Masteruppsatser / Medicinsk strålningsvetenskap

Permanent URI for this collectionhttps://gupea-staging.ub.gu.se/handle/2077/66471

Browse

Now showing 1 - 20 of 48

Optimization of Kidney Dosimetry: Determination of Recovery Coefficient for Kidneys and SPECT/CT Imaging of [177Lu]Lu-DOTATATE/TOC
(2025-10-06) Becker, Vanessa; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
Purpose: Peptide receptor radionuclide therapy (PRRT) with [177Lu]Lu-DOTATATE/TOC is an effective treatment for neuroendocrine tumors, but this treatment also involves an unwanted radiation dose to the kidneys. Dosimetry can be used to optimize treatment and reduce kidney toxicity, and therefore, obtaining an accurate estimate of activity concentration in the kidneys is essential. One of the main challenges in quantitative SPECT imaging for dosimetry is the partial volume effect, which leads to an underestimation of activity. This effect can be corrected for by using a recovery coefficient (RC). The study aimed to obtain an RC for kidneys using a software developed by Hermes for organ and tumor dosimetry. The study investigates how the RC is affected by different activity concentrations in the kidneys and the ratio between the activity concentration in the kidney and the background. It was also studied how the RC is affected by different reconstruction methods and two different projection sets, 30 and 60 projections. Method: Five 3D-printed kidney phantoms were filled with [177Lu]Lu-DOTATATE/TOC and placed in either a Jaszczak phantom or a NEMA phantom. Different activity concentrations were investigated as well as different ratios between the activity concentration in the kidney and the background. Images were collected with a SPECT/CT system (GE Healthcare Discovery NM/CT 670 Pro) and reconstructed in Hermes (Hermia Hybrid Recon 5.0.) using Ordered Subset Expectation Maximization (OSEM) and Relative Difference Prior (RDP). The kidneys were manually segmented, and the RC was determined for each measurement. In addition, the RC was compared between different segmentation methods i.e. the whole kidney segmented volume of interest (VOI), and the small VOI method was examined. Results: The RC was not statistically significantly affected by variations in activity concentration or kidney-to-background ratios. However, a statistically significant difference was shown between the two reconstruction methods, where reconstruction using RDP resulted in a higher RC than when reconstructing images with OSEM. A statistically significant difference was also found between 30 and 60 projections, probably due to increased noise in 30 projections. When comparing various activity concentrations, the whole kidney segmentation approach’s standard deviation was smaller than that of the small VOIs method. At Sahlgrenska University Hospital, the RC was determined to be 0.80 with whole kidney segmentation, 60 projections collected and reconstructed with RDP. Conclusion: The RC for kidneys in SPECT/CT imaging with [177Lu]Lu-DOTATATE/TOC are stable across the investigated range of activity concentrations and background ratios, supporting the use of a standardized RC for all imaging time points for whole kidney segmentation in dosimetry. At Sahlgrenska University Hospital, the RC was determined to be 0.80 and the method for determination of the RC is considered reasonable.
Evaluation of Synthetic Ground Truth for Semantic Brain Image Segmentation - Developing a Database of Synthetic Images with Ground-Truth Segmentations for Practical Application Assessment
(2025-09-10) Mardo, Mardo; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
Introduction: Automated brain image segmentation plays an important role in large-scale neuroimaging research and shows growing potential for clinical applications. Traditional segmentation pipelines rely on manually annotated MR-images as atlases, but such data are time-consuming to acquire and may not be readily available in sufficient quantity or variety. This thesis documents a research project where the use of synthetically generated MRimages with known ground truth was explored to support evaluation and benchmarking of automated segmentation methods. Aim & Purpose: The aim of this thesis is to evaluate the performance and robustness of the multi-atlas segmentation with enhanced registration (MAPER) algorithm using leave-one-out cross-validation, facilitated by the generation of synthetic MR images with controlled structural and intensity alterations. Additionally, this study assesses whether the synthetic images can be used as effective atlases compared to conventional MR-images and find a suitable benchmark for evaluation of different MAPER versions; onepad (master) and no-onepad. Method: Five types of synthetic MR-images were generated based on the IXI and Hammers datasets, each with different modes of spatial and intensity modification. Experiments were designed to (1) compare segmentation performance between synthetic atlases and unmodified conventional atlas (2) validate segmentation performance and robustness using leave-one-out cross-validation on synthetic targets, (3) quantify the convergence rate for the different ground truth benchmarks based on a parametric model and perform bootstrapping as a benchmarking measure to find the most sensitive benchmark. Then (4) apply this benchmark to compare two MAPER configurations. Results and conclusion: Segmentation performance was highest using conventional MRimages atlases, followed by statistical and scrambled synthetic variants. Leave-one-out cross-validation (LOOCV) confirmed similar trends. In benchmarking, the statistical smoothed image (statsmooth) type exhibited the fastest convergence rate, making it the most sensitive benchmark. Application of the benchmark revealed that the no-onepad version of MAPER outperformed the standard master configuration. Discussion: The results highlight that although synthetic images underperform compared to conventional data as atlases, they provide meaningful insights into algorithm robustness and segmentation behaviour. The ground truth benchmark shows potential as a controlled and reproducible framework for comparing segmentation methods and identifying subtle differences in performance.
Automatic exposure control for digital radiography without a physical anti-scatter grid
(2025-09-04) Ring, Louise; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
Purpose: This study aimed to find a factor R for different parameter settings that can be used to adjust the AEC preset value, in order to maintain a constant amount of primary radiation to the detector when the anti-scatter grid is removed. Method and materials: A factor R was calculated as the fraction between dose-area product (DAP) with and without an anti-scatter grid, from each exposure on polymethylmethacrylate (PMMA) phantoms, multiplied with a primary transmission factor Tp. These measurements lead to a deeper understanding of how varying PMMA thickness, kV and field size would influence R. Second, anthropomorphic phantoms were examined, with a built-in AEC assistance detector (BiAA), while increasing their thicknesses and using different kV and field sizes. An adult chest phantom were examined in anteroposterior (AP) and lateral (LAT) projections (AP/LAT thickness: 21/29, 27/35, 32/39 cm). An adult pelvis and hip phantom (thickness: 23, 29, 35 cm) was examined in AP projection. A pediatric chest phantom (thickness: 15 cm) was examined in AP projection. Results: The Rs from examination of the PMMA resulted in an increase as the thickness increased. This trend was also observed when the field size increased. However, as the kV increased, R decreased. The adult chest (thickness: 21, 27 and 32 cm) AP examination resulted in Rs of 1.13, 1.36 and 1.59. The adult chest (thickness: 29, 35, 39 cm) LAT examination resulted in Rs of 1.20, 1.36 and 1.51, with 141 kV. The adult pelvis (thickness: 23, 29, 35 cm) AP examination resulted in Rs of 2.89, 3.49 and 4.13 with 70 kV. The Rs for the hip examination were 2.34, 2.95 and 2.62. The pediatric chest (thickness: 15 cm) AP examination resulted in an R of 1.08. Discussion and conclusion: R had a similar relative increase for the examination of the adult chest and pelvis with increasing thickness, but R was higher for the pelvis. The change in R between the different thicknesses and phantoms was substantial, which implies the importance of optimizing the examination protocol according to patient size during a gridless examination with AEC. The difference in R also means that the protocol should include a combination of R for a specific anatomic region and patient size, for a finer adjustment of the AEC preset dose. The results of the pediatric phantom measurement, where two different mA were used, the lower mA increased the exposure time. However, the exposure time remained shorter than the minimum recommended by the generator manufacturer. This implies that for smaller thicknesses, the AEC may not be reliable. This study can be used as a base for further studies, and the performance of virtual anti-scatter grids and its impact on image quality needs to be investigated.
Characterization of spinal tissue in lumbar spinal stenosis by advanced image analysis of MR images
(2025-03-24) Nilsfors, Ella; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
Background. Lumbar spinal stenosis (LSS) is known for causing back and leg pain due to nerve compression but the impact on structures such as intervertebral discs (IVDs), vertebrae, and ligaments is still poorly understood. This knowledge gap may limit the diagnostics and the optimization of the treatment plans for patients with LSS. The aim of the present study was to investigate whether patients with LSS exhibit tissue changes at index levels and at lumbar levels designated for surgery. Method. In 350 LSS-patients awaiting surgery, the IVDs (L1/L2 – L5/S1) and vertebrae (L1 – S1) in both T1- and T2-weighted MR images were segmented. Mean signal intensities (SIs) and standard deviations (SDs) of the means were determined globally and regionally within the segmented IVDs and globally within the vertebrae. Additionally, Δμ (a metric of disc degeneration) was determined and the entropy of the vertebral SI was introduced as a novel metric to evaluate disorganization. All metrics were analyzed for index level and levels planned for surgery and compared between groups. Additionally, Δμ and entropy were correlated to age. Results. Statistical differences in the metrics mean SI and SD were observed between index and non-index levels, as well as between segments planned and not planned for surgery. The largest differences were found in the nucleus pulposus, located in the central regions of the IVD and including subregion 2,3 and 4 (p <0.04). For vertebrae statistical differences were observed in SD and entropy for both index- and surgery level. Vertebral entropy demonstrated the strongest correlation with index levels (r = -0.166; p= 0.002). While Δμ did not display a statistical difference at a specific index level, it displayed an age dependence (e.g., IVD L1/L2; r = 0.399, p <0.001) at most lumbar levels, except at L4/L5 (r = -0.0975, p= 0.070). Δμ did also depend on specific lumbar level. In contrast, vertebral entropy displayed an age dependence at all levels, but no level-specific variation. Conclusions. This explorative study shows distinct differences in tissue characteristics in patients with LSS that correlate with index level and levels planned for surgery. As expected, Δμ was found to depend on both age and spine level. In IVDs at level L4/L5, however, an age-dependency could not statistically be established. Notably, vertebral tissue seemed to display a more homogenous pattern with age, as determined by a decreased entropy. This may reflect the transition in the bone marrow composition in older individuals. Further research is needed to explore the combined impact of multiple metrics and develop predictive tools for earlier intervention and improved clinical decision-making.
Moving towards clinical implementation of a deep-learning brain tumour segmentation algorithm: DICOM server integration
(2025-03-05) Liljas, Camilla; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
Aim: This research project is part of a larger initiative where the aim is to develop a software tool that supports radiologists by providing more accurate tumour volume measurements on magnetic resonance imaging (MRI) scans. A specific aim within this larger endeavour is to integrate a deep-learning algorithm for brain tumour segmentation with a digital imaging and communications in medicine (DICOM) server. Method: The DeepMedic segmentation model was trained and tested using the brain tumour segmentation (BraTS) 2024 dataset, which included post-treatment glioma images from four MRI sequences. The model was trained to accurately identify distinct tumour regions. Orthanc was configured and used as a DICOM server to handle medical image retrieval, storage, and reintegration. A Python-based pipeline, built using existing Python packages and libraries, was developed to enable communication between DeepMedic and Orthanc, facilitating automated segmentation and data management. Result: The model was successfully trained on the BraTS 2024 dataset, without overfitting. Testing with the saved model demonstrated its ability to accurately segment the various tumour regions, as evaluated using the Dice similarity coefficient and the 95th percentile of the Hausdorff distance (HD95). While the model performed well in most cases, some discrepancies between regions were observed. An automated workflow integrates DeepMedic with Orthanc. The user enters a study id, after which Orthanc retrieves the corresponding DICOM series, converts, and prepares the images for DeepMedic. After segmentation, the results are converted back to DICOM format and sent to Orthanc for storage and availability for further use. Conclusion: The project demonstrated the potential of the technical implementation of a deep-learning-based brain tumour segmentation algorithm. However, further work is needed to enhance the robustness of the pipeline and ensure its applicability across a wider range of clinical settings before it can be tested on in-house clinical data.
Assessment of bioaccessibility of naturally occurring radionuclides from snus to consumers
(2025-03-05) Sanderson, Emma; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
Purpose: Naturally occuring radionuclides can be found in everything we consume, including the tobacco used to make snus. This study aimed to investigate the presence of naturally occurring radionuclides in snus, their bioaccessibility during consumption, and the associated radiological risks to consumers. Methods: A total of 16 samples were sourced from Swedish stores to reflect the national market. A radiological characterization of the samples was conducted through digestion, radiochemical processing, and alpha spectrometry analysis, focusing on 210Po, 232Th, 234U, and 238U. The activity concentrations of these radionuclides in the snus were calculated. A bioaccessibility study analyzed the same radionuclides, simulating real-life snus use to estimate the fraction extracted by saliva. Both used snus and saliva were processed and analyzed using alpha spectrometry. The committed effective dose from annual snus consumption was then calculated. Results: The characterization revealed that 210Po ranged from 2.4–3.0 mBq/g in loose snus and 1.2–5.1 mBq/pouch in pouched snus. For 232Th, levels varied between 0.22–0.42 mBq/g in loose snus and 0.1–0.66 mBq/pouch in pouched snus. 234U ranged from 0.22–0.27 mBq/g in loose snus and 0.07–0.21 mBq/pouch in pouched snus, while 238U was between <0.04–0.23 mBq/g in loose snus and 0.08–0.26 mBq/pouch in pouched snus. Minimal or no activity was detected in tobacco-free snus. The bioaccessibility study demonstrated that all samples transferred 210Po into saliva, with leaching values ranging from 9 ± 3% to 34 ± 7%. Only a limited number of samples showed leaching of other radionuclides, and just one sample appeared to have transferred all radionuclides into saliva. The committed effective dose from one year of snus consumption attributed to 210Po ranged between 1.3 - 8.3 μSv, averaging at 4.55 μSv. Discussion and conclusion: This study conclusively demonstrates that snus use contributes to the annual radiation dose, providing a basis for future research. However, it has limitations, including unexplored factors such as the packaging and design of pouched snus. While the contribution to the annual dose is present, its specific implications for human health remain uncertain. Addressing these knowledge gaps by refining the study design, the inclusion of additional variables, and further investigation into the radiological risks of snus could enhance understanding and potentially lead to strategies that mitigate associated risks in the future.
Opportunistic screening for osteoporosis with CT performed before dose planning for radiation therapy
(2025-03-05) Ljung, Frida; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
Purpose: Radiotherapy can negatively impact bone health, affecting the activity of osteoblasts, osteoclasts, and osteocytes, leading to dysfunction in bone remodeling and reduced bone mineral density (BMD). This increases the risk of fractures in irradiated bone. Preexisting osteoporosis is a well-known risk factor of radiotherapy induced fractures. While Dual energy X-ray Absorptiometry (DXA) is the gold standard for diagnosing osteoporosis, it is not routinely performed before radiation therapy. This study aimed to develop and validate a method for opportunistic screening for osteoporosis by estimating the bone mineral density from computed tomography (CT) scans used for radiation treatment planning. The study also sought to categorize patients into low, moderate, or high risk for osteoporosis. Method: A total of 179 women (mean age 67.8 ± 8.6, range 35-85 years) treated for adjuvant breast cancer were included in the study. Of these, 101 patients underwent both DXA and CT scans within a four-month period. A spherical volume of interest (VOI) was used to extract HU values from the trabecular region of Th4-L1. BMD was estimated using two methods: a translation curve derived from known calcium content and HU values for fourteen reference bone tissues, and internal calibration. Eight additional patients treated for gynecological cancer were included to study how BMD varied in the spine and to develop thoracic BMD cutoff values. Validation was performed using a European Spine Phantom (ESP). Results: A translation curve from HU values to BMD was derived as BMD = 0.7161 · HU − 1.212. Using this equation, BMD estimates in the ESP showed good agreement with actual values. Thoracic BMD cutoff values for osteopenia and osteoporosis were derived as 91 mg/cm3 and 124 mg/cm3, respectively. A moderate correlation was found between BMD estimated using DXA, areal BMD (aBMD), and CT based BMD (r=0.51, p<0.001). The CT-based method detected more cases with low BMD compared to DXA. Conclusion: CT images acquired for radiotherapy planning can be used to assess BMD, with good agreement to phantom measurements and higher sensitivity compared to DXA. Further research is needed to evaluate the method at different anatomical sites and validate thresholds for low bone density. Future research should also identify and quantify the uncertainties related to the method.
Can cooking practices affect dietary exposure to 210Po in seafood?
(2024-11-13) Nordqvist, Timmy; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
Background: The estimated yearly worldwide average exposure to radiation is about three mSv, with 2.4 mSv coming from natural radiation sources. 210Po is a highly radiotoxic decay product of 238U and a major contributor to internal radiation dose in humans, especially from seafood consumption. Swedish dietary guidelines recommend frequent seafood consumption, and thus, monitoring 210Po levels is important due to significant dose contributions and lack of guidelines on intake restrictions. Studies show that the Swedish population's yearly average radiation dose from seafood is about 100 μSv, with high consumption groups receiving above 500 μSv for children and 300 μSv for adults. The choice of method to cook seafood can influence the 210Po levels, which impacts the dose assessments from seafood ingestion. Aim: The aim of this master thesis was to determine the impact of cooking on the activity concentration of 210Po in seafood, and to calculate a committed effective dose to humans for consumption of seafood purchased in Sweden. Method: Blue mussels and Norwegian lobsters were bought from selected fish stores in Gothenburg, Sweden. Blue mussels were cooked by boiling, steaming, roasting, or by heating in a microwave. Norwegian lobsters were cooked by boiling, steaming, roasting, or pan-frying. 210Po in blue mussels and Norwegian lobsters was separated using a radiochemical separation method and then measured using alpha spectrometry. The committed effective dose per kg of blue mussels or Norwegian lobster tail meat for an adult was calculated based on the activity concentration of 210Po measured in said seafood. Results: The activity concentration of 210Po varied by cooking method for mussels and lobster. For raw mussels, the average committed effective dose was 44 ± 6 μSv per edible kg, and for boiled mussels 68 ± 8 μSv per edible kg. For raw lobster tail, the committed effective dose ranged from 3.1 ± 1.1 μSv per edible kg to an average of 5 ± 1 μSv per edible kg. The higher doses are likely due to the water content lost during cooking. Discussion and Conclusion: The study observed some variation in 210Po activity concentrations between replicates for blue mussels and Norwegian lobsters for all cooking methods. Raw blue mussels had 210Po levels consistent with previous studies, but more replicates are needed for greater accuracy and to compare different cooking methods. Boiling transferred more of 210Po to the water compared to steaming. Future studies should include more replicates and study e.g., how cooking time affects the activity concentration of 210Po.
Developing an Energy Resolution Model for Monte Carlo-Based SPECT Simulation and Reconstruction Using a GE Discovery NM/CT 670 Pro System
(2024-09-23) Andelin, Alexandra; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
Aim: In order to apply Monte Carlo (MC)-based algorithms for simulations of gamma camera acquisitions, the intrinsic characteristics of the system must be properly defined. This study aimed to determine the energy resolution of a GE Discovery NM/CT 670 Pro gamma camera system, and to develop a nonlinear model for integration in the MC-based program Sahlgrenska Academy Reconstruction Code (SARec). A part of the study also aimed to evaluate different measuring conditions, such as dead time, source geometry, and set-up geometry, and their effect on the energy resolution. Methods and materials: Measurements were performed using said gamma camera system. Spectra were retrieved by extracting listmode data. A MATLAB script was created to determine the full-width half-maximum (FWHM) of the recorded photopeaks using a Gaussian fitting technique. To determine optimal measurement conditions, variations in FWHM under different acquisition circumstances were studied. The effects of dead time were examined by measuring a 131I source at various distances and at different time points. The impact of source geometry and measurement set-up were evaluated using 177Lu, in a syringe and a vial. An evaluation of count statistics was performed by measuring a 241Am source as well as a 177Lu syringe using different acquisition times. To develop the energy resolution model, intrinsic measurements were performed using 241Am, 177Lu, 99mTc, 133Ba, and 131I. The recorded FWHM was fitted as a nonlinear function of photon energy, thus obtaining the energy resolution model. Results: Significant dead time effects arose at counts rates over 180 kcps. Set-up geometry did not affect the measured energy resolution. A small difference in FWHM was observed between the 177Lu syringe and vial (0.11 and 0.43 percentage points at 113 and 208 keV, respectively). Calculated FWHM values did not change significant after acquiring 1,500 kcounts in the photopeak window. The measured energy resolution was 11–8.3% at E = 59–364 keV. The resulting energy resolution relationship was "FWHM = 0.1239 + 0.5428" √(E+0.021E^2 ) with a R2 of 0.89. Conclusion: The evaluated measurement conditions did not affect the energy resolution, although a connection between low count statistics and variations in the calculated FWHM of the peak was observed. The established nonlinear relationship between photon energy and energy resolution provided a better fit to the measured energy resolution, compared to the linear model employed in SARec today, and would thus provide better perquisites for accurate MC-based simulation and reconstruction. However, validation of the model should be performed to evaluate performance.
Treatment Plan Quality for Deep Inspiration Breath-Hold Breast Cancer Radiation Therapy - Comparison between Conventional Technique and Modulated Approaches
(2024-09-23) Yaqoubi, Atefa; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
Introduction: Breast cancer remains a significant health concern globally, impacting millions of individuals each year [1]. Radiation therapy, an adjuvant to surgery and chemotherapy, aims to eliminate residual cancer cells and inhibit tumor growth [2]. The most common breast cancer radiation therapy technique in Sweden is 3D Conformal Radiation Therapy (3DCRT) with static photon beams. 3DCRT is a widely accessible and robust treatment technique concerning positioning uncertainties, respiratory movements, and breast swelling. However, the dose is not well-conformed to the target volume, resulting in greater exposure to nearby healthy tissues such as heart and lungs. [3]. Thus, exploring new techniques is crucial, especially for advanced stages. Intensity Modulated Radiation Therapy (IMRT) and Volumetric Modulated Arc Therapy (VMAT) offer precise radiation delivery and high conform dose to the target region while minimizing damage to nearby healthy tissues, however, they also increase low-dose exposure to nearby healthy tissues[3]. Additionally, there has been concern about the robustness of these modulated techniques, which has been a significant barrier to their widespread adoption. Therefore, hybrid techniques combining 3DCRT with IMRT (Hybrid-IMRT) or VMAT (Hybrid-VMAT) have been developed to improve dose conformity and reduce off-target absorbed dose exposure. The aim of this study is to compare the quality of treatment plans for left-sided breast cancer across four techniques: 3DCRT, Hybrid-IMRT, Hybrid VMAT, and VMAT. The assessment will include homogeneity and conformity indices, dose distribution to target volumes, and nearby organs at risk (OARs) as planned dose distribution. Robustness against positioning uncertainties and breast swelling as delivered dose distribution. Additionally, the feasibility of Deep inspiration breath hold (DIBH) for each technique will be analyzed. Material and Methods: Computed tomography (CT) data from nine female patients under 40 years with left-sided breast cancer previously treated at Sahlgrenska University Hospital were analyzed. The prescribed absorbed dose was 40.05 Gy to the whole breast and 48 Gy to the lumpectomy cavity. All CT data was acquired during DIBH. Treatment planning was conducted for a Varian TrueBeam linear accelerator in the Varian Eclipse (Varian Medical Systems, Siemens Healthineers) treatment planning system (TPS). Four treatment techniques were evaluated: 3DCRT, Hybrid-IMRT, Hybrid-VMAT, and VMAT. The data is a subset from the study by Svensson et al. [4]. Dose distributions to target volumes were evaluated according to Sweden national guidelines [2], with OARs including the heart, lungs, left anterior descending artery (LAD), and contralateral breast [2]. Homogeneity- and Conformity-index were calculated using Dose Volume Histogram (DVH) data. Robustness was assessed by shifting the treatment plan isocenter, 3 mm and 5 mm, following in six directions (±x, ±y, ±z), and variations in D98% for CTV and CTV-T, and mean doses for OARs were examined. For breast swelling, a surface contour of 5 mm thickness, called CTVsurface, was delineated along the edge of the CTV. The isocenter was shifted 1.1 cm medially and dorsally within the CTVsurface. Changes in D98% and V93% were studied for CTVsurface. The number of DIBH and mean DIBH time were analyzed for each technique. Results and Discussion: All four treatment techniques adhered to national guidelines for the target volumes. 3DCRT and Hybrid-IMRT presented significantly lower mean doses to OARs compared to Hybrid-VMAT and VMAT, with VMAT resulting in higher doses to the LAD due to increased low-dose spillage [3]. The Hybrid techniques, especially in Hybrid-IMRT, provided superior dose homogeneity and conformity compared to 3DCRT and VMAT. The analysis of robustness to positioning uncertainties showed no significant differences in dose distribution for CTV and OARs among the four techniques. For CTV-T, Hybrid-IMRT was not as robust as 3DCRT, Hybrid-VMAT, and VMAT. This can be explained by the absence of a margin outside the breast that accounts for breathing motion and breast swelling, leading to loss of dose fluence in Hybrid-IMRT.[5]. VMAT and Hybrid-VMAT presented equal robustness to breast swelling as 3DCRT, while a large under dosage for CTVsurface was observed for Hybrid-IMRT. 3DCRT had shorter breath-hold time which is beneficial for patients with breath-holding challenges. In conclusion, the quality of dose distribution was evaluated for four different radiotherapy techniques. The quality of planned dose distribution was highest for Hybrid-IMRT, both for target volumes and organs at risk. The most robust dose distributions, considering positioning uncertainty and breast swelling, were observed with 3DCRT, VMAT, and Hybrid-VMAT.
Age-Related Late Effects in Rat Thyroids After Internal 131I Exposure
(2024-09-23) Parling, Ludwig; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
The Chernobyl nuclear accident left many questions in its wake. One of these questions was why children affected by the radioactive fallout seemed to have a higher incidence of thyroid cancer compared with adults in the same area. The purpose of this study was to compare proteomic changes in internally 131I irradiated rats of two age groups, one young and one adult. This was performed through analysis of fresh frozen thyroid tissue from a previous animal study where young and adult rats were exposed internally to various amounts of 131I. Analysis of the rats’ thyroid proteome was achieved through liquid chromatography tandem mass spectrometry and analysed with Welch’s t-test with Benjamini-Hochberg correction. Protein functional analysis and gene ontology (GO) term analysis were done with UniProt. Large variations in mainly the control group’s data made comparison at the group level difficult. One protein’s relative abundance, TRXR3, was found significantly changed in the young population at 0.5 kBq. TRXR3 is the rat equivalent to the TRXR3 found in humans and is a part of cellular oxidant detoxification and redox homeostasis. This would imply that oxidative stress had taken place, but the lack of other markers makes it difficult to draw a solid conclusion. Due to the lacking appearance at other doses, TRXR3 was not considered a good candidate biomarker for 131I radiation exposure or absorbed dose
VALUATION OF CHALLENGES IN SPECT-BASED DOSIMETRY IN THE TREATMENT OF METASTATIC PROSTATE CANCER WITH 177LU-PSMA-617
(2024-09-23) Engström, Axel; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
Introduction: In nuclear medicine, targeted radioligand therapy is an established treatment option for metastatic cancer disease. A radioactive isotope is attached to a carrier molecule that will specifically bind to a certain target. One such example is the prostate-specific membrane antigen (PSMA); a transmembrane glycoprotein often highly expressed in metastatic prostate cancer tumors. Recently, 177Lu-PSMA-617 was approved by the Food and Drug Administration (FDA) for clinical treatment of prostate cancer patients under the name Pluvicto. This treatment was implemented at Sahlgrenska University Hospital in 2023 and to establishing a method of dosimetry for 177Lu-PSMA-617 is thus of great importance. The overall aim of the study was to evaluate some factors thar are challenging in 177Lu-PSMA-617 SPECT-based (single photon emission tomography) dosimetry. Method: This study included four patients, injected with 7.4 GBq of 177Lu-PSMA-617 for 1–6 treatment cycles, with 1 or 4 SPECT/CT images per treatment cycle (day 1 or day 0, 1, 2, and 7). All patients underwent a pre-therapeutic 68Ga-PSMA-11 PET/CT scan (positron emission tomography/computer tomography). A new SPECT protocol was established using three image field-of-views (FOVs), with 30 projections per FOV. The kidneys, submandibular, and parotid glands were manually outlined in the CT images. An analysis was performed to assess the uncertainties associated with manual segmentation. Subsequently, an attempt at establishing a threshold-based method for organ volume estimations using the pre-therapeutic 68Ga-PSMA-11 PET and the therapeutic 177Lu-PSMA-617 SPECT imaging was made. Repeated segmentations of the submandibular and parotid glands were performed in the CT images to obtain reference volumes. Varying thresholds of maximum activity were used to determine gland volumes in the PET and SPECT images. The volume differences could then be analyzed, and the optimal threshold decided. Furthermore, SPECT-based kidney dosimetry was conducted for each patient. Segmented kidney volumes were applied to the SPECT reconstruction. Activity was quantified and time-activity curves generated. These was then used to calculate the average absorbed kidney dose. Lastly, the correlation between absorbed dose in kidneys from SPECT and signal in kidneys from PET was evaluated. Results: Manual segmentations of the left and right parotid glands showed mean uncertainties of 2.6–19%. For the four patients, the optimal segmentation thresholds for the submandibular glands in SPECT were 57.4, 40.7, 39.9, and 35.8%, respectively, with mean uncertainty 21.2–49.6%. In PET, optimal thresholds for the submandibular and parotid glands were 34.1% and 30.2%, respectively, with a mean uncertainty of 4.9–13.9%. The average absorbed kidney doses were 3.0, 1.2, 3.8, and 3.3 Gy, respectively. There was a positive correlation between absorbed dose in SPECT and signal in PET with an R-value of 0.88. Conclusions: The accuracy of manual segmentations can be largely affected by artefacts, anatomical changes, and the judgement of the operator. This study demonstrates that a threshold-based method has the potential to be used for volume determination of the submandibular and parotid gland in PET imaging, with uncertainty similar to that of manual segmentation. A threshold-based segmentation would allow for a more efficient volume determination which would simplify the absorbed dose calculations. The calculated average absorbed kidney doses conformed to earlier studies, reinforcing current knowledge on the effects of 177Lu-PSMA-617. A correlation between absorbed dose in SPECT and signal in PET was observed, supporting the use of 68Ga-PSMA-11 PET as a predictor of 177Lu-SMA-617 absorbed dose.
Investigating the Impact of Imaging Parameters on Quantification and Interpretation in Brain PET
(2024-06-19) Zellaya, Rim; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
Rationale CortexID software (GE Healthcare) assists in interpreting and assessing FDG and beta-amyloid brain Positron Emission Tomography (PET) images for neurological disorders. It provides Z-scores that indicate abnormal radioactive tracer uptake. These Z-scores are based on the patient’s standardized up take value ratios (SUVRs) compared to a chosen reference brain region. They reflect deviations from SUVRs in healthy individuals of the same age. A positive Z-score indicates an increased uptake and a negative Z-score indicates a decreased uptake of the radiopharmaceutical relative to the normal database. Questions have been raised whether image reconstruction parameters affect the expressed Z-scores, potentially impacting the diagnostic accuracy. This study investigates how the use of the Bayesian Penalized Likelihood (BPL) reconstruction algorithm in CortexID, with various beta-values, and the effects of Anatomical Tracers for Longitudinal Assessment and Synthesis (ATLAS) and Zero-echo-time (ZTE) attenuation correction, influence the interpretation and quantification of brain PET/MR images. Methods List-mode PET data from 25 patients who underwent 2-deoxy-2-[18F]fluoro-D-glucose (FDG) 3D static brain exams using a PET/MRI Signa 3T scanner (GE Healthcare) were reconstructed with the BPL algorithm at 𝛽-values of 𝛽50, 100, 200, 400, 600, 1000, and 1500. Attenuation correction (AC) methods, ATLAS AC and ZTE AC, remained constant as 𝛽-values varied, resulting in 14 images per patient. CortexID analyzed these images, obtaining Z-Scores. The study assessed Z-Score variations with different 𝛽-values and AC choices. 𝑠𝑦𝑛𝑔𝑜.via displayed images from one patient to visually assess parameter impact on quality and interpretation. Each image underwent analysis with two Volume-of-interests (VOIs): one on a uniform background and one on a region with high tracer uptake. Extracted max-, mean-, and peak-SUV values provided quantitative data. Results In brain PET/MR images, higher 𝛽-values led to lower Z-Scores, and ATLAS AC showed slightly lower Z-Scores than ZTE AC. Lower 𝛽-values resulted in sharper but noisier images, while higher 𝛽-values yielded blurrier but less noisy ones. ATLAS AC produced slightly sharper images compared to ZTE AC, with an almost imperceptible difference. Max-, mean-, and peak-SUV decreased with increasing 𝛽-values in high tracer uptake regions. Mean-SUV in the uniform background remained nearly constant. Optimal image quality to the naked eye was achieved at 𝛽100-𝛽200. Conclusion In CortexID, higher 𝛽-values in BPL reconstruction for FDG brain PET/MR images lead to decreased Z-Scores. ZTE AC shows higher Z-Scores compared to ATLAS AC, with minimal differences in image comparison. Higher 𝛽-values lead to a reduction in max-, mean-, and peak-SUV. Additional research is needed to assess the optimal 𝛽-value in CortexID image quantification.
Compartment model for radon-induced absorbed dose in lungs
(2024-04-23) Hosseini, Sofia; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
Inhalation of radon gas and radon daughters has since a long time been reported as the second leading cause of lung cancer after smoking. The biological hazards make it therefore important to quantify the absorbed doses from radon and radon daughters. The first aim of the study was to reconstruct the Human Respiratory Tract Model, presented by ICRP, to describe the biokinetics, i.e. movement of radon and radon daughters after inhalation for members of the public. The biokinetic model was implemented in the software tool Ecolego and was supplemented with Human Alimentary Tract Model and systemic models of radon, polonium, lead, and bismuth for a complete description of internal exposure. The final biokinetic model enables the user to acquire time-integrated activity (Ã) for combinations of 1) radon and radon daughters, 2) different modes of particle size distribution of radon daughters in air, 3) for members of the public of ages 3 months old, 1 year old, 5 years old, 10 years old, 15 years old and adults defined by ICRP and 4) genders for 15 years old and adults. The model was in a first step bench-marked against the internal dose calculation software Taurus, through comparison of Ã in the lungs, which resulted in approximately 20% relative bias between the implemented biokinetic model and Taurus. For an overall evaluation of the biokinetic model, the effective dose for radon daughters was calculated to be 18.22 mSv per WLM, which is in the range of values (7–21.1 mSv per WLM) obtained by other researchers. A secondary aim was to study how absorbed doses from emitted α-and β particles following the decay of radon and radon daughters are distributed between tissues of the lungs. Absorbed doses were generated using the software program IDAC-Dose 2.1. In addition, values of S coefficients for combinations of source regions and target tissues inside the respiratory tract were incorporated in the software giving approximative values on absorbed doses and associated uncertainties. The absorbed dose from inhaled radon gas were found to be two orders of magnitude smaller than absorbed doses from inhaled radon daughters. The highest absorbed doses were, as expected, found to be from emitted α particles from 218Po- and 214Po-decay in nearly all tissues of the lungs.
Evaluating aperture shape controller (ASC) in modulated radiation treatments
(2023-09-28) Andrén Cederholm, Vilma; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
To decrease aperture complexity, the aperture shape controller (ASC) is a recently introduced function in Eclipse external beam planning (Varian Medical Systems, Palo Alto, USA). ASC works within the photon optimizer, minimizing distances between adjacent leaf tips of the MLC. The user can define via settings, ranging from Very Low to Very High, at which level the ASC is allowed to influence the optimization. Varian also suggests using another feature, convergence mode (CM), when using high settings of ASC. Since both experience and results of planning with ASC are limited, this work aims to examine the effects of ASC use in clinical treatment planning by investigating plan quality. The project was divided into two sub-studies. Sub-study 1 investigated plan quality parameters of executed treatment plans extracted pre- and post-ASC introduction. In sub-study 2, plans were created with different ASC and CM settings to evaluate how these settings affected the same plan quality parameters. The plan quality parameters used for both sub-studies were Edge Area Metric (EAM), gamma evaluation comparing planned and delivered dose distributions and plan quality index (PQI). Executed treatment plans for cancer diagnoses, rectum, tonsil, and prostate, were extracted from the database ARIA at Sahlgrenska University Hospital for sub-study 1. The resulting comparison groups were ASC None and Very Low for the prostate and tonsil and Lower versus Higher for the rectum. The rectum groups Lower and Higher includes ASC settings Very Low, Moderate, and High, Very High. A system for calculating a plan quality index was developed using clinical objectives, data from clinically accepted plans, and consults from clinicians. In sub-study 2, 15 prostate group plans were re-planned with different ASC and CM. When re-planning, the objectives were to have similar dose distribution between plans of the same patient case and fulfill clinical objectives. Time spent planning was also recorded during sub-study 2. Results were evaluated using Mann-Whitney U-test and comparing group medians. The results of sub-study 1 indicated a difference in medians and results of the Mann-Whitney test for rectum EAM and prostate PQI. For EAM, results suggested that a higher ASC results in a lower EAM. In the PQI results, ASC Very Low resulted in a lower PQI compared to ASC None for prostate plans. The results were inconclusive for all other ASC group comparisons or showed no significant differences. In sub-study 2, results indicated that a lower MLC complexity is obtained when using a higher ASC setting. No difference was seen when comparing the EAM results of different CM settings. The gamma pass rate and PQI results showed no significant difference between groups of different ASC and CM settings. Results of recorded time spent planning in sub-study 2 showed higher variances between individual plans than between groups of other ASC and CM settings. This concluded that using ASC at a high level could decrease aperture complexity. The ASC usage did not seem to affect the treatment deliverability negatively. ASC was found in one sub-study to have a negative effect on dose distribution quality for prostate plans. No other results in dose distribution quality indicated a difference between ASC settings. In addition to these results, using ASC and CM was not found to impact time spent planning negatively.
Anatomical segmentation of the human brain: comparative assessment of two automatic methods
(2023-08-22) von Dorrien, Carl; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
Magnetic Resonance Imaging (MRI) is a robust and versatile imaging modality and an integral component of a lot of studies, especially when performing quantitative analysis. MRI is the preferred method of imaging the brain because of its excellent soft tissue contrast. Accurate segmentation of the brain into its anatomical regions enables accurate quantitative analysis of the brain. Three software programs that perform automatic anatomical segmentation of the human brain are FreeSurfer, FastSurfer and MAPER. The purpose of this study was to use FreeSurfer as a baseline, and to investigate how well FastSurfer and MAPER segmentations conform to FreeSurfer’s outputs on the same dataset. 185 T1-weighted 3D MR images from the IXI Dataset were segmented using FreeSurfer, FastSurfer, and MAPER. Default training checkpoints were used for FastSurfer and FreeSurfer outputs of the IXI Dataset, along with corresponding brain MR images, were used as a source atlas for MAPER. The FastSurfer and MAPER segmentations were then compared with the FreeSurfer segmentations using the Jaccard Similarity Coefficient. MAPER performed better than FastSurfer at replicating FreeSurfer-conform outputs for subcortical regions. MAPER and FastSurfer performed similarly for the cortical regions.
Apoptosis induction in breast cancer cells after radiotherapy and potential radiosensitizers
(2023-06-27) Simonsson, Klara; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
Background. Today breast cancer is the foremost cancer death type amongst women around the world. At present some of the toughest challenges in the clinic is recurrent, radioresistant breast cancer, and metastatic breast cancer where low cure rates are observed after surgery and radiotherapy. One way to possibly increase survival rates, is combining radiotherapy and systemic therapy that doubles as a radiosensitizer. The aim of this thesis was to identify pathways that increase radiosensitivity in breast cancer cells after combination treatment with external irradiation and anti-cancer drugs, specifically through increased apoptosis. Method. Three different human breast cancer cell lines were cultured: MCF-7, MDA-MB-453 and HCC-1806. A dose-response study with irradiation was performed on the cells to determine a radiation absorbed dose and time-point after irradiation for the following combination treatment. Lastly, cells were treated with anti-cancer drugs +/– irradiation, and then stained with the fluorescent dyes Annexin V-DyLight 650 (apoptosis marker) and Hoechst 33342 (cell viability marker) whereafter flow cytometry was performed. Results. A dose and time dependency was observed in the dose-response study. In total, 29 of 31 anti-cancer drugs exhibited a synergistic effect with radiation on any cell line at any concentration. The drugs birinapant (with target IAP) and ganetespib (with target HSP90) showed statistically significant radiosensitizing effects on all three cell lines. Conclusions. Possible radiosensitizing targets for all three cell types included in this study are HSP90 and IAP. In addition, BH3 mimetics are potential radiosensitizing targets for cell lines MDA-MB-453 and HCC-1806, and the DNA repair machinery was a radiosensitizing target for cell line HCC-1806.
Preprocessing for intravoxel incoherent motion analysis in the brain - Signal drift correction
(2023-04-17) Warsame, Amina; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
Diffusion magnetic resonance imaging (dMRI) is a diagnostic imaging technique that is sensitized to the movement of water molecules in tissues. Intravoxel incoherent motion (IVIM) analysis can be used to estimate blood flow in capillaries from weakly diffusion-weighted data. dMRI images can be impacted by confounding factors and are usually preprocessed to improve their quality. This study reviewed the preprocessing pipeline for dMRI in relation to IVIM analysis in the brain. The results showed that the necessary preprocessing steps for IVIM analysis can be challenging due to a lack of consensus in this field, and may vary depending on the data and circumstances. While various tools for preprocessing dMRI exist, most of these tools are designed for datasets that have stronger diffusion-weighted data and more diffusion-encoding gradient directions than those typically used in IVIM. The results showed that only a limited number of these preprocessing tools can be directly applied to IVIM datasets. Based on observerations in previously acquired data, an in-depth analysis of the effect of signal drift on IVIM data was also investigated, and two correction methods were evaluated: temporal correction and spatio-temporal correction. Results from the signal drift study showed that the temporal and the spatio-temporal correction methods can reduce or amplify the effects of signal drift in IVIM data. This may indicate that additional methods may be needed to fully correct for this issue and obtain reliable results from IVIM imaging. In conclusion, it is crucial to have a thorough understanding of the data and desired results in order to accurately correct data.
Brain morphometry in Parkinson’s disease
(2023-03-02) Olivia, Lönkvist; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
Abstract Purpose The purpose of this brain morphometry study was to examine the volumes of different regions of the brain by research participants with Parkinson’s disease. Method To carry out the study, MR (magnetic resonance) images from 956 research participants from the Parkinson’s Progression Markers Initiative (PPMI) were used. The research participants were divided into 6 different cohorts. Before the MR images were used in this study, they had gone through preprocessing and segmentation using MAPER, where 121 regions including background were segmented. Once the data were obtained, the segmented regions were converted to volumes. The investigation was divided into two parts, first it was studied whether ICV (intracranial volume) differed between the different cohorts. This was done with the help of a violin plot and two-sided t-tests for each cohort. In the second part of the study, the individual volumes in the brain were examined. To accomplish this, hypothesis generation was used on half of each group’s subjects and then the hypotheses were tested on the other half. Two-tailed t-tests were performed to examine statistical differences and similarities between the cohorts. Result In the study, it was seen that the ICV differed between the different cohorts and thus on the imaged person’s head. None of the tested regions in the PD (Parkinsons disease) and SWEDD (scans without evidence for dopaminergic deficit) cohort showed any significant volume difference. The other three cohorts had two to five brain regions that showed a significant volume difference Conclusion In this brain morphometry study, it was shown that a correction for ICV is needed to get comparable results for the different cohorts. For the individual volumes, the results obtained in this study do not agree with previously obtained results. In further studies, it would therefore have been interesting to divide the brain into fewer regions, to add together regions that belong together.
Monte Carlo evaluation of static and dynamic 6FFF treatments - Evaluation of dose distributions calculated with AAA, Acuros XB, and Collapsed Cone (RayStation and DoseCheck)
(2023-02-23) Blombäck, Alma; University of Gothenburg/Institute of Clinical Sciences; Göteborgs universitet/Institutionen för kliniska vetenskaper
There are several benefits to removing the standard flattening filter and using so-called flattening filter free (FFF) treatments in external radiation therapy with photons. The main advantage is the possibility of significantly shortening the treatment time, de spite this FFF is not widely used yet. This thesis evaluates static and dynamic 6FFF dose distributions calculated with algorithms AAA, Acuros XB, and two implementations of Collapsed Cone (RayStation, and DoseCheck) with Monte Carlo calculations. A Monte Carlo 6FFF accelerator model was first validated in a water phantom and a previously developed Monte Carlo system was calibrated for 6FFF. Eight clinically planned patient plans, including prostate plans, brain plans, and plans in the thorax region, as well as one lung phantom plan were recalculated with all algorithms. The resulting dose distri butions were then evaluated against Monte Carlo using DVH evaluations. It was found that Acuros XB had the best agreement with Monte Carlo followed by AAA, both algo rithms having a comparable agreement as previously evaluated 6X plans. DoseCheck’s implementation of Collapsed Cone was found to underestimate the dose and RaySta tion’s differed the most from Monte Carlo. In conclusion, this study suggests that dose distributions calculated with AAA and Acuros XB do not lose accuracy when the flat tening filter is removed. Further, the algorithms used in DoseCheck and Raystations should be more investigated.

Browse

Recent Submissions