Tuija Keinänen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Tuija Keinänen has a total of 78 co-authors. Here are the top co-authors by number of shared publications:
- Vesa Korhonen
Oulu Functional Neuroimaging, Research unit of Health Sciences and Technology, Faculty of Medicine, Medical Research Center, University of Oulu
5 shared publications Juha Nikkinen
Affiliations Department of Oncology and Radiotherapy, Oulu University Hospital, Oulu, Finland, Research Unit of Health Sciences and Technology, University of Oulu, Oulu, Finland, Medical Research Center Oulu, Oulu University Hospital and University of Oulu, Oulu, Finland
3 shared publications- Niina Salokorpi
Department of Neurosurgery, Oulu University Hospital, Oulu, Finland and Research Unit of Clinical Neuroscience, Medical Research Center, Oulu University, Oulu, Finland
3 shared publications - Niko Huotari
Oulu Functional Neuroimaging, Research unit of Health Sciences and Technology, Faculty of Medicine, Medical Research Center, University of Oulu
3 shared publications Osmo Tervonen
Oulu Functional NeuroImaging (OFNI), Department of Diagnostic Radiology, Oulu University Hospital, Oulu, Finland
3 shared publicationsEveliina Lammentausta
Department of Diagnostic Radiology, Faculty of Medicine, Research Unit of Health Sciences and Technology, Oulu University Hospital, University of Oulu, Oulu, Finland
2 shared publications
Latest Publications
- Open Access
- 13 September 2026
- Published by Elsevier in Physica Medica
- Vol. 150
Tight collimation substantially increases air kerma rate and decreases image contrast during selected fluoroscopy procedures
- 13 September 2026
- Published by Elsevier in Physica Medica
- Vol. 150
Open AccessTight collimation substantially increases air kerma rate and decreases image contrast during selected fluoroscopy procedures
- Open Access
- 17 June 2026
- Published by Springer Nature
- p. 297-304
Multiparametric MRI Analysis in Glioma Grading: Added Value of ADC Maps
The current standard in glioma diagnosis and grading relies on histopathological analysis of tumor tissue obtained through biopsy or surgical resection. These invasive procedures may expose patients to increased risk of complications. In this study, we investigated the potential of using multi-sequence MRI as a noninvasive method for glioma grading. The analysis was conducted on a dataset of 36 glioma patients with contrast-enhanced T1 (T1c) images, Apparent Diffusion Coefficient (ADC) maps, tumor annotations, and patient demographics. The imaging data was preprocessed using intra-subject registration, brain extraction, and resampling. A subset of 19 T1c and 14 ADC radiomics was selected for further analysis. After dimensionality reduction, bias-corrected and accelerated bootstrapping was utilized alongside Firth logistic regression for developing a T1c-based model, and a multi-sequence model. The two models were compared using the likelihood ratio test and the Akaike information criterion. The inclusion of ADC parameters significantly improved the model in terms of the quality of fit to the data. The results further implied that increased tumor texture heterogeneity and decreased ADC values could be associated with higher glioma grade. These findings highlight the potential utility of multi-sequence MRI for digital biopsy in glioma grading, although validation in a larger patient cohort is required.
- 17 June 2026
- Published by Springer Nature
- p. 297-304
Open AccessMultiparametric MRI Analysis in Glioma Grading: Added Value of ADC Maps
The current standard in glioma diagnosis and grading relies on histopathological analysis of tumor tissue obtained through biopsy or surgical resection. These invasive procedures may expose patients to increased risk of complications. In this study, we investigated the potential of using multi-sequence MRI as a noninvasive method for glioma grading. The analysis was conducted on a dataset of 36 glioma patients with contrast-enhanced T1 (T1c) images, Apparent Diffusion Coefficient (ADC) maps, tumor annotations, and patient demographics. The imaging data was preprocessed using intra-subject registration, brain extraction, and resampling. A subset of 19 T1c and 14 ADC radiomics was selected for further analysis. After dimensionality reduction, bias-corrected and accelerated bootstrapping was utilized alongside Firth logistic regression for developing a T1c-based model, and a multi-sequence model. The two models were compared using the likelihood ratio test and the Akaike information criterion. The inclusion of ADC parameters significantly improved the model in terms of the quality of fit to the data. The results further implied that increased tumor texture heterogeneity and decreased ADC values could be associated with higher glioma grade. These findings highlight the potential utility of multi-sequence MRI for digital biopsy in glioma grading, although validation in a larger patient cohort is required.
- Open Access
- 15 May 2026
- Published by Elsevier in World Neurosurgery
- Vol. 212, 125057
The Role of Magnetic Resonance Spectroscopy in Brain Tumor Biopsy Planning: Insights from Neurosurgeons' Perspectives
- 15 May 2026
- Published by Elsevier in World Neurosurgery
- Vol. 212, 125057
Open AccessThe Role of Magnetic Resonance Spectroscopy in Brain Tumor Biopsy Planning: Insights from Neurosurgeons' Perspectives
- Open Access
- 14 November 2025
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 20 (11) , e0336387
Improved brain tumor diagnostics and follow-up with novel magnetic resonance imaging methods: A single center study protocol
This protocol outlines a prospective study aimed at enhancing the diagnosis and monitoring of brain tumors through advanced non-invasive imaging techniques. While magnetic resonance imaging (MRI) is a cornerstone of brain tumor diagnostics, it often lacks the specificity required for definitive diagnosis, which typically relies on invasive tissue sampling. To address this, the study will evaluate advanced MRI techniques—such as perfusion, diffusion, blood-oxygen-level-dependent imaging, magnetic resonance spectroscopy, and amide proton transfer-weighted imaging— that offer valuable physiological and molecular insights, beyond conventional anatomical imaging. Despite their potential, clinical adoption of these methods remains limited. MRI also plays a central role in treatment response assessment and follow-up, yet conventional anatomical sequences may not detect early physiological changes or differentiate true progression from pseudoprogression. Advanced imaging methods have shown promise in addressing these limitations, and predictive models for recurrence risk could further personalize treatment strategies. In this study, imaging will be performed using a standardized 3T MRI scanner at multiple time points: preoperatively, before radiotherapy, during treatment, and throughout follow-up. This protocol aims to establish a multiparametric imaging framework capable of capturing dynamic physiological and molecular changes in brain tumors. The primary goal is to determine whether combining advanced sequences improves diagnostic accuracy compared to conventional MRI, using histopathology as the reference. Secondary objectives include predicting treatment response, distinguishing true progression from pseudoprogression, and modeling spatial recurrence risk based on quantitative imaging biomarkers. We hypothesize that a multiparametric imaging approach will, enable earlier detection of tumor progression and support more precise, individualized treatment decisions.
- 14 November 2025
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 20 (11) , e0336387
Open AccessImproved brain tumor diagnostics and follow-up with novel magnetic resonance imaging methods: A single center study protocol
This protocol outlines a prospective study aimed at enhancing the diagnosis and monitoring of brain tumors through advanced non-invasive imaging techniques. While magnetic resonance imaging (MRI) is a cornerstone of brain tumor diagnostics, it often lacks the specificity required for definitive diagnosis, which typically relies on invasive tissue sampling. To address this, the study will evaluate advanced MRI techniques—such as perfusion, diffusion, blood-oxygen-level-dependent imaging, magnetic resonance spectroscopy, and amide proton transfer-weighted imaging— that offer valuable physiological and molecular insights, beyond conventional anatomical imaging. Despite their potential, clinical adoption of these methods remains limited. MRI also plays a central role in treatment response assessment and follow-up, yet conventional anatomical sequences may not detect early physiological changes or differentiate true progression from pseudoprogression. Advanced imaging methods have shown promise in addressing these limitations, and predictive models for recurrence risk could further personalize treatment strategies. In this study, imaging will be performed using a standardized 3T MRI scanner at multiple time points: preoperatively, before radiotherapy, during treatment, and throughout follow-up. This protocol aims to establish a multiparametric imaging framework capable of capturing dynamic physiological and molecular changes in brain tumors. The primary goal is to determine whether combining advanced sequences improves diagnostic accuracy compared to conventional MRI, using histopathology as the reference. Secondary objectives include predicting treatment response, distinguishing true progression from pseudoprogression, and modeling spatial recurrence risk based on quantitative imaging biomarkers. We hypothesize that a multiparametric imaging approach will, enable earlier detection of tumor progression and support more precise, individualized treatment decisions.
- Open Access
- 1 October 2025
- Published by Elsevier in Brain and Spine
- Vol. 5
Exploring the role of volumetric and radiological features in meningioma behaviour and follow-up strategies
- 1 October 2025
- Published by Elsevier in Brain and Spine
- Vol. 5
Open AccessExploring the role of volumetric and radiological features in meningioma behaviour and follow-up strategies
- Open Access
- 1 January 2019
- Published by Elsevier in NeuroImage: Clinical
- Vol. 22, 101763
Spectral entropy indicates electrophysiological and hemodynamic changes in drug-resistant epilepsy – A multimodal MREG study
- 1 January 2019
- Published by Elsevier in NeuroImage: Clinical
- Vol. 22, 101763
Open AccessSpectral entropy indicates electrophysiological and hemodynamic changes in drug-resistant epilepsy – A multimodal MREG study
- 14 May 2018
- Published by Wiley in Journal of Neuroscience Research
- Vol. 96 (10) , 1689-1698
Fluctuations of the EEG‐fMRI correlation reflect intrinsic strength of functional connectivity in default mode network
Both functional magnetic resonance imaging (fMRI) and electrophysiological recordings have revealed that resting‐state functional connectivity is temporally variable in human brain. Combined full‐band electroencephalography‐fMRI (fbEEG‐fMRI) studies have shown that infraslow (<.1 Hz) fluctuations in EEG scalp potential are correlated with the blood‐oxygen‐level‐dependent (BOLD) fMRI signals and that also this correlation appears variable over time. Here, we used simultaneous fbEEG‐fMRI to test the hypothesis that correlation dynamics between BOLD and fbEEG signals could be explained by fluctuations in the activation properties of resting‐state networks (RSNs) such as the extent or strength of their activation. We used ultrafast magnetic resonance encephalography (MREG) fMRI to enable temporally accurate and statistically robust short‐time‐window comparisons of infra‐slow fbEEG and BOLD signals. We found that the temporal fluctuations in the fbEEG‐BOLD correlation were dependent on RSN connectivity strength, but not on the mean signal level or magnitude of RSN activation or motion during scanning. Moreover, the EEG‐fMRI correlations were strongest when the intrinsic RSN connectivity was strong and close to the pial surface. Conversely, weak fbEEG‐BOLD correlations were attributable to periods of less coherent or spatially more scattered intrinsic RSN connectivity, or RSN activation in deeper cerebral structures. The results thus show that the on‐average low correlations between infra‐slow EEG and BOLD signals are, in fact, governed by the momentary coherence and depth of the underlying RSN activation, and may reach systematically high values with appropriate source activities. These findings further consolidate the notion of slow scalp potentials being directly coupled to hemodynamic fluctuations.
- 14 May 2018
- Published by Wiley in Journal of Neuroscience Research
- Vol. 96 (10) , 1689-1698
Fluctuations of the EEG‐fMRI correlation reflect intrinsic strength of functional connectivity in default mode network
Both functional magnetic resonance imaging (fMRI) and electrophysiological recordings have revealed that resting‐state functional connectivity is temporally variable in human brain. Combined full‐band electroencephalography‐fMRI (fbEEG‐fMRI) studies have shown that infraslow (<.1 Hz) fluctuations in EEG scalp potential are correlated with the blood‐oxygen‐level‐dependent (BOLD) fMRI signals and that also this correlation appears variable over time. Here, we used simultaneous fbEEG‐fMRI to test the hypothesis that correlation dynamics between BOLD and fbEEG signals could be explained by fluctuations in the activation properties of resting‐state networks (RSNs) such as the extent or strength of their activation. We used ultrafast magnetic resonance encephalography (MREG) fMRI to enable temporally accurate and statistically robust short‐time‐window comparisons of infra‐slow fbEEG and BOLD signals. We found that the temporal fluctuations in the fbEEG‐BOLD correlation were dependent on RSN connectivity strength, but not on the mean signal level or magnitude of RSN activation or motion during scanning. Moreover, the EEG‐fMRI correlations were strongest when the intrinsic RSN connectivity was strong and close to the pial surface. Conversely, weak fbEEG‐BOLD correlations were attributable to periods of less coherent or spatially more scattered intrinsic RSN connectivity, or RSN activation in deeper cerebral structures. The results thus show that the on‐average low correlations between infra‐slow EEG and BOLD signals are, in fact, governed by the momentary coherence and depth of the underlying RSN activation, and may reach systematically high values with appropriate source activities. These findings further consolidate the notion of slow scalp potentials being directly coupled to hemodynamic fluctuations.
- Open Access
- 20 March 2017
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 12 (3) , e0174072
Real-time monitoring of human blood-brain barrier disruption
Chemotherapy aided by opening of the blood-brain barrier with intra-arterial infusion of hyperosmolar mannitol improves the outcome in primary central nervous system lymphoma. Proper opening of the blood-brain barrier is crucial for the treatment, yet there are no means available for its real-time monitoring. The intact blood-brain barrier maintains a mV-level electrical potential difference between blood and brain tissue, giving rise to a measurable electrical signal at the scalp. Therefore, we used direct-current electroencephalography (DC-EEG) to characterize the spatiotemporal behavior of scalp-recorded slow electrical signals during blood-brain barrier opening. Nine anesthetized patients receiving chemotherapy were monitored continuously during 47 blood-brain barrier openings induced by carotid or vertebral artery mannitol infusion. Left or right carotid artery mannitol infusion generated a strongly lateralized DC-EEG response that began with a 2 min negative shift of up to 2000 μV followed by a positive shift lasting up to 20 min above the infused carotid artery territory, whereas contralateral responses were of opposite polarity. Vertebral artery mannitol infusion gave rise to a minimally lateralized and more uniformly distributed slow negative response with a posterior-frontal gradient. Simultaneously performed near-infrared spectroscopy detected a multiphasic response beginning with mannitol-bolus induced dilution of blood and ending in a prolonged increase in the oxy/deoxyhemoglobin ratio. The pronounced DC-EEG shifts are readily accounted for by opening and sealing of the blood-brain barrier. These data show that DC-EEG is a promising real-time monitoring tool for blood-brain barrier disruption augmented drug delivery.
- 20 March 2017
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 12 (3) , e0174072
Open AccessReal-time monitoring of human blood-brain barrier disruption
Chemotherapy aided by opening of the blood-brain barrier with intra-arterial infusion of hyperosmolar mannitol improves the outcome in primary central nervous system lymphoma. Proper opening of the blood-brain barrier is crucial for the treatment, yet there are no means available for its real-time monitoring. The intact blood-brain barrier maintains a mV-level electrical potential difference between blood and brain tissue, giving rise to a measurable electrical signal at the scalp. Therefore, we used direct-current electroencephalography (DC-EEG) to characterize the spatiotemporal behavior of scalp-recorded slow electrical signals during blood-brain barrier opening. Nine anesthetized patients receiving chemotherapy were monitored continuously during 47 blood-brain barrier openings induced by carotid or vertebral artery mannitol infusion. Left or right carotid artery mannitol infusion generated a strongly lateralized DC-EEG response that began with a 2 min negative shift of up to 2000 μV followed by a positive shift lasting up to 20 min above the infused carotid artery territory, whereas contralateral responses were of opposite polarity. Vertebral artery mannitol infusion gave rise to a minimally lateralized and more uniformly distributed slow negative response with a posterior-frontal gradient. Simultaneously performed near-infrared spectroscopy detected a multiphasic response beginning with mannitol-bolus induced dilution of blood and ending in a prolonged increase in the oxy/deoxyhemoglobin ratio. The pronounced DC-EEG shifts are readily accounted for by opening and sealing of the blood-brain barrier. These data show that DC-EEG is a promising real-time monitoring tool for blood-brain barrier disruption augmented drug delivery.
- Open Access
- 1 March 2017
- Published by Elsevier in NeuroImage
- Vol. 148, 352-363
Combined spatiotemporal ICA (stICA) for continuous and dynamic lag structure analysis of MREG data
- 1 March 2017
- Published by Elsevier in NeuroImage
- Vol. 148, 352-363
Open AccessCombined spatiotemporal ICA (stICA) for continuous and dynamic lag structure analysis of MREG data
- Top Cited Papers
- 21 December 2015
- Published by SAGE Publications in Journal of Cerebral Blood Flow & Metabolism
- Vol. 36 (6) , 1033-1045
Ultra-fast magnetic resonance encephalography of physiological brain activity – Glymphatic pulsation mechanisms?
The theory on the glymphatic convection mechanism of cerebrospinal fluid holds that cardiac pulsations in part pump cerebrospinal fluid from the peri-arterial spaces through the extracellular tissue into the peri-venous spaces facilitated by aquaporin water channels. Since cardiac pulses cannot be the sole mechanism of glymphatic propulsion, we searched for additional cerebrospinal fluid pulsations in the human brain with ultra-fast magnetic resonance encephalography. We detected three types of physiological mechanisms affecting cerebral cerebrospinal fluid pulsations: cardiac, respiratory, and very low frequency pulsations. The cardiac pulsations induce a negative magnetic resonance encephalography signal change in peri-arterial regions that extends centrifugally and covers the brain in ≈1 Hz cycles. The respiratory ≈0.3 Hz pulsations are centripetal periodical pulses that occur dominantly in peri-venous areas. The third type of pulsation was very low frequency (VLF 0.001–0.023 Hz) and low frequency (LF 0.023–0.73 Hz) waves that both propagate with unique spatiotemporal patterns. Our findings using critically sampled magnetic resonance encephalography open a new view into cerebral fluid dynamics. Since glymphatic system failure may precede protein accumulations in diseases such as Alzheimer's dementia, this methodological advance offers a novel approach to image brain fluid dynamics that potentially can enable early detection and intervention in neurodegenerative diseases.
- 21 December 2015
- Published by SAGE Publications in Journal of Cerebral Blood Flow & Metabolism
- Vol. 36 (6) , 1033-1045
Top Cited PapersUltra-fast magnetic resonance encephalography of physiological brain activity – Glymphatic pulsation mechanisms?
The theory on the glymphatic convection mechanism of cerebrospinal fluid holds that cardiac pulsations in part pump cerebrospinal fluid from the peri-arterial spaces through the extracellular tissue into the peri-venous spaces facilitated by aquaporin water channels. Since cardiac pulses cannot be the sole mechanism of glymphatic propulsion, we searched for additional cerebrospinal fluid pulsations in the human brain with ultra-fast magnetic resonance encephalography. We detected three types of physiological mechanisms affecting cerebral cerebrospinal fluid pulsations: cardiac, respiratory, and very low frequency pulsations. The cardiac pulsations induce a negative magnetic resonance encephalography signal change in peri-arterial regions that extends centrifugally and covers the brain in ≈1 Hz cycles. The respiratory ≈0.3 Hz pulsations are centripetal periodical pulses that occur dominantly in peri-venous areas. The third type of pulsation was very low frequency (VLF 0.001–0.023 Hz) and low frequency (LF 0.023–0.73 Hz) waves that both propagate with unique spatiotemporal patterns. Our findings using critically sampled magnetic resonance encephalography open a new view into cerebral fluid dynamics. Since glymphatic system failure may precede protein accumulations in diseases such as Alzheimer's dementia, this methodological advance offers a novel approach to image brain fluid dynamics that potentially can enable early detection and intervention in neurodegenerative diseases.