscow) and South America (Argentina)] have joined forces in a project called flowCLUSTER, dedicated to foster standardization and automation of MFC-MRD analysis in pedAML. They used machine learning technologies (61, 65), similarly to what already pursued in automated MFC-MRD data analysis of acute lymphoblastic leukemiasamples (57, 60, 64). It can be assumed that such an automated tool, together with central review and a program of continuous quality assessment, will provide standardization and high resolution in MFC-MRD assessment.…”

Section: Standardization Efforts For Mfc-mrd In Amlmentioning

confidence: 99%

Flow-Cytometric Monitoring of Minimal Residual Disease in Pediatric Patients With Acute Myeloid Leukemia: Recent Advances and Future Strategies

et al. 2019

View full text Add to dashboard Cite

Minimal residual disease (MRD) by multiparametric flow cytometry (MFC) has been recently shown as a strong and independent prognostic marker of relapse in pediatric AML (pedAML) when measured at specific time points during Induction and/or Consolidation therapy. Hence, MFC-MRD has the potential to refine the current strategies of pedAML risk stratification, traditionally based on the cytogenetic and molecular genetic aberrations at diagnosis. Consequently, it may guide the modulation of therapy intensity and clinical decision making. However, the use of non-standardized protocols, including different staining panels, analysis, and gating strategies, may hamper a broad implementation of MFC-MRD monitoring in clinical routine. Besides, the thresholds of MRD positivity still need to be validated in large, prospective and multi-center clinical studies, as well as optimal time points of MRD assessment during therapy, to better discriminate patients with different prognosis. In the present review, we summarize the most relevant findings on MFC-MRD testing in pedAML. We examine the clinical significance of MFC-MRD and the recent advances in its standardization, including innovative approaches with an automated analysis of MFC-MRD data. We also touch upon other technologies for MRD assessment in AML, such as quantitative genomic breakpoint PCR, current challenges and future strategies to enable full incorporation of MFC-MRD into clinical practice.

show abstract

WGAN Latent Space Embeddings for Blast Identification in Childhood Acute Myeloid Leukaemia

Cited by 11 publications

References 24 publications

UMAP Based Anomaly Detection for Minimal Residual Disease Quantification within Acute Myeloid Leukemia

UMAP Based Anomaly Detection for Minimal Residual Disease Quantification within Acute Myeloid Leukemia

Automated Identification of Cell Populations in Flow Cytometry Data with Transformers

Flow-Cytometric Monitoring of Minimal Residual Disease in Pediatric Patients With Acute Myeloid Leukemia: Recent Advances and Future Strategies

Contact Info

Product

Resources

About