|
Lääketieteellisen fysiikan ja tekniikan yhdistys (LFTY)
Finnish Society for Medical Physics and Medical Engineering
In English
|
VII Finnish Medical Physics and Medical Engineering Day, 7.2.2008, University of Oulu
The seventh Finnish Medical Physics and Medical Engineering Day was held 7 February 2008 at University of Oulu, Oulu,
Finland. The annual event gathered this year 136 student and researcher participants and seven Finnish companies from the
area of medical physics and medical engineering. The traditional poster exhibition, where the best Master's theses and
diploma theses finished in 2007 were awarded had 23 participants. The award for the best theses was this year 2007 €,
which was donated by
Polar Electro Ltd..
From the 23 participants, three theses were reckoned to be above the others. These three theses were:
Each of the three participants was rewarded with a 569 € scholarship. In addition, the jury rewarded two other
theses with 150 € scholarships. The two theses were:
![](lftp08_palkitut.jpg) |
Award presentation for the best Master's theses and diploma work in 2007
(from left to right): Prof. Pasi Karjalainen (chairman of Finnish Medical Physics and Medical Engineering
Society), Maarit Kangas, Matti Varanka, Mikko Nissi, Niina Nöjd, Janne Karjalainen, and Hannu Kinnunen (Polar Electro). |
Abstracts of the awarded theses
Methodology for Use in Ultrasound Backscatter-Based Osteoporosis Diagnostics
Karjalainen Janne, University of Oulu
Introduction: Osteoporosis is a disease characterized by loss of bone mass and a deterioration of tissue structure,
exposing individuals with this condition to increased risk of fractures. The initial changes in density and structure are
generally understood to occur in trabecular bone. However, the amount of compact bone is also severely diminished in
osteoporosis and may be detected as thinning of cortical layer in the shafts of long bones. The goal of this thesis was
to assess the applicability of novel ultrasound techniques for use in diagnostics of osteoporosis. The main aim (I.) of
this thesis work was to validate a novel dual frequency ultrasound (DFUS) technique for combined determination of soft
tissue composition and underlying bone properties by means of an in vivo case study. The DFUS method was further
developed towards use of single broadband ultrasound transducer. The second aim (II.) was to design a simple ultrasonic
in vivo methodology for determination of the thickness of the cortical bone layer.
Materials and methods:
I. In the first part of this study, the DFUS technique was applied to monitor the changes
in soft tissue composition and correction of the errors induced by soft tissues on the measurements of integrated
reflection coefficient (IRC). Measurements were conducted on a volunteer body builder during a 21 week training and
dieting period, inducing a weight loss of 16.5 kg (18%). For reference, dual energy X-ray absorptiometry (DXA)
measurements were conducted to follow the changes in soft tissue composition.
II. In the second part of this study, signal processing techniques for determination of cortical bone thickness were
first validated in vitro with six bovine samples. The in vivo measurements were conducted on 20 healthy volunteers. The
cortical thicknesses were determined from ultrasound signal with envelope and cepstral methods and compared to the
peripheral quantitative computed tomography (pQCT) and caliper measurements in vivo and in vitro, respectively. The
accuracy and precision of the ultrasound techniques were assessed.
Results:
I. The DXA measurements indicated that significant changes in quantity and composition of soft tissue,
but not in bone density, took place during the diet. As compared with DXA, the single transducer DFUS could determine
local soft tissue composition (r^2 = 0.88, n=8, p < 0.01). The change in uncorrected IRC associated significantly with the
change in body composition (r^2 = 0.56, n=8, p < 0.05). The IRC values, corrected by DFUS, showed only minor variation (SD
= ±1.26 dB) during the diet.
II. Values of cortical thickness, as determined with both techniques, showed high linear
correlations (r >= 0.95) with the thickness values obtained from in vitro measurements with a caliper or in vivo
measurements by pQCT. No systematic errors that could be related to cortical thickness were found. The in vivo accuracy
of the measurements was 6.6% and 7.0% for envelope and cepstral methods, respectively. Further, the in vivo precision for
the envelope and cepstral methods were 0.26 mm and 0.28 mm, respectively.
Conclusions: Both of these studies were conducted with the main aim of developing a new approach for pulse-echo
based ultrasound diagnostics of osteoporosis. The novel ultrasound techniques were further developed and validated with
two separate in vivo studies. These techniques may enable reliable assessment of both, trabecular and cortical bone
properties in addition to the determination of the composition of the soft tissue. As verified by this study, simple
ultrasound measurement method represents an accurate and feasible means for clinical estimation of cortical bone
thickness. Moreover, the DFUS method showed clear potential in the determination of soft tissue composition and therefore
could provide a solution for correction of the ultrasound parameters. A combination of these ultrasound techniques could
provide a novel approach for multi-site osteoporosis diagnostics using pulse-echo ultrasound. Overall, the techniques
presented in this thesis may have significant clinical value. The data gathered during this study will be published in
two international peer reviewed papers.
The Effect of Collagen Network on T2 Relaxation Time in Articular Cartilage
Nissi Mikko, University of Kuopio
Osteoarthritis is a severe joint disease that has high prevalence, especially in elderly people. Osteoarthritis causes
pain in it's end stages and finally disables the patient. Current methods for the diagnosis of osteoarthritis, especially
in it's initial stages, are rather insensitive . Earliest possible diagnosis would enable more efficient treatment and
thus the development of new methods is critically important.
In the present study, properties of articular cartilage were investigated using magnetic resonance imaging
(MRI). Parameters obtained from MRI, describing the properties of articular cartilage were compared to corresponding
parameters from polarized light microscopy (PLM). PLM is a very accurate reference method for studying optical
properties, such as fibril orientation of a given sample. However, due to it's invasive nature, PLM is not suitable for
clinical use.
Articular cartilage with different structural properties and varying maturational level was investigated in the present
study. Samples were harvested from human (n = 11), bovine (n = 12) and porcine (n = 11) patellae. A clear connection
between T2 relaxation time and structural properties of articular cartilage was observed. Three different structural
types of cartilage were equally detected by both methods: samples exhibited three, five or seven structural zones. The
locations of the boundaries of these histological zones, as determined from T2 measurements or PLM measurements, showed a
significant correlation (r = 0.952). Differences between species were studied in samples that had similar
structure. Differences were observed mostly between human and animal tissue, although the orientation angle of collagen
fibrils showed marked differences between bovine and porcine samples.
According to the results of this study, T2 relaxation time is capable of detecting changes in the histological zones of
articular cartilage that are related to properties and maturation of the tissue. As T2 relaxation time is sensitive to
the structural properties of articular cartilage, it is also a potential method for early diagnosis of osteoarthritis.
Optimal Electrode Positions for Facial EMG and EOG Measurements
Nöjd Niina, Tampere University of Technology
There are people suffering from different motor disabilities and thus, they are unable to use the normal control devices
of the computer. That is why there is a need for assistive devices, which make the use of the computer possible without
fine motorik or hands. We have constructed a wireless head cap that enables the measurements of facial muscle activations
and movements of the eyes. Our ultimate goal is to develop the measurement system so that it would suit in the control of
a computer interface: the gaze direction could move the cursor with some facial expressions to correspond clicking. The
wireless data transmission of the head cap restricts the number of measurement channels to six. To be able to get a good
signal quality with only a few electrodes the electrode positions on the forehead should be planned carefully. The main
purpose of this work is to use modeling in design of the optimal electrode positions for the measurements of the eye
movements (electro-oculography, EOG) and the muscle activations (electromyography, EMG).
In our work a realistic volume conductor model of the head was used. The model is maybe the most accurate used in the
world so far. Medical images such as CT slices and anatomical cryosection images from Visible Woman -project were used in
the segmentation. The software made in our institute was used to construct a finite difference method -model (FDM) of the
anatomical model. Calculations were based on reciprocity theorem and lead field concept. We modeled the activations of
frontalis and corrugator muscles and altogether 49 different gaze directions. One muscle or a specific gaze direction was
used as a source at a time and surface EMG or EOG were simulated. The optimal electrode positions were selected to be
those on the forehead that gave the strongest potential values and were the most sensitive for small changes of the angle
of gaze. A variety of different inverse solutions were presented and considered as to how they would suit the localizing
of the active muscle and the detection of the gaze direction. The use of the least squares fitting method in the inverse
solution was tested. Some noise was added to the simulated surface potentials and the effect of it on the localization
accuracy was considered. The inverse solution was also used to define the gaze directions, and it was estimated how the
errors in the gaze direction defining affect the position of the cursor.
The optimal positions for the electrodes were defined. For the EMG measurements it resulted that the best measurement
sensitivity is achieved by placing the electrodes parallel to the muscle fibers of the measured muscle. Anyway, better
separating capability for frontalis and corrugator activation is achieved by placing the electrodes more orthogonally
i.e., the electrode pair measuring frontalis lies orthogonally to the electrode pair measuring corrugators. Optimal
electrode positions for EOG measurements were similar to those found in the literature. The exception was the electrodes
measuring the horizontal eye movement: the head cap doesn't enable the use of electrode under the eye and thus new
positions for horizontal EOG measurements were found. The least squares fitting method did not work well when more than
one dipole source was localized. The method was too slow. Anyhow with the method the gaze direction was defined
moderately: The error of the position of the cursor was about 4 cm.
Next we should make real measurements to make certain that the obtained electrode positions really provide a better
signal quality. Also the source models for the EOG and EMG will be better validated using real measurements. EMG and EOG
models that are based on anatomical realistic model do not exist in the world that much. A new accurate model is now in
our use for modelling purposes such as bioelectric field problems. The model has over 160 million elements and exceeds
the standard computer resources.
Fall Detection with Body Attached Accelerometers
Kangas Maarit, University of Oulu
Fall related injuries are a central problem for elderly people. Elderly desire to live at home, and thus, new
technologies, such as automated fall detectors, are needed to support their independence and security. The aim of this
study was to evaluate different low-complexity fall detection algorithms, using triaxial accelerometers attached at the
waist, wrist, and head. The fall data were obtained from intentional fall in three middle-aged subjects and data from
activities of daily living were used as reference. Three different detection algorithms with increasing complexity were
investigated using two or more of the following phases of a fall event: beginning of the fall, falling velocity, fall
impact, and posture after the fall. The results indicated that fall detection using a triaxial accelerometer worn at the
waist or head is efficient, even with quite simple threshold-based algorithms, with a sensitivity of 97-98% and
specificity of 100%. In this study, the wrist did not appear to be an applicable site for fall detection. Since a head
worn device includes limitations concerning usability, a waist worn accelerometer, using an algorithm that recognizes the
impact and the posture after the fall, might be optimal for fall detection.
Estimating Running Performance from Physiological Data
Varanka Matti, University of Oulu
Running performance can be assessed in a laboratory environment by running an exhausting maximal treadmill test while
wearing a gas analyser mask. This may not be a convenient method for a hobbyist runner, since he may not have access to
costly treadmill testing. Because of this, simple submaximal measurement methods are needed which can assess the daily
running performance reliably and do not depend on the laboratory environment.
In this work, a multiple linear regression model is created which can estimate the daily running performance as an
endurance time. Endurance time is defined as the time until exhaustion when the subject runs on the treadmill whose
speed is increased by half a kilometer per hour each minute. The initial speed is eight kilometers per hour. This test
measures the aerobic system, running economy, and neuromuscular properties. The model uses the Running Index developed by
Polar Electro Oy and mean RR interval while standing as the predictor variables.
The model can estimate the daily fitness level in this data set with a mean estimation error of ±6.5%, and the standard
estimate of the error is 77 seconds. The data contained measurements from 27 subjects whose endurance time varied from
665 to 1260 seconds. As a second step, the created model was used to estimate the daily running performances in a four
week training intervention. The performance of the model remained constant when the estimated values were compared to
interpolated endurance times. Also, in the majority of cases (18 of 27), the estimated fitness trends were similar to
the interpolated endurance times.
The model created can estimate the daily running performance fairly reliably by using variables obtained with a heart
rate meter and a running speed sensor. Also, long-term changes of the running performance can be assessed by fitting
trendlines to the daily fitness estimates.