A major health concern for subjects with diabetes is weaker bones and increased fracture risk. Current clinical assessment of the bone strength is performed by measuring Bone Mineral Density (BMD), where low BMD-values are associated with an increased risk of fracture. However, subjects with Type 2 Diabetes (T2D) have been shown to have normal or higher BMD-levels compared to healthy controls, which does not reflect the recognized bone fragility among diabetics. Thus, there is need for more research about diabetes-related bone fragility to find other factors of impaired bone health. One potential biomarker that has recently been studied is Bone Marrow Fat (BMF). The data in this project consisted of whole-body water-fat Magnetic Resonance Imaging (MRI) volumes from the UK Biobank Imaging study (UKBB). Each subject in this data has a water volume and a fat volume, allowing for a quantitative assessment of water and fat content in the body. To analyze and perform quantitative measurements of the bones specifically, a Deep Learning (DL) model was trained, validated, and tested for performing fully automated and objective skeleton segmentation, where six different bones were segmented: spine, femur, pelvis, scapula, clavicle and humerus. The model was trained and validated on 120 subjects with 6-fold cross-validation and tested on eight subjects. All ground-truth segmentations of the training and test data were generated using two semi-automatic pipelines. The model was evaluated for each bone separately as well as the overall skeleton segmentation and achieved varying accuracy, performing better on larger bones than on smaller ones. The final trained model was applied on a larger dataset of 9562 subjects (16% type 2 diabetics) and the BMF, as well as bone marrow volume (BMV) and cortical bone volume (CBV), were measured in the segmented bones of each subject. The results of the quantified biomarkers were compared between T2D and healthy subjects. The comparison revealed possible differences between healthy and diabetic subjects, suggesting a potential for new findings related to diabetes and associated bone fragility.

Concussions occur frequently as a result of tackles in ice hockey. Analysis of video material may provide an understanding of the relationship between the kinematics of collisions and the risk for injury. In this thesis, two video analysis methods were used to estimate the impact velocities of 22 ice hockey tackles that resulted in concussions. The Point tracking method uses tracking of user-defined object points on the players and ice to estimate the velocities. It was used in an earlier thesis. A deep learning-based method was implemented in this thesis. It uses a pre-trained deep learning model to detect the players in each frame of the video. Both methods were validated in this thesis using soccer videos containing accelerometer data from the players. The mean error was 25.6 % for the Point tracking method and 43.1 % for the Deep learning method. The difference was not significant. Both methods calculate the player velocity as a mean from a given number of video frames before impact. The choice of the number of frames did not significantly affect the difference in estimated velocities between the Point tracking method and the Deep learning method. The Point tracking method succeeded in estimating velocities in 17 cases. The mean velocities for the attacking and injured players were 10.5 m/s and 9.3 m/s, respectively. The Deep learning method succeeded in 9 cases, and the mean velocities were 9.7 m/s and 9.5 m/s. The velocities are higher than what has been found in earlier research, suggesting that both methods may be biased towards estimating too high velocities. More investigation needs to be done to evaluate the methods’ performance, possibly by comparing with accelerometer data from ice hockey.

Pancreatic ductal adenocarcinoma (PDAC) is one of the most difficult type of cancer to treat, due to late diagnosis which is a result of vague symptoms and lack of biomarkers, as well as refractory behavior toward current treatment protocols. Imaging of the disease progression therefore plays a crucial role in identifying potentially curable PDAC patients at an early stage. Nanoparticle-based contrast agents have shown multimodal capabilities and potential to enhance the contrast of previously undetectable pathological changes, including PDAC. In this master’s thesis study, an iron oxide nanoparticle (IONP) was evaluated as a potential multimodal contrast agent for both photoacoustic imaging (PAI) and magnetic resonance imaging (MRI). The investigated particle was composed of Fe3O4 with a hydrodynamic size of 418.5 nm and a zeta potential of -27.7 mV. In the agarose suspended IONP phantom studies, the IONP demonstrated a two-fold higher T2 contrast compared to commercial IONP VivoTrax (Magnetic Insight), as well as generating strong and stable photoacoustic signal throughout the first near-infrared window (700 to 1000 nm). Based on this thesis’ proof of concept study, Fe3O4 IONP showed good potential as multimodal contrast agent for MRI and PAI. Future work consists of modification of the particle composition and in vivo imaging on animals to evaluate the application in PDAC diagnostics.

The life expectancy of the elderly population is expected to increase with 22 % by 2050. As one grows older, the body starts to deteriorate, which can lead to a higher risk for diseases and accidents. During recent years shoulder surgeries have increased dramatically, and to assess the shoulder function the most common technique is the use of camera-based motion capture systems. However, this is very time consuming and does not completely represent the real shoulder performance. Therefore, this study was aimed to validate a new technique, the force range monitor (FRM). Thirteen volunteers participated in this study, which was divided into two sessions. Session one included abduction and flexion strength measurements using the FRM, as well as six mobility measurements with the inertial measurement unit (IMU) of the FRM. The second session was conducted in the same manner, with the addition of the Vicon system (motion capture system). In this study a control session for FRM and the Vicon system was also performed for abduction and flexion movements with two participants, where the placement of the IMU was modified. As indicated by the results, FRM and the Vicon system do not measure the same parameters. Even if the FRM does not measure the same shoulder joint angles as the Vicon System, the FRM can still be of clinical importance when evaluating position deviation during strength and mobility measurements. Moreover, the FRM had a high repeatability for a number of participants, and most of the mobility measurements presented a distinct patterns for various activities. Hence, it can be concluded that the FRM is a potential technique to evaluate shoulder strength and mobility.

På Sachsska barn- och ungdomssjukhuset vid Södersjukhuset fanns det ett upplevt problem vid akut datortomografi av huvudet på barn mellan 0-1 år. Problemet var att barn ofta vaknar när de ska förflyttas till röntgenbordet. Barn i detta åldersspann placeras då på en specialanpassad Baby Mattress. Arbetets primära mål var därför att ta fram ett lösningsförslag i syfte att underlätta denna förflyttning så att risken för att barnet vaknar reduceras i största möjliga mån. Arbetet grundade sig på metoden tjänstedesign där det inledningsvis samlades in data med hjälp av intervjuer och observationer för att förstå det verkliga behovet. Utifrån detta togs fyra lösningsförslag fram som slutligen vägdes mot varandra. Två av dessa valdes ut då en kombination av dessa ansågs vara mest fördelaktigt. De slutgiltiga lösningsförslagen var produkterna Baby-cocoon 2 respektive Baby-wagon 2. Baby-cocoon 2 innebar optimal komfort för barnet och minimerar känslan av förflyttning till Baby Mattress. Baby-wagon 2 innebar att barnet kan somna i Baby Mattress redan på avdelningen för att sedan transporteras till röntgenavdelningen. Med hjälp av Baby-wagon 2 kan Baby Mattress smidigt lossas på röntgenbordet. På så sätt elimineras momentet med en direkt fysisk förflyttning av ett redan sovande barn. Det sekundära målet var att undersöka stråldosen som ett barn under ett år exponeras för vid datortomografi av huvudet. Detta gjordes med hjälp av insamlade data för 15 patienter. Resultatet var att barn exponeras för en effektiv stråldos på 2,1 mSv. Det har även konstaterats att denna stråldos kan vara skadlig vid upprepade bildtagningar eftersom barn är mycket strålningskänsliga. Detta är något de framtagna lösningsförslagen motverkar.

Muscle morphological parameters such as fascicle length (FL), pennationangle (PA) and physiologic cross-sectional area (PCSA) can provide an insightinto the reasons of the deteriorated muscle functions caused by pathologies.This study investigates the 3D structure of the lower leg muscles using 3Dfreehand ultrasound (3DfUS). This imaging modality uses a motion capturesystem to track the position of the US probe during acquisition and thusreconstruct the structure of the tissues in 3D. In this study, two subjects werescanned on the medial gastrocnemius (MG) and tibialis anterior (TA) musclesin the lower leg using 3DfUS system. The FL and PA of the muscles werecalculated and compared with the values previously measured using diffusiontensor imaging (DTI). The results using 3DfUS were averagely 19.2% largerin FL and 2.9%larger in PA. In conclusion, 3DfUS can successfully determinemuscle morphological parameters within a physiologically acceptable range.But the differences in FL observed between the two imaging modalities werequite big, which probably was due to the differences in sample size and area.The values can also differ greatly within the 3DfUS measurements as a resultof different manipulations during data processing, and the 3DfUS protocolneeds to be further improved in future studies.

Electromyography (EMG) is a common tool in electrical muscle activity measurement and can be used in multiple areas of clinical and biomedical applications, mainly in identifying neuromuscular diseases, analyzing movement or in human machine interfaces. Traditionally a pair of electrodes were used to measure the signals, but in recent years the use of high density surface EMG (HD-sEMG) gained more popularity as it can sample myoelectric activities from multiple electrodes in an array on a single muscle and provide more information. In this thesis a measurement setup and protocol is proposed that can provide a reliably measurement, furthermore multiple features are extracted from the collected signals to characterise the major muscles around the ankle. 5 healthy subjects were tested using an ankle dynamometer with 5 HD-sEMG placed on the Tibialis Anterior, the Gastrocnemius Medialis, the Soleus, the Gastrocnemius Lateralis, and on the Peroneus Longus. Several tests were conducted using different initial angle of the ankle joint and different percentages of the maximum voluntary contraction. The reliability of the setup was assessed by comparing the variance between the collected signals of the same subject in a repeated test, and by comparing different subjects to each other. Results show a reasonably good reliability with less than $10\%$ variance, and adequate selectivity as well. To examine the muscle characteristics, 7 features were extracted from the collected and processed signals, then the features were plotted and compared to signs for muscle characteristics such as muscle fatigue, activation, and spatial distribution of activation. Correlations between features of mean average value (MAV) and zero crossing (ZC), and different muscle characteristics could be observed.

Natural Killer cells are a type of lymphocyte belonging to the innate immune system and they operate cell-mediated cytotoxicity and release of pro-inflammatory cytokines against cancerous cells. However, in vivo testings have shown a reduced activity of NK cells against solid tumors probably due to the negative influence of the immunosuppressive tumor microenvironment. Multicellular tumor spheroids may constitute an advantageous model in cancer biology for studying the mechanisms behind cancer immune editing since it more closely mimics the complexity of the human body compared with the 2D model counterpart. This study investigated the interaction between NK cells isolated from blood and tumor spheroids obtained from A498 renal carcinoma cells, using light-sheet microscopy imaging which allows satisfactory cell tracking in the inner layers of the spheroids. NK cells not only indeed interact with tumor spheroids, but many of them were able to penetrate the spheroids inducing some changes in the structure of the latter. NK cells were also tracked over time, displaying the migration path and calculating the speed. The fluorescence intensity of the NK cells was found reduced as soon as they penetrate the spheroid but, conversely, the speed seems to increase inside the spheroid, a possible sign of the fallibility of the tracking algorithm in this specific case. We propose solutions for more sophisticated future implementations, involving the use of marks during the experimental phase and drift corrections at the data analysis level.

In many developing countries health care conditions are poor and there is a lack of healthcare professionals and diagnostics tools. Cheap and easy-to-use diagnostics tools have been developed to make practicing medicine easier under these conditions. However, signal monitors can be many and spread out, making it hard for the limited number of medical workers to handle. The monitors are also stationary, making mobile supervision impossible. In this thesis a solution is suggested, made of a hardware setup consisting of an Arduino UNO and Bluetooth module paired with an application, capable of analog to digital conversion, wireless transfer and display of medical signals. Furthermore, two different QRS detection algorithms are tested, a larger and accurate model called Pan-Tompkins and a smaller and faster, moving average based filtering system. The transmission circuit as well as the signal displayed showed promise. However, the analog to digital conversion was noisy due to the power source. The tested algorithms showed that speed and low computational requirements are traded for precision.

This thesis project evaluates and compares different deep learning based segmentation tools for acquiring 3D segmentations of mitochondria. These segmentations could then hopefully be used in the future to quantify the mitochondria dynamics, which is vital for the survival of human cells. Four different models were evaluated and compared using the metrices Intersection over Union (IoU) and Dice, and a measurement of the quantity and area of the segmented mitochondria. The four different models were from the Fiji U-Net plugin, MitoSegNet, EmbedSeg 2D and EmbedSeg 3D. The data used was microscopic images of transfected MDCKII cells taken using a Lattice light-sheet microscope. Processing of the data was done in Fiji, which included manual annotation of the images in order to acquire ground truth segmentations. The results showed that the most suited model for this task was the model from the Fiji U-Net plugin. The other models also generated adequate segmentations, but could not adapt to images from a different cell. It was also concluded that stacking together 2D segmentations in order to achieve a 3D segmentations was successful. / Detta examensarbete utvärderar och jämför olika djupinlärningsbaserade segmenteringsverktyg för att få 3D-segmenteringar av mitokondrier. Dessa segmenteringar kan sedan förhoppningsvis användas i framtiden för att kvantifiera mitokondriernas dynamik, vilken är avgörande för de mänskliga cellernas överlevnad. Fyra olika modeller utvärderades och jämfördes med hjälp av måtten IoU och Dice, samt en mätning av kvantiteten och arean av de segmenterade mitokondrierna. De fyra olika modellerna var från en Fiji U-Net-plugin, MitoSegNet, EmbedSeg 2D och EmbedSeg 3D. Datan som användes var mikroskopbilder av transfekterade MDCKII-celler tagna med ett Lattice light-sheet mikroskop. Processeringen av datan gjordes i Fiji, som inkluderade manuell annotering av bilderna för att få ground truth segmenteringar. Resultaten visade att modellen som var bäst lämpad för denna uppgift var modellen från Fiji U-Net-pluginen. De andra modellerna genererade också adekvata segmenteringar, men kunde inte anpassa sig till bilder av en annan cell. En slutsats var också att stapla samman 2D-segmenteringar för att få 3D-segmenteringar var en lyckad metod.