| 1. |
- Ali, Muhaddisa Barat, 1986, et al.
(författare)
-
A novel federated deep learning scheme for glioma and its subtype classification
- 2023
-
Ingår i: Frontiers in Neuroscience. - 1662-4548 .- 1662-453X. ; 17
-
Tidskriftsartikel (refereegranskat)abstract
- Background: Deep learning (DL) has shown promising results in molecular-based classification of glioma subtypes from MR images. DL requires a large number of training data for achieving good generalization performance. Since brain tumor datasets are usually small in size, combination of such datasets from different hospitals are needed. Data privacy issue from hospitals often poses a constraint on such a practice. Federated learning (FL) has gained much attention lately as it trains a central DL model without requiring data sharing from different hospitals. Method: We propose a novel 3D FL scheme for glioma and its molecular subtype classification. In the scheme, a slice-based DL classifier, EtFedDyn, is exploited which is an extension of FedDyn, with the key differences on using focal loss cost function to tackle severe class imbalances in the datasets, and on multi-stream network to exploit MRIs in different modalities. By combining EtFedDyn with domain mapping as the pre-processing and 3D scan-based post-processing, the proposed scheme makes 3D brain scan-based classification on datasets from different dataset owners. To examine whether the FL scheme could replace the central learning (CL) one, we then compare the classification performance between the proposed FL and the corresponding CL schemes. Furthermore, detailed empirical-based analysis were also conducted to exam the effect of using domain mapping, 3D scan-based post-processing, different cost functions and different FL schemes. Results: Experiments were done on two case studies: classification of glioma subtypes (IDH mutation and wild-type on TCGA and US datasets in case A) and glioma grades (high/low grade glioma HGG and LGG on MICCAI dataset in case B). The proposed FL scheme has obtained good performance on the test sets (85.46%, 75.56%) for IDH subtypes and (89.28%, 90.72%) for glioma LGG/HGG all averaged on five runs. Comparing with the corresponding CL scheme, the drop in test accuracy from the proposed FL scheme is small (−1.17%, −0.83%), indicating its good potential to replace the CL scheme. Furthermore, the empirically tests have shown that an increased classification test accuracy by applying: domain mapping (0.4%, 1.85%) in case A; focal loss function (1.66%, 3.25%) in case A and (1.19%, 1.85%) in case B; 3D post-processing (2.11%, 2.23%) in case A and (1.81%, 2.39%) in case B and EtFedDyn over FedAvg classifier (1.05%, 1.55%) in case A and (1.23%, 1.81%) in case B with fast convergence, which all contributed to the improvement of overall performance in the proposed FL scheme. Conclusion: The proposed FL scheme is shown to be effective in predicting glioma and its subtypes by using MR images from test sets, with great potential of replacing the conventional CL approaches for training deep networks. This could help hospitals to maintain their data privacy, while using a federated trained classifier with nearly similar performance as that from a centrally trained one. Further detailed experiments have shown that different parts in the proposed 3D FL scheme, such as domain mapping (make datasets more uniform) and post-processing (scan-based classification), are essential.
|
|
| 2. |
- Ge, Chenjie, 1991, et al.
(författare)
-
Enlarged Training Dataset by Pairwise GANs for Molecular-Based Brain Tumor Classification
- 2020
-
Ingår i: IEEE Access. - 2169-3536 .- 2169-3536. ; 8:1, s. 22560-22570
-
Tidskriftsartikel (refereegranskat)abstract
- This paper addresses issues of brain tumor subtype classification using Magnetic Resonance Images (MRIs) from different scanner modalities like T1 weighted, T1 weighted with contrast-enhanced, T2 weighted and FLAIR images. Currently most available glioma datasets are relatively moderate in size, and often accompanied with incomplete MRIs in different modalities. To tackle the commonly encountered problems of insufficiently large brain tumor datasets and incomplete modality of image for deep learning, we propose to add augmented brain MR images to enlarge the training dataset by employing a pairwise Generative Adversarial Network (GAN) model. The pairwise GAN is able to generate synthetic MRIs across different modalities. To achieve the patient-level diagnostic result, we propose a post-processing strategy to combine the slice-level glioma subtype classification results by majority voting. A two-stage course-to-fine training strategy is proposed to learn the glioma feature using GAN-augmented MRIs followed by real MRIs. To evaluate the effectiveness of the proposed scheme, experiments have been conducted on a brain tumor dataset for classifying glioma molecular subtypes: isocitrate dehydrogenase 1 (IDH1) mutation and IDH1 wild-type. Our results on the dataset have shown good performance (with test accuracy 88.82%). Comparisons with several state-of-the-art methods are also included.
|
|
| 3. |
- Gryska, Emilia, 1992, et al.
(författare)
-
Deep learning for automatic brain tumour segmentation on MRI: evaluation of recommended reporting criteria via a reproduction and replication study.
- 2022
-
Ingår i: BMJ open. - : BMJ. - 2044-6055. ; 12:7
-
Tidskriftsartikel (refereegranskat)abstract
- To determine the reproducibility and replicability of studies that develop and validate segmentation methods for brain tumours on MRI and that follow established reproducibility criteria; and to evaluate whether the reporting guidelines are sufficient.Two eligible validation studies of distinct deep learning (DL) methods were identified. We implemented the methods using published information and retraced the reported validation steps. We evaluated to what extent the description of the methods enabled reproduction of the results. We further attempted to replicate reported findings on a clinical set of images acquired at our institute consisting of high-grade and low-grade glioma (HGG, LGG), and meningioma (MNG) cases.We successfully reproduced one of the two tumour segmentation methods. Insufficient description of the preprocessing pipeline and our inability to replicate the pipeline resulted in failure to reproduce the second method. The replication of the first method showed promising results in terms of Dice similarity coefficient (DSC) and sensitivity (Sen) on HGG cases (DSC=0.77, Sen=0.88) and LGG cases (DSC=0.73, Sen=0.83), however, poorer performance was observed for MNG cases (DSC=0.61, Sen=0.71). Preprocessing errors were identified that contributed to low quantitative scores in some cases.Established reproducibility criteria do not sufficiently emphasise description of the preprocessing pipeline. Discrepancies in preprocessing as a result of insufficient reporting are likely to influence segmentation outcomes and hinder clinical utilisation. A detailed description of the whole processing chain, including preprocessing, is thus necessary to obtain stronger evidence of the generalisability of DL-based brain tumour segmentation methods and to facilitate translation of the methods into clinical practice.
|
|
| 4. |
- Lee, SangWook, et al.
(författare)
-
Network analyses identify liver-specific targets for treating liver diseases
- 2017
-
Ingår i: Molecular Systems Biology. - : EMBO. - 1744-4292. ; 13:8
-
Tidskriftsartikel (refereegranskat)abstract
- We performed integrative network analyses to identify targets that can be used for effectively treating liver diseases with minimal side effects. We first generated co-expression networks (CNs) for 46 human tissues and liver cancer to explore the functional relationships between genes and examined the overlap between functional and physical interactions. Since increased de novo lipogenesis is a characteristic of nonalcoholic fatty liver disease (NAFLD) and hepatocellular carcinoma (HCC), we investigated the liver-specific genes co-expressed with fatty acid synthase (FASN). CN analyses predicted that inhibition of these liver-specific genes decreases FASN expression. Experiments in human cancer cell lines, mouse liver samples, and primary human hepatocytes validated our predictions by demonstrating functional relationships between these liver genes, and showing that their inhibition decreases cell growth and liver fat content. In conclusion, we identified liver-specific genes linked to NAFLD pathogenesis, such as pyruvate kinase liver and red blood cell (PKLR), or to HCC pathogenesis, such as PKLR, patatin-like phospholipase domain containing 3 (PNPLA3), and proprotein convertase subtilisin/kexin type 9 (PCSK9), all of which are potential targets for drug development.
|
|
| 5. |
- Kallak, Theodora Kunovac, 1985-, et al.
(författare)
-
Higher than expected estradiol levels in aromatase inhibitor-treated, postmenopausal breast cancer patients
- 2012
-
Ingår i: Climacteric. - London, United Kingdom : Informa Healthcare. - 1369-7137 .- 1473-0804. ; 15:5, s. 473-480
-
Tidskriftsartikel (refereegranskat)abstract
- Objective: Vaginal estradiol is considered contraindicated in aromatase inhibitor (AI)-treated patients because of the risk of elevated estrogen levels. This leaves limited treatment options for patients experiencing gynecological symptoms. However, in clinical practice, no precise estimation has been performed of circulating estrogens and aromatase index in postmenopausal breast cancer patients on long-lasting AI or tamoxifen treatment.Methods: Steroid hormones were measured using liquid chromatography tandem mass spectrometry (LC-MS/MS) and extraction radioimmunoassay (RIA). Postmenopausal AI-treated patients (n =33) were compared with tamoxifen-treated patients (n =34) and controls without vaginal treatment (n =56), with vaginal estradiol (n =25), or with estriol (n =11) treatment.Results: By use of LC-MS/MS, median (range) estradiol plasma concentrations were 16.7 (2.4-162.6), 31.0 (13.4-77.1), 27.2 (7.8-115.8) and 33.3 (20.3-340.1) pmol/l in AI-treated breast cancer patients, tamoxifen-treated breast cancer patients, postmenopausal controls and postmenopausal controls on vaginal estradiol, respectively. The AI-treated group and subgroups had significantly lower estradiol and estrone concentrations than all other groups (p <0.05). There was extensive interindividual variation in estradiol concentration within the AI-treated group, measured using both LC-MS/MS (2.3-182.0 pmol/l) and extraction RIA (2.4-162.6 pmol/l). The AI-treated group had lower aromatase index compared to all other groups (p <0.05-0.001).Conclusion: Circulating estrogen levels may have been underestimated in previous longitudinal studies of AI-treated breast cancer patients. Additional studies are required to further evaluate the role of circulating estrogens in breast cancer patients suffering from gynecological symptoms.
|
|
| 6. |
- Zanni, Giulia, et al.
(författare)
-
Lithium Accumulates in Neurogenic Brain Regions as Revealed by High Resolution Ion Imaging
- 2017
-
Ingår i: Scientific Reports. - : Springer Science and Business Media LLC. - 2045-2322. ; 7
-
Tidskriftsartikel (refereegranskat)abstract
- Lithium (Li) is a potent mood stabilizer and displays neuroprotective and neurogenic properties. Despite extensive investigations, the mechanisms of action have not been fully elucidated, especially in the juvenile, developing brain. Here we characterized lithium distribution in the juvenile mouse brain during 28 days of continuous treatment that result in clinically relevant serum concentrations. By using Time-of-Flight Secondary Ion Mass Spectrometry-(ToF-SIMS) based imaging we were able to delineate temporospatial lithium profile throughout the brain and concurrent distribution of endogenous lipids with high chemical specificity and spatial resolution. We found that Li accumulated in neurogenic regions and investigated the effects on hippocampal neurogenesis. Lithium increased proliferation, as judged by Ki67-immunoreactivity, but did not alter the number of doublecortin-positive neuroblasts at the end of the treatment period. Moreover, ToF-SIMS revealed a steady depletion of sphingomyelin in white matter regions during 28d Li-treatment, particularly in the olfactory bulb. In contrast, cortical levels of cholesterol and choline increased over time in Li-treated mice. This is the first study describing ToF-SIMS imaging for probing the brain-wide accumulation of supplemented Li in situ. The findings demonstrate that this technique is a powerful approach for investigating the distribution and effects of neuroprotective agents in the brain.
|
|
| 7. |
- Bayadsi, Haytham, 1987-, et al.
(författare)
-
The correlation between small papillary thyroid cancers and gamma radionuclides Cs-137, Th-232, U-238 and K-40 using spatially-explicit, register-based methods
- 2023
-
Ingår i: Spatial and Spatio-Temporal Epidemiology. - : Elsevier. - 1877-5845 .- 1877-5853. ; 47
-
Tidskriftsartikel (refereegranskat)abstract
- A steep increase of small papillary thyroid cancers (sPTCs) has been observed globally. A major risk factor for developing PTC is ionizing radiation. The aim of this study is to investigate the spatial distribution of sPTC in Sweden and the extent to which prevalence is correlated to gamma radiation levels (Caesium-137 (Cs-137), Thorium-232 (Th-232), Uranium-238 (U-238) and Potassium-40 (K-40)) using multiple geospatial and geo-statistical methods. The prevalence of metastatic sPTC was associated with significantly higher levels of Gamma radiation from Th-232, U-238 and K-40. The association is, however, inconsistent and the prevalence is higher in densely populated areas. The results clearly indicate that sPTC has causative factors that are neither evenly distributed among the population, nor geographically, calling for further studies with bigger cohorts. Environ-mental factors are believed to play a major role in the pathogenesis of the disease.
|
|
| 8. |
- Ge, Chenjie, 1991, et al.
(författare)
-
Cross-Modality Augmentation of Brain Mr Images Using a Novel Pairwise Generative Adversarial Network for Enhanced Glioma Classification
- 2019
-
Ingår i: Proceedings - International Conference on Image Processing, ICIP. - 1522-4880.
-
Konferensbidrag (refereegranskat)abstract
- © 2019 IEEE. Brain Magnetic Resonance Images (MRIs) are commonly used for tumor diagnosis. Machine learning for brain tumor characterization often uses MRIs from many modalities (e.g., T1-MRI, Enhanced-T1-MRI, T2-MRI and FLAIR). This paper tackles two issues that may impact brain tumor characterization performance from deep learning: insufficiently large training dataset, and incomplete collection of MRIs from different modalities. We propose a novel pairwise generative adversarial network (GAN) architecture for generating synthetic brain MRIs in missing modalities by using existing MRIs in other modalities. By improving the training dataset, we aim to mitigate the overfitting and improve the deep learning performance. Main contributions of the paper include: (a) propose a pairwise generative adversarial network (GAN) for brain image augmentation via cross-modality image generation; (b) propose a training strategy to enhance the glioma classification performance, where GAN-augmented images are used for pre-training, followed by refined-training using real brain MRIs; (c) demonstrate the proposed method through tests and comparisons of glioma classifiers that are trained from mixing real and GAN synthetic data, as well as from real data only. Experiments were conducted on an open TCGA dataset, containing 167 subjects for classifying IDH genotypes (mutation or wild-type). Test results from two experimental settings have both provided supports to the proposed method, where glioma classification performance has consistently improved by using mixed real and augmented data (test accuracy 81.03%, with 2.57% improvement).
|
|
| 9. |
- Haider, Zahra, 1988-
(författare)
-
DNA methylation signatures in precursor lymphoid neoplasms : with focus on clinical implications & the biology behind
- 2019
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Precursor lymphoid neoplasms, namely acute lymphoblastic leukemias (ALL) and lymphoblastic lymphomas (LBL), are characterized by an aggressive proliferation of malignant progenitor B- or T-cells. To improve risk classification at diagnosis, better prognostic and treatment stratifying biomarkers are needed. Altered DNA methylation pattern is a hallmark of neoplastic transformation, and has been employed as a molecular prognostic and predictive marker in various cancers, including hematological malignancies. Our research group previously identified a CpG island methylator phenotype (CIMP) panel that classified pediatric T-ALL patients into prognostic subgroups.The aim of this thesis was to evaluate distinct DNA methylation signatures in precursor lymphoid neoplasms, and to validate the prognostic value of CIMP classification in separate patient cohorts. Additionally, the biological mechanisms underlying the distinct CIMP methylation signatures in these malignancies were investigated.The prognostic relevance of CIMP classification was validated in an independent Nordic cohort of pediatric T-ALL patients. Combination of CIMP status with minimal residual disease (MRD) status, could further dissect the high-risk MRD positive T-ALL patients into two CIMP subgroups with significantly distinct outcomes. Furthermore, CIMP classification at diagnosis was shown to predict overall survival in relapsed BCP-ALL patients. CIMP methylation signatures were also identified in T-LBL patients, indicating a broader relevance of CIMP based classification in lymphoid malignancies. Investigating the biology behind CIMP methylation signatures showed the association of CIMP status with the proliferative history of the leukemic cells. A differential transcriptomic analysis revealed a correlation of CIMP subgroups with known T-ALL drivers, as well as with novel genes in T-ALL biology. Finally, we identified distinct DNA methylation patterns and genetic aberrations in T-ALL and T-LBL that might contribute to the different clinical presentation of these two diseases. In conclusion, we validated the prognostic significance of CIMP methylation signature in precursor lymphoid malignancies and identified transcriptomic profiles that associated with the subgroups. DNA methylation is a strong candidate for further risk classification in lymphoid neoplasms and our findings can contribute to the identification of new potential targets for treatment.
|
|
| 10. |
- Sadeghi, B., et al.
(författare)
-
Early-phase GVHD gene expression profile in target versus non-target tissues : kidney, a possible target?
- 2013
-
Ingår i: Bone Marrow Transplantation. - : Springer Science and Business Media LLC. - 0268-3369 .- 1476-5365. ; 48:2, s. 284-293
-
Tidskriftsartikel (refereegranskat)abstract
- GVHD is a major complication after allo-SCT. In GVHD, some tissues like liver, intestine and skin are infiltrated by donor T cells while others like muscle are not. The mechanism underlying targeted tropism of donor T cells is not fully understood. In the present study, we aim to explore differences in gene expression profile among target versus non-target tissues in a mouse model of GVHD based on chemotherapy conditioning. Expression levels of JAK-signal transducers and activators of transcription (STAT), CXCL1, ICAM1 and STAT3 were increased in the liver and remained unchanged (or decreased) in the muscle and kidney after conditioning. At the start of GVHD the expression levels of CXCL9, ITGb2, SAA3, MARCO, TLR and VCAM1 were significantly higher in the liver or kidney compared with the muscle of GVHD animals. Moreover, biological processes of inflammatory reactions, leukocyte migration, response to bacterium and chemotaxis followed the same pattern. Our data show that both chemotherapy and allogenicity exclusively induce expression of inflammatory genes in target tissues. Moreover, gene expression profile and histopathological findings in the kidney are similar to those observed in the liver of GVHD mice. Bone Marrow Transplantation (2013) 48, 284-293; doi:10.1038/bmt.2012.120; published online 23 July 2012
|
|
