Infotech Oulu Annual Report 2014 - Biomedical Engineering Research Group (BME)

Professor Tapio Seppänen, Department of Computer Science and Engineering, and Professor Timo Jämsä, Research Center for 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

Cardiovascular and Respiratory System Research

Biobank ECG Digitalization

Cardiovascular diseases (CVDs) are the leading cause of death globally, and sudden cardiac death (SCD) is the most common mode of death in western societies. SCD affects 400 000 persons annually in Europe, causing more deaths than AIDS, lung and breast cancer and stroke together. Similar SCD figures have been estimated for the USA as well.

In Europe, more than half of all the cardiovascular deaths are sudden and unexpected estimated to account for 1-2 deaths/1000 adults over the age of 35. A large majority of SCDs occurs as the first clinical expression of previously silent heart disease or among the subjects with a known coronary artery disease without an evidence of heart failure.

Even for high-risk subgroups within the populations suffering from SCD – like those with left ventricular ejection fraction less than 35 % after a myocardial infarction – the positive predictive accuracy of clinical phenotype for individuals remains less than desirable. Moreover, the burden of the CVDs leads to 17-53/1 000 000/year EMS-treated cardiac arrest cases varying regionally.

To have an impact on these large public health problems, better individual risk prediction methods and risk stratification strategies are needed. Longitudinal medical studies spanning over decades can shed important light on the dynamics of CVDs and SCD, and enable development of early stage risk markers. A fact so important and congruent with other diseases as well that archives of data are now being combined to large biobanks with their own specific legislation.

For the aforementioned public health problems especially, distillation of the information contained in ECG measurements together with health information and outcomes over long time periods is expected to provide useful new tools and methods with quantifiable predictive power. The problem, however, is that much of the older material is only available in non-digital data formats.

The Biosignal processing team of the Biomedical Engineering Research Group has been developing an array of novel methods for ECG sheet digitalization suitable for the sheer amount of information contained in Biobanks. We have been working closely with the National Institute for Health and Welfare as well as cardiologists from both Northern Ostrobothnia Hospital District and Päijät-Häme social and health care group to develop highly automated tools that combine tools from research fields such as signal processing, machine vision and pattern recognition.

As our initial target, we have been digitizing the data of Mini-Finland health survey which is the first comprehensive nationwide study combining interview, questionnaire, and examination methods. Its baseline field work was carried out by the Social Insurance Institution’s Mobile Clinic Unit during 1978–1980 collecting a representative sample of 8 000 adults aged 30 years. In addition to a complete register-based follow-up, there exists a standard 12-lead ECG on four paper sheets and a longer vectorcardiogram (VCG) sheet for each person.

All the ECG and VCG material has been first scanned with the help of ScanningAid (SkannausApuri) software developed by the team to facilitate the scanning process with automated image and metadata quality checking. The resulting near four terabytes of image data are now being converted into digital signal formats using the innovative and highly automated ECGTraceTool program also developed by the team. The software has gathered international interest already.

Figure 1. Tracking of ECG plots.

 

Ambulatory ECG and VCG

In addition to the western view that was presented in the previous chapter, it should be noted that over 80 % of the CVD deaths occur in low- and middle-income countries, where chronic diseases are often diagnosed at an advanced stage due to a lack of primary care capacity and infrastructure. Hence, the cost of treatment and rehabilitation can be prohibitive at the time of diagnosis.

However, advances in technology have resulted in the availability of low-cost portable devices that can be used to provide early-stage detection and intervention utilizing the existing infrastructure. To facilitate the departure from the clinical setting, new kinds of signal processing and analysis methods that can operate reliably in real-time or near real-time with non-professional users in ambulatory conditions are needed. The foreseen cost reductions and population health benefits drive forward this type of a model also in more developed countries but with differing application scenarios and requirements. We have joined our forces with key Finnish parties, and started a collaborative research project together with colleagues in India to explore easily configurable and affordable technological solutions to these issues.

When making ECG measurements outside laboratory conditions during mundane activities, the number of ECG electrodes should preferably be kept to minimum for the reasons of e.g. price, convenience and reliability. We have been studying methods to synthesize VCG-like signals from measurements with reduced number of leads. In this way, fewer electrodes than normally need to be used in measurements, but the signal can be reconstructed with high enough accuracy to perform standard analyses that have become the de facto tools of the trade due to the vast amounts of research behind them. In our research, we have been exploring approaches based on novel non-linear systems theory and it has shown good promise in our tests.

Respiratory System Research

Early diagnostics, prediction, and treatment of several diseases – such as sleep apnea, asthma, cardiac disorders, and obstructive pulmonary diseases – can benefit from respiratory information. Especially, indirect methods, such as deriving respiratory signals algorithmically from ECG signals, align well with more and more typical ambulatory measurement scenarios.

Recent trends in in this research field have included applications of decomposition techniques such as principal component analysis (PCA), and its nonlinear version, kernel PCA (KPCA), both of which have been used to derive respiration signals from single-lead ECGs.

Our team has built on top of the state of the art, and developed a new adapted version of independent component analysis algorithm called AICA to obtain respiration signals. Additionally, we have devised a better performing extended version, APCA, of the normal linear PCA technique that utilizes our novel best component selection algorithm. Our studies have also revealed that smoothing spline resampling and band pass filtering can improve the performance of many traditional methods of the literature as measured by correlation coefficient and magnitude squared coherence to a reference respiration signal measured using spirometer or respiratory effort belts.

Allergic rhinitis is a major global health problem since over 500 million people suffer from it around the world.  It causes, in addition to individual impacts, a substantial economic burden to societies. It is a systemic inflammatory condition which is inheritable and links with other illnesses like asthma. Therefore, specific programs and guidelines on the problem have been released, for example by World Health Organization and European Union. There is a lack of an objective measurement method producing a continuous, accurate and reliable measurement data about the dynamic changes in nasal function. We have presented a new method to assess nasal function that produces nasal airflow resistance as a continuous signal at any sampling frequency allowing for analysis of dynamic changes in the resistance. The method was used to study nasal responses of test subjects from birch pollen allergic group and control group during the nasal birch pollen provocation test.

Our measurement system acquires the flow signal from the calibrated respiratory effort belts and the pressure signal by using a thin nasopharyngeal catheter inserted transnasally into the nasopharynx. A calibration of the respiratory effort belt signals is commonly done with the spirometer signal as reference signal and by applying the method of multiple linear regression, which produces the estimate of the respiratory airflow. However, this method yields only crude estimates of the waveforms in the airflow signal. If the breathing style or body position changes from the calibration, predicted airflows get even worse. We have made an extension to the conventional multiple linear regression method so that: 1) it uses a number of N consecutive signal samples and linear filtering for each prediction, and 2) it is based on polynomial regression to model different transfer functions between the input and output. This method is robust to breathing style changes and body position changes improving greatly the accuracy of the calibration of respiratory effort belts over the conventional method. In this study, only linear terms were used, because they provided sufficient performance to our respiratory effort belts based on our previous study. Figure 2 depicts a block diagram of the used method as a MISO (multiple input, single output) system consisting of FIR filters and a delay element.

Figure 2. Extended linear model.

 

Commonly, the resistance R is defined by Ohm’s law as R = P / F, where P and F refer pressure and flow, respectively. The Broms model offers parametric means to describe the nonlinear pressure / flow relationship, see Figure 3. However, the Broms model expects the data to be stationary, while the nasal system is non-stationary. We made an extension to the Broms model so that it can be used for calculating a continuous nasal airflow resistance value through model adaptation to varying signal statistics. Hence, instantaneous resistance values can be calculated over the whole measurement data and shown as dynamic plots over time.

Figure 3. Diagram of Broms model.

 

Ten birch pollen allergic and eleven non-allergic volunteers were challenged with control solution and allergen solution. Control solution was used before the actual allergen solution to rule out the non-specific nasal hyper-reactivity, which is usual procedure in provocation tests. Pressure and respiratory effort belt signals were recorded 10 minutes in baseline (Phase 1), 5 minutes after the control challenge (Phase2) and 20 minutes after the birch pollen challenge (Phase 3). At first, the respiratory effort belt signals were calibrated from the 1 min calibration recording. Then, continuous nasal airflow resistance signals were computed from the pressure and calibrated respiratory effort belt signals.

Continuous nasal airflow resistance signals were analyzed for the dynamic changes in the nasal airflow resistance. Examples of the continuous signals can be seen in Figure 4. Small gaps in the resistance signals can be seen in the figures due to removing of artifacts during manual validation. Artifacts can originate from example from sneezing, snuffling, mouth opening and moving. In Figure 4, the top and middle resistance curves are from birch pollen allergic subjects and the bottom resistance curve is from non-allergic subject. The resistance curve at the top in Figure 4 shows a non-specific response after the challenge with the control solution (Phase 2). After the reaction in the beginning, the resistance curve returns approximately to the baseline level (Phase 1). After the birch pollen challenge (Phase 3), a significant allergic reaction can be seen. The resistance curve rises immediately after the birch pollen challenge and continues rising still about 8 minutes before settling down to the much higher level than in baseline (Phase 1). The resistance curve in the middle in Figure 4 shows another kind of allergic reaction after the birch pollen challenge (Phase 3). The resistance starts to rise from the same level than in Phase 2 and the rise is much slower than in the previous case. The resistance rises about 11 minutes before settling down. The resistance curve at the bottom in Figure 4 is from non-allergic subject and it can be seen that there are no reactions to either of the challenge solutions.

Figure 4. Resistance curves for (top) allergic subject with a short non-specific response to the control solution and fast and significant reaction to the birch pollen challenge, (middle) allergic subject with a slower significant reaction to the birch pollen challenge, and (bottom) non-allergic subject with no reactions during measurement. Each pair of vertical bars between the phases marks the period of a challenge. Phase 1 = baseline. Phase 2 = after the control challenge. Phase 3 = after the birch pollen challenge.

 

The derived signals show in great detail the timing and intensity differences in subjects’ reactions. To our knowledge, this is the first time that it is possible to estimate accurately from the continuous nasal airflow resistance: (1) how strong and fast the allergic reaction occurs, (2) how long it takes the reaction to settle, and (3) whether short non-specific hyper-reactive reactions occur with test subjects. Quantitative results of resistance changes indicated that allergic and non-allergic subjects can be differentiated in a statistically significant degree using the this method. The method opens entirely new possibilities to research accurately the dynamic changes in non-stationary nasal function. It could increase the accuracy and reliability of diagnostics and assessment of the effect of nasal treatments.

Central Nervous System Research

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 clinical examinations (e.g. reflex testing), neurophysiological measurements (e.g. somatosensory evoked potential (SEP)), laboratory tests (e.g. S -NSE and S -100), and imagining modalities that, 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 EEG-based technology for detecting and measuring HIE. We have successfully carried out a pilot study with ten CA patients and designed a protocol for discriminating the subjects with good and poor neurological prognosis in the early phase of recovery.

Slow oscillations in direct current EEG (DC-EEG) background have been shown to reflect potential over blood brain barrier (BBB). Animal experiments have shown that the manipulation of BBB integrity could be monitored using DC-EEG. In collaboration with Oulu University Hospital, we monitored human subjects during intra-arterial mannitol induced BBB disruption (BBBD) using DC-EEG and synchronous brain near infra-red spectroscopy (NIRS). BBBD is used in humans clinically for therapy in primary central nervous system lymphoma due to its ability to increase drug penetrance into the brain. DC-EEG was monitored continuously with NIRS and anesthesia monitoring during clinical BBBD procedures in eight measurements carried out with two patients. EEG was recorded according to the 10-10 international system from the scalp to assess DC-EEG topography dynamically during mannitol infusion. NIRS was used for deoxy/oxyhemoglobin (Hb/HbO) monitoring from the forehead. During the BBBD, the subjects were in isoelectric anesthesia due to propofol and thiopental administration and mechanically ventilated. We detected robust DC-EEG potential shifts following intra-arterial mannitol infusion in all of the measurements. The shifts followed topographically the treated arterial territories of the brain and enabled quantitative mapping of the BBB opening. The study was the first one to show successful monitoring of the human BBBD in a clinical environment.

Somatosensory evoked potentials (SEPs) are widely used in the clinical environment and research for studying the functional integrity of the different parts of sensory pathways. However, most of the general anesthetics, such as the inhalational anesthetic isoflurane, are known to suppress SEPs, which affects the interpretation of these recordings. In animal studies, the usage of anesthetics during SEP measurements is, however, inevitable due to which one should know the detailed effect of these drugs on the signal. In collaboration with Singapore Institute for Neurotechnology (SINAPSE), National University of Singapore, we studied the effect of isoflurane on SEPs in a rat model. Time and frequency properties of the cortical SEP recordings generated by electrically stimulating the tibial nerve of rat’s hindlimb were investigated at three different isoflurane levels. While the anesthetic agent was shown to generally suppress the amplitude of the SEPs, the effect was found to be nonlinear affecting more substantially the latter part of signal waveform. The finding will help us in future in separating the effects of anesthetics on SEP from those due to injury in the ascending neural pathways.

Figure 5. Determination of somatosensory evoked potential (SEP) parameters from raw data.

 

Figure 6. The effect of anesthetic isoflurane on the time and frequency characteristics of the somatosensory evoked potentials (SEPs). The average SEP waveforms are given for five rats at light, moderate, and deep anesthesia. Mean spectrograms calculated from the average SEP waveforms are given as well.

 

Neuroprosthetic devices interfacing with the nervous system to restore functional motor activity offer a viable alternative to nerve regeneration. This is especially the case with proximal nerve like brachial plexus injuries where muscle atrophy potentially sets in before re-innervation occurs. So far, control signals from muscle or cortical activity have been used for the restoration of functional motor activity. However, nerve signals offer an interesting alternative for this as they permit more natural and precise control when compared to muscle activity. They also can be accessed with much lower risk than cortical activity. The usage of recordings from the nerve requires identification of the signals that control the appropriate muscles. In collaboration with Singapore Institute for Neurotechnology (SINAPSE), National University of Singapore, we investigated the correlation between muscle and nerve signals responsible for hand grasping in a non-human primate. Recordings were performed simultaneously using a 4-channel thin-film longitudinal intra-fascicular electrode (tf-LIFE) and nine bipolar endomysial muscle electrodes while the animal performed grasping movements. A high degree of correlation (r > 0.6) between nerve signals from the median nerve and movement-dependent muscle activity from the flexor muscles of the forearm was observed, with a delay that corresponded to 25 m/s nerve conduction velocity. By applying a wavelet approximation to the neural signal, the phase of the flexion could be identified. The results supported the usage of the proposed approach for future neuroprosthetic device for the treatment of peripheral nerve injuries.

Furthermore, using the above described data, we used waveform feature extraction and clustering techniques to identify a discrete set of events in the intraneural recordings. The approach allowed the classification of the different phases of hand grasping. The waveform features were found to significantly differ at each phase of grasping. These waveforms can be seen as the minimal signal components that result from the activation of a group of nerve fibers and they were thus denominated as miniature compound nerve action potentials (mCNAPs). In future, the correlation between mCNAPs and the different stages of movement can be utilized when designing high-performance neuroprosthetic therapies.

Figure 7. Miniature compound nerve action potential (mCNAP) clusters and their corresponding mean waveforms extracted from the thin-film longitudinal intra-fascicular electrode (tf-LIFE) recordings of the median nerve of a non-human primate.

 

Moreover, we have studied the properties of the critically sampled functional magnetic resonance imaging (fMRI), also known as magnetic resonance encephalography (MREG). The research focus was to explore the methods as well as improvements of existing tools, to detect and remove the effects of the physiological noise such as respiration and cardiac components. This will efficiently lead to more reliable analysis of brain functionality. We used technologies like group ICA (Figure 8) to separate the signal from vasomotor fluctuations, cardiac and respiratory physiological noise, in a manner that we are able to detect certain time points in blood oxygen level-dependent (BOLD) data which refer to neuronal activity peaks in selected parts of the DMN (Default Mode Network). We can assume that the detected peaks are with some notable probability parts of neuronal activity avalanches starting from (or crossing) the selected parts of the DMN (Figure 9). This opens a possibility to follow up neuronal activity avalanches in hope to understand more about the system wide properties of diseases related to DMN dysfunction.

Figure 8. Group ICA analysis, example components, posterior cingulate cortex (pcc), and ventro-medial prefrontal cortex (vmpf).

 

Figure 9. An example for single activity avalanche in component 64 (DMN pcc).

 

Digital Watermarking

The focus of our research was on digital watermarking under heavy distortions, watermarking biomedical signals and signal processing for functional MRI. We continued our work on the resilience of hidden information in physical printouts. Furthermore, the year 2014 was also a turn point in our watermarking research. The team has a long history on biosignal processing, and a natural shift was to explore the possibilities of applying watermarking techniques on medical data. Co-operation with the Department of Radiology in ELIXIR (European life science infrastructure for biological information) and Multimodal Scanning of Neuronal Avalanche Dynamics project (Academy of Finland) projects has deepened the signal level understanding of radiological data, especially MRI and functional MRI. Watermarking has a variety of applications, ranging from copyright protection to metadata embedding, from tamper detection to data recovery, from integrity checking to source identification. As confidentiality of medical data is essential, watermarking might have great advantages. As an example, in biobanks such as Northern Finland Birth Cohort Studies and biobank Borealis, there is an obvious need for managing security and confidentiality, where watermarking methods might be applicable.

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. 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. To be able to read the hidden information with a camera phone is not yet well studied but would open multiple different application areas from games and other recreational applications to authentication and other security related applications (Figure 10).

Figure 10. 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. Here an augmented reality application is launched in which the application launches a view according to the watermark, in this case, a frog.

 

In these application cases, 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. 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. Therefore, there is a need for strict security requirements as data must be protected, e.g., against attacks towards integrity and tampering.

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 choice of paper quality of physical location of the watermarked image causes variance to the watermark process.

It has been already shown that print-cam robust watermarking is possible and robust to common signal processing techniques and user input but before commercializing, the method needs to be extremely robust, unobtrusive to eye and fast to calculate even in a mobile device. In 2014, several previously proposed print-cam robust watermarking methods were implemented on a high end mobile phone and assessed. The experiments showed that intended application area of the watermarking method is highly important for selecting which methods to be used. On the other hand, many of our methods are no longer restricted by phone specifications.

In the near future, the research will focus on achieving faster and more robust methods with the aid of computer vision. Especially some problems inherent to print-cam process, such as camera lens focusing problems are addressed (Figure 11). 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 solving these problems is under way and the first results will be published soon.

Figure 11. Focusing of a camera may not always be perfect and unfocused areas make watermark extraction difficult. Here, left part of the printed image is unfocused while the rest is in focus.

 

Medical signals need special care when watermarking methods are applied. Data hiding is performed by introducing small imperceptible modifications in the host image, but for medical purposes the challenge is to be able to recover these modifications after extracting hidden data (Figure 12). The reason behind 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. Considering these requirements, we have developed a reversible data hiding technique for digital medical images. Practical experiments were realized using MRI and X-Ray images, but the method can also be extended for more general applicability.

Figure 12. Embedding procedure outputs the image containing data and extracting/reversing procedure outputs hidden data and a reversed image as close as possible to the original.

 

In our proposed technique, we use the header data of MRI images in NIFTI format and X-Ray images in DICOM format as the information piece to be hidden. Concerning the MRIs a cross-section of the 3D structure was used as the host (Figure 13). Data such as information about the origin is embedded into image structure itself providing inseparability, privacy and authentication capabilities.

The technique is based on the Least Significant Bit (LSB) scheme. Initially the host image is segmented into the Region of Interest (ROI) and Region of Non Interest (RONI). Next, data is embedded and hidden in LSBs of ROI pixels’ LSBs while backup data for reversibility purposes is embedded in RONI pixels’ LSBs. Since RONI does not have a supporting role in the diagnosis it can be safely used for imperceptible modifications restricting reversibility to the ROI. As for the sequence of pixels which carry hidden data, it can be defined by exploiting a Just Noticeable Distortion (JND) model along with LSB elimination to output the same model regardless the LSB modifications. RONI pixels are accessed in a consecutive or pseudo-random order. In the experiments with MRI and X-Ray images we proved excellent fidelity (Figure 14) and 100% reversibility of the ROI.

Figure 13. Data Hiding Scheme.

 

Figure 14. Pairs of original images (left) and the same images carrying hidden data (right).

 

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. We showed that textural analysis enables discrimination of patients at risk for femoral neck fracture, even with better prediction accuracy than bone mineral density (BMD). We also demonstrated that changes in bone texture in knee OA can be quantitatively evaluated from plain radiographs using advanced image analysis. We have developed computational finite element (FE) models and shown that FE analysis can yield reasonable accuracy in the assessment of experimental failure load. We presented a novel method to automatically create personalized 3D models from standard 2D radiographs. We also demonstrated that the novel axial transmission ultrasound technique can discriminate fracture subjects from controls with similar accuracy to that of DXA.

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 high-impact exercise increases bone mass, whereas no changes occurred in the biochemical composition of the cartilage, as investigated by advanced MRI techniques. In the study with prediabetic subjects, we showed that even light physical activity as determined by a wearable motion sensor decreases insulin resistance, improves lipid homeostasis and reduces visceral fat. The data obtained during the incremental set of studies suggest that automatic accelerometric fall detection systems might offer a tool for improving safety among older people.We also presented a wireless body area network (WBAN) utilizing the IEEE802.15.6 standard as applied to a multi-accelerometer system for monitoring Parkinson’s disease and fall detection.

 

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. 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

5

senior research fellows

1

postdoctoral researchers

8

doctoral students

17

other research staff

8

total

39

person years for research

31

 

External Funding

Source

EUR

Academy of Finland

404 000

Tekes

148 000

other domestic public

325 500

domestic private

80 000

international

193 000

total

1 150 500

 

Doctoral Theses

Finnilä M (2014) Bone toxicity of persistent organic pollutants. Acta Univ Oul. Medical D 1254.

Manninen AL (2014) Clinical applications of radiophotolu-minescence (RPL) dosimetry in evaluation of patient radiation exposure in radiology. Determination of absorbed and effective dose. Acta Univ Oul. Medica D 1265.

Väyrynen E (2014) Emotion recognition from speech using prosodic features. Acta Univ Oul. Technica C 487.

 

Selected Publications

Danso EK, Honkanen JTJ, Saarakkala S, Korhonen RK. Comparison of nonlinear mechanical properties of bovine articular cartilage and meniscus. J Biomech 47:200-6, 2014.

Hirvasniemi J, Immonen V, Thevenot J, Liikavainio T, Pulkkinen P, Jämsä T, Arokoski J, Saarakkala S (2014) Quantification of differences in bone texture from plain radiographs in knees with and without osteoarthritis. Osteoarthritis Cartilage 22(10): 1724-31

Herzig KH, Ahola R, Leppäluoto J, Jokelainen J, Jämsä T, Keinänen-Kiukaanniemi S (2014) Light physical activity determined by a motion sensor decreases insulin resistance, improves lipid homeostasis and reduces visceral fat in high-risk subjects: PreDiabEx study RCT. Int J Obesity 38(8):1089-96

Husso M, Sipola P, Kuittinen T, Manninen H, Vainio P, Hartikainen J, Saarakkala S, Töyräs J, Kuikka J. Assessment of myocardial perfusion with MRI using modified dual bolus method. Physiol Meas 35:533-47, 2014.

Huttu MRJ, Puhakka J, Mäkelä JTA, Takakubo Y, Tiitu V, Saarakkala S, Konttinen YT, Kiviranta I, Korhonen RK. Cell-tissue interactions in osteoarthritic human hip joint articular cartilage. Connect Tissue Res 55:282-91, 2014.

Hyppönen H, Reponen J, Lääveri T, Kaipio J. User experiences with different regional health information exchange systems in Finland. Int J Med Inform 83(1):1-18, 2014.

Jämsä T, Kangas M, Vikman I, Nyberg L, Korpelainen R (2014) Fall detection in the older people: From laboratory to real-life. Proceedings of the Estonian Academy of Sciences 63(3): 253-257

Kallio-Pulkkinen S, Haapea M, Liukkonen E, Huumonen S, Tervonen O, Nieminen MT. Comparison of consumer grade, tablet and 6MP-displays: Observer performance in detection of anatomical and pathological structures in panoramic radiographs. Oral Surg Oral Med Oral Pathol Oral Radiol 118:135-41, 2014.

Kiviniemi AM, Hintsala H, Hautala AJ, Ikäheimo TM, Jaakkola JJ, Tiinanen S, Seppänen T, Tulppo MP. Impact and management of physiological calibration in spectral analysis of blood pressure variability. Frontiers in Physiology, 5:473. (2014)

Kortelainen J, Vipin A, Thow X, Mir H, Thakor N, Al-Nashash H, All A (2014) Effect of isoflurane on somatosensory evoked potentials in a rat model. Proc. of the 36th Annual International Conference of the IEEE EMBS, Chicago, USA, 2014, August 26-30, 4286-9.

Kretzschmar W, Juuso I. Simulation of the complex system of speech interaction: digital visualizations. Lit Linguist Computing, 29(3):432-42. (2014)

Kretzschmar W, Juuso I, Bailey C. Computer simulation of dialect feature diffusion. Journal of Linguistic Geography, 2(1):41-57. (2014)

Kretzschmar W, Juuso I. Cellular automata for modeling language change. Lecture Notes in Computer Science, 8751:339-48. (2014)

Liukkonen J, Lehenkari P, Hirvasniemi J, Joukainen A, Virén T, Saarakkala S, Nieminen MT, Jurvelin JS, Töyräs J. Ultrasound Arthroscopy of Human Knee Cartilage and Subchondral Bone In vivo. Ultrasound in Medicine and Biology 40:2039-47, 2014.

Maksimow A, Silfverhuth M, Långsjö J, Kaskinoro K, Georgiadis S, Jääskeläinen S, Scheinin H. Directional connectivity between frontal and posterior brain regions is altered with increasing concentrations of propofol. PLoS One, 9(11):e113616. (2014)

Manninen A-L, Ojala K, Nieminen MT, Perälä J. Fetal Radiation Dose in Prophylactic Uterine Arterial Embolization. Cardiovasc Intervent Radiol 37:942-8, 2014.

Multanen J, Nieminen MT, Häkkinen A, Kujala U, Jämsä T, Kautiainen H, Lammentausta E, Ahola R, Selänne H, Ojala R, Kiviranta I, Heinonen A (2014) Effects of high-impact bone exercise on bone and articular cartilage: a 12 months randomized controlled quantitative magnetic resonance imaging study. J Bone Miner Res 29(1):192-201

Munukka M, Waller B, Multanen J, Rantalainen T, Häkkinen A, Nieminen MT, Lammentausta E, Kujala UM, Paloneva J, Kautiainen H, Kiviranta I, Heinonen A. Relationship between lower limb  neuromuscular performance and bone strength in postmenopausal women with mild knee osteoarthritis. J Musculoskelet Neuronal Interact 14:418-424, 2014.

Määttä M, Moilanen P, Timonen J, Pulkkinen P, Korpelainen R, Jämsä T (2014) Prediction of fractures using low-frequency ultrasound – comparison with DXA-based BMD. BMC Musculoskelet Disord 15(1):208

Pramila A. Benchmarking and Modernizing Print-Cam Robust Watermarking Methods on Modern Mobile Phones. M.Sc. thesis, Department of Information Processing Science, University of Oulu, Finland. 2014

Puhakka PH, Ylärinne JH, Lammi MJ, Saarakkala S, Tiitu V, Kröger H, Virén T, Jurvelin JS, Töyräs J. Dependence of light attenuation and backscattering on collagen concentration and chondrocyte density in agarose scaffolds. Phys Med Biol 59:6537-48, 2014.

Rautiainen J, Nissi MJ, Liimatainen T, Herzog W, Korhonen RK, Nieminen MT. Adiabatic Rotating Frame Relaxation of MRI Reveals Early Cartilage Degeneration in a Rabbit Model of Anterior Cruciate Ligament Transection. Osteoarthritis Cartilage 22:1444-52, 2014.

Rieppo L, Saarakkala S, Jurvelin JS, Rieppo J. Optimal variable selection for Fourier transform infrared spectroscopic analysis of articular cartilage composition. J Biomed Opt 19:027003, 2014.

Seppänen TM, Alho O-P, Seppänen T (2014) Dynamic Changes of Nasal Airflow Resistance during Provocation with Birch Pollen Allergen. Biomedical Systems and Technologies, Communications in Computer and Information Science 452:225-239.

Sheshadri S, Kortelainen J, Nag S, Ng K, Bazley F, Michoud F, Patil A, Orellana J, Libedinsky C, Lahiri A, Chan L, Chng K, Bossi S, Micera S, Thakor N, Delgado-Martinez I, Yen S (2014) Correlation between muscular and nerve signals responsible for hand grasping in non-human primates. Proc. of the 36th Annual International Conference of the IEEE EMBS, Chicago, USA, 2014, August 26-30, 2314-7.

Sheshadri S, Kortelainen J, Cutrone A, Bossi S, Micera S, Thakor N, Delgado-Martínez I, Yen S. 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. Accepted.

Siirtola P, Pyky R, Ahola R, Koskimäki H, Jämsä T, Korpelainen R, Röning J (2014) Detecting and profiling sedentary young men using machine learning algorithms. IEEE Symposium on Computational Intelligence and Data Mining (CIDM). IEEE Conference Proceedings, 296-303

Särestöniemi M, Niemelä V, Hämäläinen M, Keränen N, Jämsä T, Reponen J, Partala J, Seppänen T, Iinatti J (2014) Receiver performance evaluation on IEEE 802.15.6 based WBAN for monitoring Parkinson's disease. 8th International Symposium on Medical Information and Communication Technology (ISMICT), 2-4 April 2014, Florence, Italy.

Takalokastari T, Alasaarela E, Kinnunen M, Jämsä T (2014) Quality of wireless electro­cardiogram signal during physical exercise. IEEE J Biomed Health Inform 18(3):1058-64

Thevenot J, Hirvasniemi J, Pulkkinen P, Määttä M, Korpelainen R, Saarakkala S, Jämsä T (2014) Assessment of risk of femoral neck fracture with radiographic texture parameters: a retrospective study. Radiology 272(1):184-191

Thevenot J, Koivumäki J, Kuhn V, Eckstein F, Jämsä T A.novel method to generate a 3D finite element model of the hip from a 2D radiograph. J Biomech 47: 438-444 (2014)

Ye L, Ferdinando H, Seppänen T, Alasaarela, E. Physical violence detection for preventing school bullying. Advances in Artificial Intelligence, 2014:9 p. 1-9.

Last updated: 29.3.2016