Infotech Oulu Annual Report 2015 - Biomedical Engineering Research Group (BME)

Professor Tapio Seppänen, Center for Machine Vision and Signal Analysis, and Professor Timo Jämsä, Research Unit of Medical Imaging, Physics and Technology, University of Oulu
tapio.seppanen(at)ee.oulu.fi, timo.jamsa(at)oulu.fi

http://www.oulu.fi/cse/bme

 

Background and Mission

The mission of the Biomedical engineering group (BME) is to carry out top-level basic, applied and translational research in biomedical engineering. The aim is to develop, apply and evaluate novel biomedical measurement technologies in health and wellbeing. The research is interdisciplinary and focuses on measurement problems with the cardiovascular system, central nervous system, respiratory system and musculoskeletal system, together with applications in biomedicine and eHealth.

The BME is organized around the professors of biomedical engineering and medical physics in two teams: Biosignal processing team (Faculty of information technology and electrical engineering) and Biomechanics research team (Faculty of medicine). The research profile is based on linking information technology and medicine, with an aim to utilize methodologies of information engineering, signal and image processing, pattern recognition, biophysics, medical imaging, applied mathematics, simulation, biomedicine and clinical medicine. BME has strong national and international collaborative networks, including partners in USA, Japan and many European universities. 

 

Scientific Progress

Cardiorespiratory System Research

Allergic rhinitis is a major global health problem affecting to more than 500 million people around the world and the prevalence is increasing in most countries. It is a major chronic respiratory disease due to its prevalence, impact on quality of life and school/work productivity and performance. Furthermore, it is a substantial economic burden to societies. Allergic rhinitis is an inheritable systemic inflammatory condition which links with other illnesses like asthma. It is very commonly underdiagnosed and undertreated. For these reasons, specific guidelines and programs on the problem have been released, e.g. by EU and WHO.

Allergic rhinitis is considered to be an immuno-neuronal disorder, but little is known about the part played by the neural system in nasal allergic reaction. This is due mainly to the lack of objective measurement techniques producing reliable, accurate and continuous measurement data about the dynamic changes in nasal respiratory function.

We have previously presented a novel method to assess nasal airflow resistance in a way that provides continuous resistance values. The nasal pressure signal is measured with a nasopharyngeal catheter and the respiratory airflow signal with the respiratory effort belts calibrated with our new method. The resistance is calculated for each signal sample at any sampling frequency, making it possible to discover rapid changes in resistance, which is essential, for example, during provocation tests. We produced the continuous resistance using above mentioned methods and dynamically changing HRV (Heart Rate Variability) parameters were computed to study their concurrent changes during an allergen provocation test. The LF (Low Frequency) and HF (High Frequency) components were computed in 1 min windows (overlap of 50 %). In each 1 min window, the Hamming window was applied to the tachogram and the spectral analysis was performed using a FFT (Fast Fourier Transform) algorithm.

To our knowledge, this is the first time that it is possible to estimate from the concurrent continuous nasal airflow resistance and HRV parameters:  are there associations between the dynamic reactions of nasal airflow resistance and HRV during the allergic reaction.

Ten birch pollen allergic volunteers were challenged with the allergen. Continuous nasal airflow resistance and spectral heart rate variability parameters were computed and analyzed for the dynamic changes. The derived signals showed in detail the timing and intensity differences in subjects’ reactions. A representative case of concomitant dynamic changes in the nasal airflow resistance and spectral features during allergen provocation is shown in Figure. 1. A small gap in the resistance signal can be seen due to removing of artifacts (sneezes) during manual validation.

After the provocation, the nasal airflow resistance rose gradually, whereas LF power and LF/HF ratio decreased gradually for all subjects. This implies gradually increasing sympathetic withdrawal in allergic patients during the provocation with allergen.

The proposed method opens entirely new possibilities to assess accurately concomitant changes in non-stationary nasal function and ANS. This could increase the accuracy and reliability of diagnostics and assessment of the effect of nasal treatments and desensitization.

Figure 1. An example of an increase of nasal airflow resistance and related LF and LF/HF ratio values. Top: resistance (blue) and robust fit of the resistance curve (red). Middle: LF (blue) and robust fit of the LF curve (red). Bottom: LF/HF ratio (blue) and robust fit of the LF/HF ratio curve.

Respiratory disorders are a very common and growing health problem globally. Signal waveforms of respiratory volume and airflow may indicate pathological signs of several diseases and, thus, it would be important to measure them accurately. Currently, devices used in respiration measurements are mostly obtrusive in nature interfering with the natural respiration.

We have presented a novel method for the measurement of accurate continuous respiratory airflow with a depth camera (Kinect) without a contact on a subject. We proposes a novel calibration method which enables accurate estimates of the respiratory airflow waveforms from the depth camera data.

Depth video sequences from a Microsoft Kinect device were used for contactless measurement of respiration movement. At the beginning of each test subject’s measurement, an image from the background was taken so that it could be excluded from the actual measurement view. The measurement data consisted of frame-synchronized consecutive point clouds from the subject in the world coordinates. Two virtual respiratory effort belts were created from the depth image sequence. They must be calibrated with the actual airflow signal from the spirometer before they can be used for detailed waveform analysis purposes.

We have previously developed the respiratory effort belt calibration method, which is robust to breathing style changes and reduces waveform error. The method was applied to the respiratory signals derived from the depth data of Kinect to produce accurate respiratory airflow estimates. The method is based on an optimally trained FIR (Finite Impulse Response) filter bank constructed as a MISO (Multiple-Input Single-Output) system between the respiratory signals from the Kinect and the spirometer. It extends the conventional multiple linear regression method in two important ways: 1) it is based on polynomial regression to model different transfer functions between the input and output, and 2) it uses a number N of consecutive input signal samples and linear filtering for estimation of each output signal sample. Five different transfer models were created and used.

Eight subjects were measured with the depth camera and spirometer at the same time using different breathing styles. Figure 2 depicts short segment of example signals from one subject.

Figure 2. Short segment of example signals from one subject: spirometer signal (black line) and the estimated respiratory airflow from Kinect (red line).

Results showed that not only the respiratory volume and respiratory rate can be computed precisely from the estimated respiratory airflow, but also the respiratory airflow waveforms are very accurate. The analysis of the thoracoabdominal asynchrony or the regional pulmonary function becomes available with depth camera techniques, as well. Hence, this measurement method could be used for example to monitor and analyze the respiratory function unobtrusively and continuously in pulmonary and critical care medicine. Accurate data on signal waveforms of respiratory volume and airflow could enable earlier diagnostics of respiratory pathology.

Postoperative respiratory complications are common in patients after the general anaesthesia. Respiratory depression often occurs in association with postoperative opioid analgesia. It can arise from pain, residual operating room anaesthetics and administration of opioids for pain management. Inadequate respiration can result in respiratory complications, morbidity, mortality and increased costs. Accurate respiration monitoring postoperatively is important, so that respiratory depression can be identified as early as possible. Currently, there is a need for an accurate, continuous and non-invasive respiratory monitoring of spontaneously breathing postoperative patients.

We used respiratory effort belts calibrated with our previously developed method for the respiratory monitoring pre- and postoperatively. Our method enables accurate estimates of the respiratory airflow waveforms. Five patients who had lumbar back surgery and were expected to need opioid analgesia postoperatively were recruited. They were measured with the spirometer and respiratory effort belts at the same time. Preoperative measurements (length 5 min) were done in the operating room just before the operation without any sedative medication. Postoperative measurements (length 3 h) were done in the recovery room as soon as possible after the operation. We compared three calibration models pre- and postoperatively.

Results showed that postoperative calibration produced much more accurate respiratory airflows.  Not only the tidal volume, minute volume and respiratory rate can be computed precisely from the estimated respiratory airflow, but also the respiratory airflow waveforms are very accurate. The method produced accurate estimates even from the following challenging respiratory signals: low airflows (Figure 3), hypopneic events (Figure 4) and thoracoabdominal asynchrony (Figure 5).

Figure 3. Short segment of example signals with low airflow: spirometer signal (black line) and the estimated respiratory airflow signal (red line).

 

Figure 4. Example signals of hypopneic event. Upper subfigure: spirometer signal (black line) and the estimated respiratory airflow signal (red line). Lower subfigure: rig cage respiratory effort belt signal (green line) and abdominal respiratory effort belt signal (green line).

 

Figure 5. Example signals of thoracoabdominal asynchrony. Upper subfigure: spirometer signal (black line) and the estimated respiratory airflow signal (red line). Lower subfigure: rib cage respiratory effort belt signal (blue line) and abdominal respiratory effort belt signal (green line).

The presented method is able to produce estimates of postoperative respiratory airflow waveforms to enable accurate, continuous, non-invasive respiratory monitoring postoperatively.

Central Nervous System Research

Estimation of Neural Damage in Brain

Hypoxic ischemic encephalopathy (HIE) is a severe complication of cardiac arrest (CA). Detecting the need and assessing the effect of different therapeutic interventions and other treatment strategies protecting CA patients from permanent neural damage requires an objective measurement of brain function in the intensive care unit (ICU). However, at the moment, no reliable technological solution for this exists. Defining the patient’s neurological condition relies on a number of different kinds of examinations which, even when applied together, are unreliable giving only a rough estimate of the damage. In collaboration with Oulu University Hospital, we are developing a novel technology for detecting and measuring HIE. Our approach combines measurement of electroencephalogram (EEG) with common pharmacological compounds and clinical procedures to empower physicians with the first truly objective and practical measurement of brain function. After finishing a successful pilot study, the technology is validated through a second clinical study with results expected in late 2016.

Figure 6. The topographic representation of an EEG-derived measure at different anesthetic levels in a patient with and without hypoxic brain injury.

 

Figure 7. The behaviour of different intrinsic mode functions (IMFs) calculated from EEG using Hilbert-Huang Transform during deepening level of anesthesia.

Slow waves (< 1 Hz) are considered to be the most important EEG signature of non-rapid eye movement sleep and have substantial physiological importance. In addition to natural sleep, slow waves can be seen in the EEG during general anesthesia offering great potential for depth of anesthesia monitoring. We applied Hilbert-Huang Transform, an adaptive data-driven method designed for the analysis on non-stationary data, to investigate the dynamical changes in the EEG slow wave activity during induction of anesthesia with propofol. The method was found to be able to extract stable signal components representing slow wave activity that were consistent between patients. The signal analysis revealed a possible specific structure between different components dependent on the depth of anesthesia on which further studies are needed.

Hypothermia is known to be neuroprotective and is one of the most effective and promising first-line treatments for central nervous system trauma. At present, induction of local hypothermia, as opposed to general hypothermia, is more desired because of its ease of application and safety; fewer side effects and an absence of severe complications have been noted. Local hypothermia involves temperature reduction of a small and specific segment of the spinal cord. The neuroprotective effect of short-term, acute moderate general hypothermia has been shown through improvements in electrophysiological and motor behavioral assessments, as well as histological examination following contusive spinal cord injury (SCI) in rats. The benefits of using short-term local hypothermia versus short-term general hypothermia post-acute SCI have also been shown. In collaboration with National University of Singapore, a series of experiments were designed to investigate the feasibility, long-term safety, as well as eventual complications and side effects of prolonged, semi-invasive, moderate local hypothermia in rats with uninjured spinal cord. The weekly somatosensory evoked potential and motor behavioral assessments of rats that underwent 5 and 8 hours of semi-invasive local hypothermia, revealed no statistically significant changes. In addition, 4 weeks after local hypothermia induction, histological examination showed no anatomical damages or morphological changes in their spinal cord structure and parenchyma. We concluded that this method of prolonged local hypothermia is feasible, safe, and has potential for clinical translation.

Figure 8. The calculation of somatosensory evoked potential (SEP) parameters used in the assessment of the severity of spinal cord injury.

Recent findings suggest that specific neural correlates for the key elements of basic emotions do exist and can be identified by neuroimaging techniques. We used EEG to explore the markers for video-induced emotions. The problem was approached from a classifier perspective: the features that perform best in classifying person’s valence and arousal while watching video clips with audiovisual emotional content were searched from a large feature set constructed from the EEG spectral powers of single channels as well as power differences between specific channel pairs. The feature selection was carried out using a sequential forward floating search method and was done separately for the classification of valence and arousal, both derived from the emotional keyword the subject had chosen after seeing the clips. The proposed classifier-based approach revealed a clear association between the increased high-frequency (15-32 Hz) activity in the left temporal area and the clips described as ‘pleasant’ in the valence and ‘medium arousal’ in the arousal scale. These clips represent the emotional keywords amusement and joy/happiness. The finding suggests the occurrence of a specific neural activation during video-induced pleasant emotion and the possibility to detect this from the left temporal area using EEG.

Figure 9. The topographic and spectral features of EEG revealing video-induced pleasant emotion.

 

Speeding-up of Brain Data Search Engines

A method and implementation were developed to achieve a thousand fold speed-up for seeking of large files in a commonly used compressed neuroimaging data format NIfTI. Such technologies are not currently available in this research field while they would make the everyday work for hundreds of researchers and experts much smoother and faster. The method includes the creation of a novel index structure for the compressed data in order to achieve the speed-up. With random seek simulations, we demonstrate that a speed-up of over hundred up to even five thousand can be reached compared to the currently available implementations. By configuring the index structure properly, one can set an operating point which optimizes the efficiency as speed-up versus index size according to the requirements by the user (Figure 10). For example, a thousand fold speed-up can be achieved with an index size of only about two percent of the original compressed data.

Figure 10. Efficiency of zindex data for different access point distances. On the vertical axis, the relative random access speed-up per megabytes of zindex data is given.

 

Neural Avalanche Detection in Brain

Recent studies pinpoint visually cued networks of avalanches with MEG/EEG data. Co-activation pattern (CAP) analysis can be used to detect single brain volume activity profiles and hemodynamic fingerprints of neuronal avalanches as sudden high signal activity peaks in classical fMRI data. We aimed to detect dynamic patterns of brain activity spreads with the use of ultrafast MR encephalography (MREG). MREG achieves 10 Hz whole brain sampling, allowing the estimation of spatial spread of an avalanche, even with the inherent hemodynamic delay of the BOLD signal. A novel computational method was developed to separate avalanche type fast activity spreads from motion artifacts, vasomotor fluctuations, and cardio-respiratory noise in human brain default mode network (DMN). A system overview is shown in Figure 11. Reproducible and classical DMN sources were identified using spatial ICA prior to advanced noise removal in order to assure that ICA converges to reproducible networks. Brain activity peaks were identified from parts of the DMN, and normalized MREG data around each peak were extracted individually to show dynamic avalanche type spreads as video clips within the DMN (Figure 12, http://youtu.be/nPGyk5p76eg). Individual activity spread video clips of specific parts of the DMN were then averaged over the group of subjects. The experiments show that the high BOLD values around the peaks are mostly spreading along the spatial pattern of the particular DMN segment detected with ICA. With also the spread size and lifetime resembling the expected power law distributions, this indicates that the detected peaks are parts of activity avalanches, starting from (or crossing) the DMN. Furthermore, the split, one-sided sub-networks of the DMN show different spread directions within the same DMN framework. The results open possibilities to follow up brain activity avalanches in the hope to understand more about the system wide properties of diseases related to DMN dysfunction.

Figure 11. Steps of the activity avalanche detection system.

 

Figure 12. Single activity avalanches for DMN ICA components with vertical time flow (top to bottom) in medial prefrontal cortex and posterior cingulate cortex. ICA components on top (z-score), corresponding avalanche group mean frames around avalanche detection points on under them (normalized BOLD).

 

Biomedical Data Security

The practice of medicine is increasingly evolving towards personalised systems. Development of telemedicine and remote health monitoring is expected to increase cost-effectiveness of healthcare. However, the application of communication technologies leads to various privacy issues. Personal medical information is highly sensitive and needs to be protected against privacy and security threats. Security technologies to counter these threats ranging from cryptography to digital watermarking have been studied by our group.

Digitalization has also entered the health area in a wider sense. Big data collections for research and commercial purposes are increasingly becoming available. Confidentiality of medical data as well as an efficient information management is essential. During year 2015 the team has been working on data management issues of large collections of medical data, their privacy issues and the development of methods for watermarking medical data. We also continued our work on the resilience of hidden information in physical printouts.

Cryptographic Techniques

Remote monitoring of health using portable devices leads to challenges related to the confidentiality, integrity and availability of the data. Cryptographic techniques are used to guarantee these properties even under attacks by malicious adversaries. The communication technologies related to remote health monitoring can be divided into three tiers shown in Figure 13. Each of these tiers has its own set of security threats.

Figure 13. The communication tiers of a remote health monitoring system. Each tier faces its own characteristic set of security threats.

Our research has focused on novel light weight cryptographic primitives and protocols related to remote health monitoring. We have also turned our focus to the novel application of biometrics for the implementation of security mechanisms as a part of the Platform technology for Affordable, Continuous Health Monitoring – VitalSens (TEKES) project.

The rise of biobanks together with modern technologies has enabled the analysis of large collections of medical data. Biobanks are becoming a more common way to store big data collections of samples taken from humans with the appropriate consent from the donor and relevant health data. Scientists are able to conduct large-scale investigations based on the clinical status of patients and the related biomedical samples. Growing interest for the usage of biobanks in both research and commercial health applications has been estimated. The STM (Ministry of Health and Social Affairs) has initialized a national work on describing the overall architecture for biobanks in Finland. The purpose is to describe the overall operational environment as well as to define requirements for efficient operation. In the overall biobank framework there exists a definite need for finding new technological solutions to support the management of data as well to ensure security and privacy.

In INKA Digital Biobanks project, started in 2015, the aim is to design an architecture supporting biobank operation.  The work is done in close relation with other established biobanks in Finland, especially Northern Finland Biobank Borealis, through parallel projects as well as through BBMRI (Biobanking and Biomolecular Resources Research Infrastructure) work. In the ongoing project the accompanying network of biobanks have purchased whole slide scanners, which are to be integrated to the overall operation workflow in the hospitals. The scanners are utilized to digitalize pathology samples. Digitalized information along with other sample information, and information collected from hospital information systems are managed to obtain a rich database of information for research and business (Figure 14). The task of maintaining both the privacy of the subjects and the usefulness of the data is not a trivial one. Our team has studied the application of cryptography for patient data anonymization. Moreover, the samples of pathology provide a challenging field of study for pattern recognition and machine vision, where several use cases have been defined together with the Northern Finland Biobank Borealis, PPSHP personnel and parallel project network.

Figure 14. Information flow in the biobank infrastructure.

 

Digital Watermarking

Digital watermarking is a method of embedding a hidden sequence of bits in the host media in such a way that it is hard to perceive or remove. It is later possible to read the hidden information with a computer. To be able to read this hidden information with a camera phone is not yet well studied but would open multiple application areas from games and other recreational applications to authentication and other security related applications.

Smartphones are changing how we interact with the world and spend our time. According to recent studies, most of the time on smartphones is used on social networking, communication and entertainment but the amount of time used on entertainment and gathering information is rising. We are just starting to discover how to use our devices in the most efficient ways.

One of the new methods in order to narrow the gap between the old analogue world and the new digital world are barcodes. In addition to games, barcodes can be seen in flight tickets and newspaper adverts, they are put on top of packages of bread in order to persuade users to participate in various competitions, and there are even some on tourist attractions working as a link to online guest books. People are getting accustomed to using barcodes and mobile services initiated by barcodes.

Watermarking relates to barcodes on the level of techniques. A simple form of a watermark could be viewed as an invisible barcode. The advantages of barcodes include high robustness, the barcode can be captured with a camera fairly carelessly and the barcode works. However, barcodes are eye-catching and easily removed a fact that is not desirable in all applications, especially if security is required.

Another rising field is the health technology the export of which has grown past telecommunication in Finland. 47% of high technology exports in Finland are now some form of health technology, including in vitro diagnostics, X-ray and imaging equipment as well as mobile applications for self-care. In many health care applications, a lot of data is being stored and transferred constantly. Especially medical documents on paper are in danger to get lost or damaged or even stolen. There is a need for strict security requirements as data must be protected, e.g., against attacks towards integrity and tampering.

From an early stage, watermarking has been designed to answer some of these demands. In watermarking, data is hidden, i.e., embedded, in the content by making very small changes in such a manner that a human cannot detect these changes but a computer can read the hidden information, resembling effectively invisible barcodes. Depending on application, the watermark can be designed such that it is fully removable so that the original data does not get disturbed in any way, or that the data is meant to be broken if content is changed in anyway. With watermarking, the additional information is carried within the content discreetly. The technology is also highly transparent, meaning that it does not conflict with other security measurements, nor does it need any specific equipment of embedding or extraction as is the case with, for example, RFID tags.

In addition to health care applications, in many commercial and recreational applications aesthetics play an important role and visible barcodes may not be desirable due to visibility constraints or confidentiality requirements.  In these applications, the data embedded can be used for example in copyright protection and authentication, to provide links to net services, in games that feature searching for hidden objects, and in augmented reality applications. Image watermarking can thus be used as a connecting step between printed materials and camera phones as in Figure 15.

Figure 15. The watermark, here resembling a barcode, is embedded in image which is then printed. The user can read the watermark with a camera phone, and the content is shown according to the watermark information.

While reading watermarks from a printed image with a camera phone, the watermark should be robust against many kinds of attacks and distortions. These attacks contain general signal processing techniques such as compression and scaling but also the distortions due to capturing the two-dimensional image with the camera in three dimensions. In the captured image, the watermark will be rotated, scaled, translated, and tilted so that the synchronization of the watermark might be lost and thus the watermark cannot be extracted properly. The user interaction means that the image may be out of focus or, for example, paper quality and alignment cause variance to the watermark process.

Most of the work about digital watermarking in the literature has focused on digital copy- and copyright protection methods for the images moving electrically in the Internet. Some research has been made on print-scan robust watermarking systems where the digital image is printed on a paper and the watermark then read with a scanner. Only few papers discuss methods to read watermarks with a digital camera or a camera phone from printed images, the process called here as the print-cam process, which is our research focus area. 

In our previous publications we developed new print-cam robust methods, compared them to the old methods, researched new formats for images and limits for camera phone implementation. In 2014 it was shown that the intended application area of the watermarking method is highly important for selecting which methods to be used. On the other hand, many of the methods are no longer restricted by phone specifications.

For high usability, the designed methods should be fast and robust. The latest research concerned increasing robustness with the aid of computer vision techniques. Especially some problems inherent to print-cam process, such as camera lens focusing problems will be addressed. The auto-focus of a camera does not always work perfectly and sometimes lens properties, such as depth of field, make focusing impossible.

The work proposing a solution to these problems is sent for review in a journal at the beginning of 2016 and another publication is being written.

Similar methods are used for data hiding in medical images and signals. Medical data need special care when data hiding methods are applied. Data hiding methods embed a payload into the host data by introducing small imperceptible modifications. The payload could be, Electronic Patients Records, metadata, IDs, or multimodal data.  For medical purposes the challenge is to be able to reverse these modifications after extracting hidden data. The reason for the reversibility requirement is that a medical diagnosis cannot be compromised by even the smallest distortions. Distortions might even cause a false diagnosis with life threatening consequences.

First, the purpose of data hiding applications in biomedical data is to improve data management efficiency by saving space and time, combining everything is a single package. Second, hidden data acts a second layer of security by enriching data with metadata. This assures the maintenance of confidentiality, availability and reliability. Access to hidden data can be restricted only to entitled users, ensuring confidentiality. Then, as data co-exist in a single package, availability is also guaranteed. Last, reliability is improved as hiding IDs or hashes along hidden data can be used to prove ownership, authenticity or integrity.

In the first implementation that we developed during 2014 a reversible data hiding method was proposed were header information was hidden in the structures of MRI and X-Ray images using Least Significant Bit (LSB) substitution techniques. During 2015 a new technique was developed where Electroencephalography (EEG) signals are reversibly hidden in MR-Encephalography (MREG) sequences maintaining temporal synchronization (Figure 16) which is a novel feature. MREG sequences are similar to fMRI enabling even faster and more sensitive monitoring of functional activation of the brain sampled every 25-100 msec. Simultaneous MREG and EEG recordings are vastly used in neurobiology, being complementary data, as MREG provides good spatial resolution and EEG good temporal resolution. So far though, they were stored and handled as separate files.

Figure 16. From an MREG segment over time we are able to extract the EEG that was simultaneously sampled and also acquire the same MREG segment with any modifications removed.

This reversible data hiding method is based on histogram shifting. Originally, histogram shifting techniques were applied for 8-bit images. In this case, as we had to use 32-bit MREG as input we applied data quantization and generated histograms that can still provide sufficient embedding capacity. Let us note though, that quantization does not actually down-sample data, but in this case the down-sampled version of data is only used for the generation of the histogram to be exploited in the data hiding scheme. Of course, there is a tradeoff because higher quantization means that capacity is increased but fidelity between the original MREG and the one containing hidden information is decreased.

Since the area outside the brain is usually extracted from any analysis in order to increase the speed and accuracy of brain analytics we also provide an option of extracting the imaged tissue and restricting data hiding only in that area which we call Region Interest (ROI). This results in loss of capacity but we are able to compensate from that loss by either using higher quantization or compressing the EEG to be used as payload. Figure 17 demonstrates one example showing an MREG segment over a frame where we use the whole MREG as host and two examples where we restrict to the ROI.

Figure 17. The first row, from left to right, shows the original MREG and the one containing 64 uncompressed EEG channels of hidden data within the entire MREG using quantization that produces data with 256 different intensity values. In the second row, 10 uncompressed EEG channels are hidden in the ROI using quantization producing 64 intensity values. In the last row, 64 uncompressed EEG channels are hidden in the ROI using quantization producing 8 intensity values

 

Musculoskeletal System Research

Current clinical diagnostic tools have only limited ability to assess fracture risk or early osteoarthritis (OA) at an individual level. Using a biomechanical approach and advanced image analysis, we have shown that structural parameters can discriminate femoral neck fractures from controls. Assessment of the trabecular structure using texture analysis of radiographs appears to be a promising method.

In collaboration with researchers from the University of Eastern Finland and Mikkeli Hospital, we have developed multimodal technology for the detection of early degenerative changes of articular cartilage. We have demonstrated quantitative magnetic resonance imaging (MRI), ultrasound, and spectral infrared spectroscopy as potential tools for improved clinical diagnostics. Novel quantitative MRI techniques have been developed and proven to be more sensitive to osteochondral degeneration when compared to established techniques. Furthermore, biomedical imaging techniques suitable for quantitative 3D imaging of osteochondral tissue samples are actively developed within the group. Specifically, micro/nano-CT and MRI based techniques are being investigated. These 3D techniques improve our understanding of the pathogenesis of early osteoarthritis.

In our collaborative study with University of Jyväskylä, we demonstrated that progressively implemented high-impact exercise increases bone mass in postmenopausal women with mild knee osteoarthritis. High-impact exercise even created enough stimuli for favorable effects on patellar cartilage quality, as investigated by advanced MRI techniques.

Falls and fall-related injuries are a serious problem in older people. The data obtained during our incremental set of studies suggest that automatic accelerometric fall detection systems might offer a tool for supporting independent living and improving safety among older people. Reasonable fall detection sensitivity and specificity have been reported in laboratory settings, but knowledge about the sensitivity and specificity of these systems in real-life conditions is still lacking. In our long-term field test with elderly subjects in Oulu and Luleå, we monitored a total of 15,500 h of real-life data from 16 older people, using an accelerometry-based sensor system with an implemented fall detection algorithm. We were able to detect 12 out of 15 real-life falls, having a sensitivity of 80.0%. The false alarm rate was 0.025 false alarms per hour, equating to 1 false fall alarm per 40 usage hours. These data suggest that automatic accelerometric fall detection systems might offer a tool for improving safety among older people, especially in frequent fallers.

Figure 18. Acceleration as a sum vector (solid line), and low-pass-filtered posture calculated from vertical acceleration (dashed line) from a real-life fall, resulting in an online fall alarm (right).

In our joint study with Kyoto University Hospital, we demonstrated the feasibility of an e-textile sensor vest measuring ECG signal. The traditional NISHIJIN textile technique is used to illustrate a wire of electric circuit pattern on e-textile by single conductive yarn. In our tests with 8 elderly persons, we obtained good quality signal in standing and lying positions, but reduced quality signal while walking.

Physical inactivity is an increasing challenge especially in young population. In the MOPO study we evaluated whether the use of an activity monitor providing feedback has an effect on physical activity (PA) in young men. A population-based sample of 276 conscription-aged men participated in a 3-month randomized controlled trial in Oulu in 2012. We obtained weekly PA data from 72 (53%) and 90 (65%) men in the intervention and control groups, respectively. Time spent in moderate-to-vigorous activity increased (p = 0.012) and sedentary behaviour decreased (p = 0.032) in the intervention group. We were able to demonstrate that a wrist-worn activity monitor providing feedback had a positive effect on PA and sedentary behaviour in young men. We also demonstrated that determinants of sedentary and non-sedentary lifestyles are multiple and partially overlapping. Recognizing individual patterns and underlying factors of the sedentary lifestyle is essential for future tailored health promotion and interventions.

Osteoarthritis Research in Vitro

Quantitative MRI of Articular Cartilage

Novel MRI biomarkers hold great promise for the early degenerative changes occurring in articular cartilage in diseases such as osteoarthritis. In this Infotech group, the development process of novel MRI methodologies ranges from use of various in vitro model systems to preclinical testing to in vivo implementation for studying human subjects and patient populations.

In the first part of the research novel MRI biomarkers were developed and validated using an ultra-high field experimental MRI system. Consequently, various established and novel MRI contrasts potentially sensitive for early osteoarthritic changes in cartilage were compared against histological and biomechanical reference techniques. Different tissue models were used: enzymatically degraded bovine articular cartilage, human osteochondral specimens and rabbit model of anterior cruciate ligaments transection. In all studies the novel rotating frame of reference relaxation employing adiabatic radio frequency preparation (adiabatic T1ρ and T2ρ relaxation) proved to be sensitive to early macromolecular changes of cartilage (Figure 19). as compared to established quantitative MRI parameters, such as T2 relaxation time mapping. The same finding was confirmed in studies employing human articular cartilage samples and enzymatically digested bovine cartilage specimens.

Figure 19. Adiabatic T1ρ relaxation time at 9.4 tesla field strength reveals early macromolecular changes in articular cartilage 4 months after anterior cruciate ligament transection (ACLT) as compared to control (CTRL).

Sweep Imaging with Fourier Transform (SWIFT) technique was implemented and validated for the evaluation of the cartilage- bone interface using an experimental MRI system. The SWIFT technique proved useful for evaluating the cartilage-bone interface in a horse model for spontaneous cartilage repair and human osteochondral specimens, as it provides detectable signal from calcified tissue structures. Conventional MRI sequences are unable to provide such signal. Furthermore, SWIFT technology enabled the quantiative measurements of articular cartilage T1 relaxation time, sensitive to tissue water content, and altered in OA.

Development of 3D Histopathological Grading for Osteoarthritis

We have developed in vitro micro-CT -based technique to detect collagen 3D distribution in articular cartilage (AC) using phosphotungstic acid (PTA) as the collagen stain (bonds covalently with collagen) (Figure 20). In addition to collagen micro-CT imaging, we optimised CA4+ staining protocol for imaging of proteoglycan 3D distribution within articular cartilage. We were the first one to show that information about both cartilage chondrons and GAG content of the extracellular matrix can be obtained from a single contrast-enhanced micro-CT scan. The used contrast agent CA4+ is clinically applicable and these results could provide further new applications for osteoarthritis research and clinical diagnostics.

Figure 20. Example of 3D reconstruction of a human osteochondral sample subjected to PTA labeling and micro-CT imaging. PTA produced strong contrast allowing 3D visualization of the superficial collagen degeneration.

A novel segmentation algorithm to characterize complex surface topology of artivular cartilage from PTA-micro-CT images was also developed. This is one essential goal for creating 3D histopathological grading for osteoarthritis. We filed a patent application related to this technique on November 2015.

We obtained beamtime from the Canadian Light Source (synchrotron micro-CT). The goal was to collect high resolution data from large PTA-stained osteochondral samples and build a prediction model based on this data that would reveal microstructural detail from clinical images from bone and cartilage and compare this to data obtained for same samples imaged with clinical cone-beam CT.  The outcome was that osteocyte morphology and subchondral plate vascularity, bone-calcified cartilage interface morphology, tidemark morphology, chondrocyte morphology and clustering in calcified cartilage was possible to distinguish (Figure 21). However, more data is needed for building the actual prediction model, and new beamtime from the Canadian Light Source will be applied for 2016.

Figure 21. Unpublished data from experiments on the BMIT 05ID-2 beamline (Canadian Light Source, Saskatoon, Canada). Visualization of osteocytes in subchondral bone and chondrocytes in calcified cartilage form synchrotron CT data (left). Preliminary results indicate strong correlation between the osteoarthritis histological grade (OARSI) and osteocyte volume fraction (right).

While micro-CT approach to image articular cartilage using contrast agents is very promising, we have observed that diffusion of agents into the cartilage is highly time-consuming (even up to 3 days with PTA). We, therefore, innovated to apply ultrasonic to enhance the penetration of contrast agents. This may shorten the time required for contrast agents to penetrate the full depth articular cartilage. This activity advances, not only the field of 3D histology, but delivery of drugs into the cartilage tissue.

Finally, in collaboration with the optoelectronics laboratory we also developed an optical clearing method to reduce light scattering within articular cartilage enhancing light penetration especially to deep cartilage layers and cartilage-bone interface. The developed approach enabled more reliable detection of arthroscopically inaccessible regions, including cartilage-bone boundary and subchondral bone.

Osteoarthritis Research in Vivo

Quantitative MRI of Osteoarthritis

Encouraged by the in vitro MRI results, adiabatic T1ρ and T2ρ relaxation were implemented on a clinical 3-tesla MRI system. The MRI methodology was succesfully implemented and sequence parameters validated for imaging the knee joint and both asymptomatic and symptomatic human subjects have been imaged in order (Figure 22).

Figure 22. T1ρ relaxation time shows an increasing trend with radiographical OA classificiation (KL-grade) in various joint compartments.

 

Ultrasonography of Osteoarthritis

We developed a novel atlas for assessment of knee osteophytes from non-invasive ultrasonography (Figure 23). Our results clearly showed that detection of knee osteophytes using the novel ultrasound atlas is as reproducible as reading conventional radiographs, and that ultrasonography is even more sensitive to detect knee osteophytes than conventional radiography.

Figure 23. The developed novel atlas for knee ultrasonography for knee osteophyte detection.

 

Conventional Radiography of Osteoarthritis

We developed new computational algorithms for quantitative analysis of subchondral bone structure and composition from plain radiographs (Figure 24). We were among the first one to report that bone density and structure in tibia from standard clinically available 2D radiographs are significantly correlated with true 3D microstructure of bone. The developed methods could enhance the clinical diagnostics of osteoarthritis already at the primary healhcare where conventional radiography devices are widely available.

Figure 24. Regions of interest (ROIs) for quantitative analysis of subchondral bone structure and composition. Fractal signature analysis (FSA) and local binary patterns (LBP) based methods were used in quantitative analyses.

 

Oulu Knee Osteoarthritis Study

Finally, we finished the data acquisition of large clinical osteoarthritis trial (Oulu Knee Osteoarthritis Study, OKOA, 160 subjects) which has been ongoing during 2013-2015. The primary motivation in this trial is to further clarify the role of non-invasive ultrasonography in knee osteoarthritis diagnostics. A subcohort of the OKOA subjects will be also used to validate the implemented MRI methodology against patient symptoms and clinical MRI findings. Data analysis from the trial started in 2015 and first results will be published in 2016.

Digital Humanities Research

Digital Humanities is an emerging research area seated firmly between traditional humanities scholarship and the engineering sciences. Research in the domain seeks to uncover new humanities knowledge and add insight to existing theories by utilizing advances made in computer-assisted analysis methods. Due to this unique multidisciplinary nature, the domain offers fruitful ground for developing and testing out many machine vision and signal analysis methods on real-world data and research questions.

Advanced signal analysis, digital media content analysis, intelligent utilization of metadata and novel information visualization techniques are at the core of our research, and have proven highly valuable in the solving of real-world research issues stemming from, for example, the fields of humanities, education and medicine.

Since 2005 we have collaborated closely with digital humanities researchers in Finland and abroad, most notably with the University of Georgia (USA), University of Glasgow (UK) and Åbo Akademi, all of which remain close partners to this day. Since 2007 the research group has included the renowned digital humanist, linguist and corpus analytics expert Professor William A. Kretzschmar of the University of Georgia (USA) as an active collaborator on a number of innovative Digital Humanities projects. Highlights of this collaboration include the GISCA simulation environment for big data analytics based on real-world linguistic survey data (Kretzschmar, 2014; Juuso, 2015) and the LICHEN framework for large-scale corpora, as exemplified by the adoption of the framework for the 2012 release of the Digital Archive of Southern Speech by the Linguistic Data Consortium (Kretzschmar, 2012). These projects were built on the strong technical expertise developed at Oulu and the equally vital domain knowledge of our international partners.

Linguistic Modelling based on Large-scale Survey Data

A persistent issue in linguistic research, and in dialect research in particular, has been how to make generalizations from survey data about where some dialect feature might be found. We have prepared a sophisticated cellular automaton (CA) for use with authentic, large-scale data collected for the Linguistic Atlas Project (www.lap.uga.edu) that can address the problems involved in this type of data visualization. The developed CA seeds the simulation using real informant data, plots the locations where survey data were elicited, and then through the application of rules creates an estimate of the spatial distributions of selected features. The CA supports a variety of rules addressing both the elicited linguistic features and the social metadata available for each informant.

In a paper published in the esteemed Journal of English Linguistics in 2015 we show that our CA can create regions based on real data with simple rules following a definite, logical procedure. The CA differs significantly from the traditional process of drawing isoglosses as it provides a rigorous, repeatable process that grows regions based on initial data positions instead of drawing divisions subjectively between them. Through comparison of corresponding CA and density estimation (DE) plots we have shown that the CA can reduce the uncertainty in drawing boundaries. In addition, our CA method identifies regions for dialect features in a way that recognizes the many features that are in use at the same time, sometimes in the same locations, and incorporates competition between features.

The research has produced a steady stream of publications including a JUFO2-rated journal paper in 2015 and is expected to ultimately result in a doctoral thesis. Elements of the developed simulation will be deployed on the Linguistic Atlas Project site, and evaluation of elements for a new linguistic collection distributed by the Linguistic Data Consortium is currently underway.


Figure 25. An example of the linguistic feature regions produced by the CA using a rule set requiring a set level of support for a feature variant in its local neighbourhood to survive and/or expand.

 

Digital Edition of a 17th Century Manuscript

Our research group provided the technical expertise in image analysis and metadata processing for a four-year project funded by the Academy of Finland (ended 3/2015). The ORATIONES project was jointly run by the University of Oulu and Åbo Akademi with development work taking place in Oulu. The project included international partners from the UK and the USA.

In the project, a 17th century handwritten manuscript composed of school plays in English, Greek and Latin was digitized, transcribed by expert scholars and turned into a state-of-the-art digital edition available online. The primary technical issues to solve included the design of the TEI-XML annotation scheme and tools, distortion correction and filtering of the obtained scans, alignment of transcripts and images and the design of the final user interfaces supporting the various modalities and palaeographic goals.

We are currently investigating the option of providing the developed software suite to other digital edition projects including a project under preparation with the Sámi Archives and the National Archives of Finland.

MORE – A Multimodal Observation System for Groups

The MORE system is designed for observation and machine-aided analysis of social interaction in real life situations, such as classroom teaching scenarios and business meetings. The system utilizes a multichannel approach to collect data whereby multiple streams of 11 data in a number of different modalities are obtained from each situation. Typically the system collects a 360-degree video and audio feed from multiple microphones set up in the space. The system includes an advanced server backend component that is capable of performing video processing, feature extraction and archiving operations on behalf of the user. The feature extraction services form a key part of the system and rely on advanced signal analysis techniques, such as speech processing, motion activity detection and facial expression recognition in order to speed up the analysis of large data sets.

The provided web interface weaves the multiple streams of information together, utilizes the extracted features as metadata on the audio and video data and lets the user dive into analysing the recorded events. The objective of the system is to facilitate easy navigation of multimodal data and enable the analysis of the recorded situations for the purposes of, for example, behavioural studies, teacher training and business development. A further unique feature of the system is its low setup overhead and high portability as the lightest MORE setup only requires a laptop computer and the selected set of sensors on site.

The MORE system has been deployed in real classroom observation cases and has a solid application potential within the fields of vocology (professional voice training), speech-language pathology (voice, speech and language assessment and therapy), medicine (physician-patient interaction), psychology (counselling and guidance), education (class-room management), forensics (interrogation and interviewing), and security research. The developed system consisting of both a hardware and a software component is presented in a paper published in the journal of Multimedia tools and applications (Springer).

We are currently investigating the possibility of applying the system and selected machine vision and signal analytics features to the measurement professional athletes in collaboration with a number of top-tier sports teams in Finland.

 

Exploitation of Results

The research team has collaboration with several health technology companies in subcontracting projects and in TEKES and EU projects, which offers an opportunity for commercial exploitation of the results. Three commercialization projects are on-going or in preparation. Oulu University Hospital and the City of Oulu as partners enable transfer of new knowledge in the clinical work and social and health care practices.

 

Future Goals

We aim at scientific breakthrough in several domains of our research: diagnostics of cardiovascular diseases, respiratory diseases, brain disorders, osteoarthritis and bone fractures. We develop improved clinically feasible tools for the identification of individuals with increased risk. Our goal is also to find optimal technologies for the assessment of physical activity in respect to different health outcomes, and technologies for improving safety and wellbeing of older citizens.

 

Personnel

professors

6

senior research fellows

1

postdoctoral researchers

2

doctoral students

11

other research staff

10

total

54

person years for research

43

 

External Funding

Source

EUR

Academy of Finland

404 000

Tekes

461 000

other domestic public

339 500

domestic private

111 000

international

193 000

total

1 508 500

 

Doctoral Theses

Hirvasniemi, Jukka (2015) Novel X-ray-based methods for diagnostics of osteoarthritis. Acta Universitatis Ouluensis, Medica D 1327.

Partala, Juha (2015) Algebraic methods for cryptographic key exchange. Acta Universitatis Ouluensis, Technica C 519.

Rautiainen, Jari (2015) Novel magnetic resonance imaging techniques for articular cartilage and subchondral bone: studies on MRI Relaxometry and short echo time imaging. Acta Universitatis Ouluensis, Medica D 1286.

Toljamo, Päivi (2015) Dual-energy digital radiography in the assessment of bone characteristics. Acta Universitatis Ouluensis, Medica D 1311.

Kallio-Pulkkinen, Soili (2015) Effect of display type and room illuminance in viewing digital dental radiography: Display performance in panoramic and intraoral radiography, Acta Universitatis Ouluensis, Medica D 1312.

 

Selected Publications

Casula V, Hirvasniemi J, Lehenkari P, Ojala R, Haapea M, Saarakkala S, Lammentausta E, Nieminen MT. Association between quantitative MRI and ICRS arthroscopic grading of articular cartilage, Knee Surg Sports Traumatol Arthrosc DOI:10.1007/s00167-014-3286-9.

Fylakis A, Keskinarkaus A, Kiviniemi V, Seppanen T (2015) Reversible blind data hiding for verifying integrity and authenticating MRI and X-Ray images. Proc. of the 9th International Symposium on Medical Information and Communication Technology (ISMICT), Kamakura, Japan, 2015, March 24-26, 185-189.

Gilman E, Keskinarkaus A, Tamminen S, Pirttikangas S, Röning J, Riekki J (2015) Personalised assistance for fuel-efficient driving. Transport. Res. Part C: Emerging Technologies, 58:681-705.

Gilstad, Heidi; Faxvaag, Arild; Nøhr, Christian; Villumsen, Sidsel; Reponen, Jarmo, Andreassen, Hege; Jervall, Lars; Pehrsson, Thomas; Kangas, Maarit; Harðardóttir, Gudrun Audur; Koch, Sabine; Hyppönen, Hannele (2015) Challenges of Comparing Medication eHealth Services in the Nordic Countries. International Conference on Global Health Challenges, pp. 33-38.

Hannila I, Lammentausta E, Tervonen O, Nieminen MT. The Repeatability of T2 Relaxation Time of Human Knee Articular Cartilage, Magnetic Resonance Materials in Physics, Biology and Medicine, MAGMA 28:547-53, 2015.

Hirvasniemi J, Thevenot J, Kokkonen HT, Finnilä MA, Venäläinen MS, Jämsä T, Korhonen RK, Töyräs J, Saarakkala S (2015) Correlation of Subchondral Bone Density and Struc-ture from Plain Radiographs with Micro Computed Tomogra-phy Ex Vivo. Annals of Biomedical Engineering (epub ahead of print).

Hyppönen, Hannele; Kangas, Maarit; Reponen, Jarmo; Nøhr, Christian; Villumsen, Sidsel; Koch, Sabine; Hardardottir, Gudrun Audur; Gilstad, Heidi; Jerlvall, Lars; Pehrsson, Thomas; Faxvaag, Arild; Andreassen, Hege; Brattheim, Berit; Vimarlund, Vimarlund; Kaipio, Johanna. Nordic eHealth Benchmarking : Status 2014. TemaNord 2015:539.

Jauho, Anna-Maiju; Pyky, Riitta; Ahola, Riikka; Kangas, Maarit; Virtanen, Paula; Korpelainen, Raija; Jämsä, Timo (2015) Effect of wrist-worn activity monitor feedback on physical activity behavior : A randomized controlled trial in Finnish young men. - Preventive Medicine Reports 2, 628-634.

Jelinek HF, Karmakar C, Kiviniemi AM, Hautala AJ, Tulppo MP, Mäkikallio TH, Huikuri HV, Khandoker AH, Palaniswami M (2015) Temporal dynamics of the circadian heart rate following low and high volume exercise training in sedentary male subjects. Eur J Appl Physiol, 115(10):2069-80.

Kangas, Maarit; Korpelainen, Raija; Vikman, Irene; Nyberg, Lars; Jämsä, Timo (2015) Sensitivity and False Alarm Rate of a Fall Sensor in Long-Term Fall Detection in the Elderly. - Gerontology 61 (1), 61-68.

Keränen, Niina S.; Hautala, Mia; Chimariya, Adesh; Kangas, Maarit; Jämsä, Timo; Kuroda, Tomohiro (2015) Validation of signal quality in NISHIJIN e-Textile ECG vest. Proceedings of the 12th International Conference on Ubiquitous Healthcare, pp. 93-96. Keskinarkaus A, Huttunen S, Siipo A, Holappa J, Laszlo M, Juuso I, Väyrynen E, Heikkilä J, Lehtihalmes M, Seppänen T, Laukka S (2015) MORE - a multimodal observation and analysis system for social interaction research. Multimedia Tools and Applications, 1-25.

Kiviniemi AM, Lepojärvi S, Kenttä TV, Junttila MJ, Perkiömäki JS, Piira OP, Ukkola O, Hautala AJ, Tulppo MP, Huikuri HV (2015) Exercise Capacity and Heart Rate Responses to Exercise as Predictors of Short-Term Outcome Among Patients With Stable Coronary Artery Disease. Am J Cardiol, 116(10):1495-501.

Kiviniemi AM, Tulppo MP, Eskelinen JJ, Savolainen AM, Kapanen J, Heinonen IH, Hautala AJ, Hannukainen JC, Kalliokoski KK (2015) Autonomic Function Predicts Fitness Response to Short-Term High-Intensity Interval Training. Int J Sports Med, 36(11):915-21.

Koli, Jarmo; Multanen, Juhani; Kujala, Urho M.; Häkkinen, Arja; Nieminen, Miika T.; Kautiainen, Hannu; Lammentausta, Eveliina; Jämsä, Timo; Ahola, Riikka; Selänne, Harri; Kiviran-ta, Ilkka; Heinonen, Ari (2015) Effects of Exercise on Patellar Cartilage in Women with Mild Knee Osteoarthritis. - Medicine and science in sports and exercise 47, 1767-1774.

Kortelainen J, Väyrynen E (2015) Assessing EEG slow wave activity during anesthesia using Hilbert-Huang Transform. Proc. of the 37th Annual International Conference of the IEEE EMBS, Proc. of the 37th Annual International Conference of the IEEE EMBS, Milan, Italy, 2015, August 25-29, 117-20.

Kortelainen J, Väyrynen E, Seppänen T (2015) High-frequency electroencephalographic activity in left temporal area is associated with pleasant emotion induced by video clips. Computational Intelligence and Neuroscience, Article ID 762769, 14 p.

Koski JM, Kamel A, Waris P, Waris V, Tarkiainen I, Karva-nen E, Szkudlarek M, Aydin SZ, Alasaarela E, Schmidt W, De Miguel E, Mandl P, Filippucci E, Ziswiler H, Terslev L, Áts K, Kurucz R, Naredo E, Balint P, Iagnocco A, Lepojärvi S, Elseoud A, Fouda M, Saarakkala S. Atlas-based knee os-teophyte assessment with ultrasonography and radiography: relationship to arthroscopic degeneration of articular cartilage. Scand J Rheumatol. 2015 Sep 1:1-7. [Epub ahead of print]

Luoto, Kim; Korpelainen, Raija; Röning, Juha; Ahola, Riikka; Enwald, Heidi; Hirvonen, Noora; Tuovinen, Lauri; Heikkinen, Hannu I (2015) Gamified persuasion: user experiences of online activation service. International Journal of Sociotech-nology and Knowledge Development 6(4) 1-17.

Multanen, J.; Heinonen, A.; Häkkinen, A.; Kautiainen, H.; Kujala, U. M.; Lammentausta, E.; Jämsä, T.; Kiviranta, I.; Nieminen, M. T. (2015) Bone and cartilage characteristics in postmenopausal women with mild knee radiographic osteoar-thritis and those without radiographic osteoarthritis. - Journal of musculoskeletal and neuronal interactions 15 (1), 69-77.

Nauha, Laura; Keränen, Niina S.; Kangas, Maarit; Ahola, Riikka; Jämsä, Timo; Reponen, Jarmo (2015) Selection and implementation of a real-world smart home for persons with memory disorder. Proceedings of the 12th International Con-ference on Ubiquitous Healthcare, pp. 83-86.

Nieminen, Heikki J.; Ylitalo, Tuomo; Suuronen, Jussi-Petteri; Rahunen, Krista; Salmi, Ari; Saarakkala, Simo; Serimaa, Ritva; Haeggstrom, Edward (2015) Delivering Agents Locally into Articular Cartilage by Intense MHz Ultrasound. Ultrasound in medicine and biology 41(8), 2259-2265.

Nieminen, H. J.; Ylitalo, T.; Karhula, S.; Suuronen, J. P.; Kauppinen, S.; Serimaa, R.; Haeggstrom, E.; Pritzker, K. P. H.; Valkealahti, M.; Lehenkari, P.; Finnilä, M.; Saarakkala, S. (2015) Determining collagen distribution in articular cartilage using contrast-enhanced micro-computed tomog-raphy. Osteo-arthritis and Cartilage 23(9), 1613-1621.

Nieminen, Petteri; Huitu, Otso; Henttonen, Heikki; Finnilä, Mikko A. J.; Voutilainen, Liina; Itämies, Juhani; Karja, Vesa; Saarela, Seppo; Halonen, Toivo; Aho, Jari; Mustonen, Anne-Mari (2015) Physiological condition of bank voles (Myodes glareolus) during the increase and decline phases of the popu-lation cycle. Comparative Biochemistry and Physiology. Part A: Molecular & Integrative Physiology 187: 141-149.

Nissi M, Lehto L, Corum C, Idiyatullin D, Ellermann J, Nieminen MT. Measurement of T1 relaxation time of oste-ochondral specimens using VFA-SWIFT. Magn Reson Med; DOI: 10.1002/mrm.25398.

Pyky, Riitta; Jauho, Anna-Maiju; Ahola, Riikka; Ikäheimo, Tiina M.; Koivumaa-Honkanen, Heli; Mäntysaari, Matti; Jä-msä, Timo; Korpelainen, Raija (2015) Profiles of seden-tary and non-sedentary young men - a population-based MOPO study. - BMC Public Health 15 (1), 1164.

Rajna Z, Kananen J, Keskinarkaus A, Seppänen T, Kiviniemi V (2015) Detection of short-term activity avalanches in human brain default mode network with ultrafast MR encephalography. Frontiers in Human Neuroscience, (448).

Rajna Z, Keskinarkaus A, Kiviniemi V, Seppänen T (2015) Speeding up the file access of large compressed NIfTI neuroimaging data. Proc. of the 37th Annual International Conference of the IEEE EMBS , Milan, Italy, 2015, August 25-29, 654-7.

Rautiainen, Jari; Nissi, Mikko J.; Salo, Elli-Noora; Tiitu, Virpi; Finnilä, Mikko A. J.; Aho, Olli-Matti; Saarakkala, Simo; Lehenkari, Petri; Ellermann, Jutta; Nieminen, Miika T (2015) Multiparametric MRI assessment of human articular cartilage degeneration: Correlation with quantitative histology and mechanical properties. Magnetic resonance in medicine 74(1), 249-259.

Seppänen TM, Alho O-P, Seppänen T (2015) Concomitant dynamic changes in autonomic nervous system function and nasal airflow resistance during allergen provocation. Proc. of the 37th Annual International Conference of the IEEE EMBS, Milan, Italy, 2015, August 25-29, 3339-42.

Seppänen TM, Kananen J, Noponen K, Alho O-P, Seppänen T (2015) Accurate measurement of respiratory airflow waveforms using depth data. Proc. of the 37th Annual International Conference of the IEEE EMBS, Milan, Italy, 2015, August 25-29, 7857-60.

Sheshadri S, Kortelainen J, Cutrone A, Bossi S, Micera S, Thakor N, Delgado-Martínez I, Yen S (2015) Classification of phases of hand grasp task by the extraction of miniature compound nerve action potentials (mCNAPs). Proc. of the 7th International IEEE EMBS Neural Engineering Conference , Montpellier, France, 2015, April 22-24, 593-6.

Thevenot, Jerome; Chen, Jie; Finnilä, Mikko; Nieminen, Mii-ka; Lehenkari, Petri; Saarakkala, Simo; Pietikäinen, Matti. Local binary patterns to evaluate trabecular bone structure from micro-CT data: Application to studies of human osteoarthritis (2015) Computer Vision: ECCV 2014 Workshops: Zurich, Proceedings, Part II, 63-79.

Villumsen, Sidsel; Harðardóttir, Guðrún Auður; Kangas, Maarit; Gilstad, Heidi; Brattheim, Berit Johanne; Reponen, Jarmo. Monitoring the Amount of Practical Use of eHealth on National Level by Use of Log Data: Lessons Learned. Studies in Health Technology and Informatics 218, 138-144.

Vipin A, Kortelainen J, Al-Nashash H, Chua S, Thow X, Manivannan J, Rusly A, Thakor N, Kerr C, All A (2015) Prolonged local hypothermia has no long-term adverse effect on the spinal cord. Therapeutic Hypothermia and Temperature Management, 5:152-62.

Villumsen, Sidsel; Harðardóttir, Guðrún Auður; Kangas, Maarit; Gilstad, Heidi; Brattheim, Berit Johanne; Reponen, Jarmo. Monitoring the Amount of Practical Use of eHealth on National Level by Use of Log Data: Lessons Learned. Studies in Health Technology and Informatics 218, 138-144.

Vihriälä, Erkki; Rinta-Paavola, Anneli; Sorvoja, Hannu; Jämsä, Timo; Myllylä, Risto (2015) Relationship between weight change and changes in 3D acceleration signals generated by walking. Journal of Mechanics in Medicine and Biology 15 (5), 1-15, 1550080.

Last updated: 22.6.2016