Coming dissertations at Uppsala university
-
Evaluation of some MR and PET techniques in the differential diagnosis of intracranial lesions
Link: http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-538142
There are many different causes of intracranial lesions including infection, inflammation, stroke, tumors, and trauma. To investigate the nature of a lesion, various MR (magnetic resonance) and PET (positron emission tomography) techniques are used that can give information about the anatomical, physiological, and metabolic properties of the lesion and thereby aid in the differential diagnostic procedure. The treatment for brain tumors can include radiation therapy, which can induce brain changes that are difficult to differentiate from a recurring tumor. This thesis evaluates the investigation of intracranial lesions with several MR and PET techniques.
Paper I delt with an evaluation of how much additional information MRS (magnetic resonance spectroscopy) provides in clinical patients compared to MRI (magnetic resonance imaging). In this study that included 208 cases, it was found that the additional information gained from MRS was beneficial or very beneficial in 15% of the cases and misleading in 17% of the cases. In Paper II, a sub-population (n = 100) of the patients in Paper I was investigated. In this paper, use of a decision-support system, INTERPRET DSS 3.1, was compared with conventional analysis of MRS with regard to the correct evaluation of focal lesions. Comparing INTERPRET DSS with conventionally analyzed MRS and MRI, the diagnostic category was correct in 67/58/52 cases, indeterminate in 5/8/20 cases, and incorrect in 28/34/28 cases. In Papers III–IV, the differentiation of tumor recurrence from treatment-induced changes was evaluated. In Paper III, MR examinations were performed for calculation of the perfusion fraction (f) using the intravoxel incoherent motion imaging (IVIM) technique and for the relative cerebral blood volume (rCBV) using dynamic susceptibility contrast (DSC) perfusion in 60 patients. The accuracy of the IVIM parameter f was similar to that of rCBV in differentiating tumor recurrence from treatment-induced changes. In Paper IV, two PET techniques (11C-methionine and 18F-fluorothymidine) were compared in 48 patients. Both techniques had similar efficacy in differential diagnosis between recurrent intracranial tumor and treatment-induced changes.
In conclusion, conventionally analyzed MRS did not add to the diagnostic value of MRI in general. In focal lesions, the INTERPRET DSS system did not improve the categorization of the lesions significantly compared to conventional analysis of MRS but did so compared to MR imaging alone. IVIM can be used to differentiate tumor recurrence from treatment-induced changes. The PET tracers 11C-methionine and 18F-fluorothymidine have a similar diagnostic ability to separate tumor recurrence from treatment-induced changes.
-
From Parking to Power : Integrating an Energy Management System in a Multifunctional Building to Enable E-mobility
Link: http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-538089
E-mobility is pivotal in enabling sustainable and technologically advanced urban environments. In line with this, Sweden's electric vehicle fleet is rapidly expanding, thereby increasing the power necessary for charging electric vehicles. If not properly managed and controlled, this increase in power can potentially threaten grid stability and exacerbate grid congestion.
The primary aim of this thesis was to assess and investigate the potential of a next-generation parking facility at a multifunctional building to be an active part of the city’s distribution grid. The research was guided by the question of to what capacity smart control of a parking facility with a technical system could assist and alter the load demand to generate benefits for both the building and the city’s distribution grid.
This was investigated at Dansmästaren, the first multifunctional building in Uppsala, Sweden. An experimental setup with an electric vehicle charging station and an energy management system was developed at the Ångström laboratory to test and verify control strategies before their implementation at the multifunctional building's parking facility. Thereafter, a second energy management system was developed and implemented at Dansmästaren with the purpose of monitoring and controlling the electric vehicle charging at the parking facility.
The findings of the included papers were divided into two categories. The charging of the electric vehicles can either be assisted by the parking facility's technical system or altered by including the electric vehicle charging in the control for the technical system. Both categories show that a parking facility with a technical system in a multifunctional building can help reduce local grid demand while also providing local benefits for the building.
While the contribution of a single multifunctional building may appear negligible from a grid perspective, the cumulative effect becomes substantial when applied across multiple buildings. Thus, the parking facility at Dansmästaren has the potential to play an active role in the city’s distribution grid through smart charging and the utilization of an energy management system.
-
Deep Learning and Explainable Artificial Intelligence for Biomedical Applications : Methods for Cytology-based Cancer Detection
Link: http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-537511
We live in an era of a rapidly evolving field of AI, where AI technologies have turned into an indispensable tool in a multitude of domains, including biomedical image analysis. Digital cytology, a field that deals with biomedical image data, could substantially benefit from AI technologies. AI techniques offer great potential to support medical experts in detecting diseases such as cancer, by alleviating the load on professionals and uncovering patterns that could go unnoticed by humans. However, AI algorithms come with ethical considerations and potential hazards that require attention and management. Recognising this problem is especially important in applications that handle patient data, given the serious consequences that could arise from mistakes. Another existing problem is that patient data often pose unique challenges that add complexity to the development of AI algorithms intended for handling such information.
This thesis comprises four papers that include methodologies for image classification of data with challenging properties, such as scarcity of fine-grained labels and complex data composition. Importantly, we explore those AI methods that are capable of addressing the lack of interpretability and trust in AI. Two of the four papers in this thesis are dedicated to making feasible end-to-end training of an interpretable multiple instance learning method on datasets with a large volume of data per patient, e.g., cytology data. The research work presented in one of the other papers of this thesis is focused on applying interpretable AI methods to analyse real-world cytology data for cancer detection. Motivated by the shortage of publicly available datasets in digital cytology and the scarcity of fine-grained labels in our real-world digital image cytology data, we investigate the role of synthetic data in the analysis of AI methods. In the fourth paper, we explore the capabilities of AI methods to analyse data with the sparseness of information relevant to a studied condition. This research question is important to answer for cytology-based early cancer detection.
Our findings indicate that while image cytology data analysis comes with challenges, AI methods can play an important role in assisting medical experts by providing information that might prove valuable to them.