Background
Vulvar squamous cell carcinoma accounts for 90% of vulvar cancers and is primarily treated by surgical excision. To reduce the risk of recurrence, surgeons aim to remove tumor-free margins of 2–3 mm, which can lead to extensive tissue loss, among others. Accurate intraoperative margin assessment is critical, but it currently relies on the surgeon’s experience and judgment, as well as time-consuming frozen section analysis, both of which are limited in precision.
Hyperspectral imaging offers a promising solution. This technique captures both spatial and spectral tissue information. It is a non-contact, non-invasive technique that is based on the interaction of light with the target object. However, implementation remains limited, as the current systems are restricted to image capture only.
This study aims to bridge the gap between hyperspectral imaging data and its practical use in surgery. The primary goal is to design and evaluate an interface that supports intraoperative decision-making for gynecological oncologists, ensuring the medical specialist remains in control. Since interpretability is essential, the second goal is to enhance the transparency of the HSI classification model by applying an explainable AI method.
Design Process
Research was conducted concerning the user, the context in which the user operates, existing medical interfaces, and optimal visualization of clinical data. A user survey and usability test were conducted to evaluate the prototype, which was developed through an iterative design process. The research produced a user persona, a context and task analysis, insights into medical interface layouts, and guidelines for visualizing clinical data. These aspects formed the basis for the requirements and wishes for the interface, in conjunction with the principles of the design frameworks applied. The user survey (n = 40, with a medical background) provided design direction and recommendations. The usability test was designed to evaluate the overall workload of the designed interface and its intuitiveness. The test was performed on nine participants.
The mean accuracy of all the tasks performed was 88%, indicating that the interface is straightforward to use. NASA-TLX scores ranged from 6.7 to 24 across six parameters, suggesting low to moderate cognitive load. Semantic differential results showed high ratings for user-friendliness (7.8), logical structure (8.2), and color aesthetics (7.9). However, improvements are still needed in visualizing classification outcomes and tissue parameters, which were partially addressed in the redesign.
Data Analysis
As a first step in distinguishing healthy from tumor tissue, hyperspectral data was collected intraoperatively before tumor excision. The data was then preprocessed by data calibration, glare removal, and data extraction. The preprocessed data was classified with a Support Vector Machine. This classification model was first optimized by tuning the kernel type and then validated through different evaluation metrics. To interpret the model’s decision-making, SHapley Additive exPlanations was applied to identify which features contributed most to the classification. Three types of input features were assessed: wavelengths (500–1000 nm), visual fractions, and tissue parameters.
The tuning of the kernel type yielded the Radial Basis Function kernel as the optimized model, with an average area under the curve of 1 and an accuracy of 99%. The SHapley Additive exPlanations analysis of the spectral dataset showed that all wavelengths consistently support the prediction for healthy tissue and oppose in predicting tumor. Suggesting that the model is biased toward classifying samples as healthy, and tumor is identified primarily by the absence of features associated with healthy tissue. Notably, the 500–595 nm range and the near-infrared region were most influential in supporting prediction for healthy tissue and opposing tumor.
Using spectral fractions and tissue parameters as input features did not yield accurate enough results(macro AUC = 0.65 and accuracy is 10%) to apply SHAP. This suggests that these features alone do not provide sufficient discriminatory information. This displays the need for additional spectral data to improve model performance for these input features.
Conclusion
The results of this study indicate that the initial probabilistic interface design is relatively intuitive, requiring only low to medium cognitive effort and workload. The application of explainable AI revealed that the 500–595 nm range and the near-infrared region are most influential in predicting healthy tissue versus tumor. This improves the transparency and interpretability of the classification model for gynecological oncologists. Together, the intuitive probabilistic interface and explainable AI support the integration of hyperspectral imaging into surgical practice.

Introduction Healthcare systems are increasingly integrating advanced technologies into daily practice, leading to a growing emphasis on the adoption of innovative high-tech medical devices. Unlike conventional device replacements, these innovations introduce new technologies that can significantly impact clinical workflows, patient outcomes, and organizational structures. However, the complexity and uncertainty surrounding their adoption present major decision-making challenges for hospitals. Despite the critical nature of these decisions, most hospitals lack a structured approach to guide them. This thesis addresses this gap by developing a decision-making framework to support hospitals in the acquisition of innovative high-tech medical devices.

Methodology This study employed a three-phase approach: analyzing the current decision-making process for acquiring innovative high-tech medical devices, developing a decision-making framework, and evaluating the framework. Semi-structured interviews with stakeholders from two hospitals were conducted to map current practices and identify challenges. Insights from the interviews and a successful case study informed the development of the framework, designed using principles from the Cynefin framework to address the complexity of innovation adoption. The framework was refined based on expert feedback and evaluated through a semi-structured questionnaire focusing on its structure, usability, and expected effectiveness.

Results The results of this study revealed several challenges in the current decision-making processes for acquiring innovative high-tech medical devices including limited exploration of broader organizational needs, premature formation of project groups and unclear early-stage leadership. The interviews and the case study highlighted the importance of iterative decision-making, early project leadership by a technically skilled project leader, and flexibility in adapting project structures as new insights emerge. Based on these findings, a decision-making framework was developed that addresses these challenges by promoting adaptive, stakeholder-driven, and strategically aligned acquisition processes. Both respondents of the evaluation questionnaire agreed that the framework is clearly structured, adds value to the hospital’s decision-making process, and supports a well-informed investment decision. Nevertheless, some limitations were identified.

Discussion The developed decision-making framework approaches the acquisition of innovative high-tech medical devices as a complex, iterative process aimed at uncovering broader organizational needs behind device requests. It emphasizes early stakeholder involvement, flexible project structures, and delayed formalization of business cases to better navigate uncertainty. While the framework offers hospitals a structured but adaptable tool to professionalize decision-making and foster innovation, its development was based on interviews at only two hospitals and has not yet been tested in real-world applications. Future research should focus on validating the framework through longitudinal case studies and further refining its usability with detailed guidance and practical examples.

Accurate needle placement is essential for effective cervical high dose-rate (HDR) brachytherapy in anatomically complex or asymmetric tumor regions. Conventional rigid obturators used with standard applicators, such as Elekta's Geneva and Venezia systems, lack the flexibility to reach off-axis targets, thereby limiting treatment precision. This study presents the development and validation of a manually operated steerable interstitial needle for gynecological brachytherapy. The device employs a compliant split-shaft mechanism that converts differential push-pull actuation into controlled tip deflection. Finite element analysis guided the selection of Grade 5 titanium and informed geometric optimization to balance flexibility with mechanical stability. A prototype was fabricated and integrated into a 3D-printed handheld actuation unit compatible with the Elekta Venezia applicator. Experimental evaluation in gelatin and silicone phantoms under computed tomography (CT) imaging demonstrated tip deflections of up to 33.0 mm with the Flexible Implant Tube 6F and up to 18.5 mm with the ProGuide 6F at an insertion depth of 80 mm. Placement accuracy remained within ± 5 mm in over 95\% of the trials. These results validate the steerable needle concept and highlight its potential to improve maneuverability during clinical brachytherapy.

Brachytherapy is a type of internal radiation therapy that is used to treat cervical cancer. It involves the application of a radioactive source in close proximity to the tumour. This can be done either by directly inserting the source into (or close to) the tumour using interstitial hollow needles, or by loading the source into an intracavitary applicator that is placed in the vaginal cavity. Standard applicator types may lead to suboptimal radioactive source placement, possibly resulting in underexposure of target volumes and overexposure of organs at risk, especially in advanced cancers. Customised applicators and optimised needle channels based on the patient's MRI/CT data could enhance conformity between target volumes and prescribed isodose. Hence, the goal of this project was to develop and validate a design for a 3D-printed brachytherapy applicator which geometry is based on the patient's vaginal cavity and which features optimised interstitial needle channels based on the patient's anatomy and tumour location.

Various analyses have been conducted which have led to the establishment of a list of requirements for the applicator design. Based on this list of requirements, two conceptual designs have been presented: one fully 3D-printed design and one design that is clicked on the Geneva ovoid tubes. Through the creation of prototypes, these conceptual designs have been refined into two final designs which were manufactured in PA12 using selective laser sintering. The dose attenuation properties of PA12 were evaluated and compared to that of water. Furthermore, the potential needle positions within the proximal end of both designs have been analysed. For both designs, a final prototype based on a phantom's vaginal cavity geometry has been created. The usability of these prototypes has been tested by three radiotherapist-oncologists, who also provided feedback on the designs. Upon analysing their feedback and the outcomes of the other evaluations, recommendations for future designs have been formulated.

The conducted dosimetry experiment yielded a maximum difference of 0.8% between the average percent dose depth curves of water and PA12, which can be considered a water-equivalent response. This allows PA12 to be used as the material for the applicator. The result of the potential needle position analysis suggest that the first design provides more space for personalised needle channels in the top of the applicator compared to the second design. The three radiotherapist-oncologists validated the usability of both final prototypes.

Two designs of a patient-tailored 3D-printed brachytherapy applicator containing optimised interstitial needle channels based on the patient's anatomy and tumour location have been presented, produced and validated. Based on the outcomes of the conducted evaluations, there can be concluded that the first concept shows the most promise to be used as a design for a patient-tailored 3D printed brachytherapy applicator. However, to ensure the proper functioning of the working principles, further development is required. If the recommended improvements are implemented, the design has the potential to be used as applicator in the treatment of cervical cancer.

The technique of ultrasound-guided needle insertion is commonly employed in various clinical fields, including biopsy, anesthesia, brachytherapy, and ablation. However, the visibility of the needle in ultrasound (US) images remains a persistent challenge. To improve the guidance accuracy of needle insertion during interventions, it is crucial to develop a reliable technique that enhances the visibility of ultrasound needles and accurately detects their position and orientation. Recently, Deep Learning (DL) based segmentation methods have drawn attention because of their high efficiency and accurate results. In order to improve the model performance on challenging datasets, previous researchers modified the segmentation models by introducing spatial attention mechanism and temporal information. However, whether the approaches are effective for ultrasound images, especially for the needle insertion tasks remains unclear.

This thesis aims to investigate whether deep learning models are able to increase segmentation accuracy as well as localization accuracy in 2D ultrasound images, specifically focusing on introducing spatial attention and optical flow information into U-Net backbone. Spatial Mask Attention U-Net (SMA-UNet) and Optical Flow Attention U-Net (OFA-UNet) were therefore proposed. The hierarchical experiments were designed to evaluate the effects of training loss, mask width and optical flow methods, and then select an optimal configuration for the segmentation models. Furthermore, U-Net, Attention U-Net and two proposed models were validated on datasets collected from pork and beef phantoms, as well as patients. The evaluation results indicate that OFA-UNet has significant improvement in terms of segmentation metrics and geometrical errors compared to the U-Net baseline and the U-Net only considering the mask attention. Specifically, the model achieved Dice of 86.7%, IoU of 88.2%, Precision of 88.6%, tip error of 2.7 mm and angular error of 0.002 radians on the pork dataset. Furthermore, the OFA-UNet shows robustness and consistency in evaluation metrics across three different datasets, indicating its ability to adapt to varying complexities of US datasets.

The goal of this study is to create a phantom which can be used to test strategies for sealing the vaginal cavity, so that ultrasound gel can be used to fill the vaginal cavity improve the visibility of the vaginal cavity in MR-images.
Silicone rubber and PolyVinyl Alcohol (PVA) were tested, to check whether their mechanical properties can be used to mimic the bio-mechanical behaviour of Human Anterior Vaginal Wall (HAVW)-tissue. Both materials were used to produce a phantom. Contrast differences in MR-imaging of PVA-hydrogel and silicone rubber with ultrasound gel and air were measured, to verify proper contrast within a phantom. Both phantoms were used to test sealing strategies.
Young's moduli for HAVW-tissue were obtained from literature, and formed a range (6.7 - 14.4 MPa). Silicon proved a closer match to the HAVW-tissue, compared to PVA-hydrogel. Calculated Young's moduli were between 1.4 - 4.2 MPa (silicone rubber) and 0.04 - 0.16 MPa (PVA-hydrogel). Contrast differences in MR-images were measured between 177.8 and 1585.7, and proved sufficient enough to be spotted by an untrained eye. The sealing strategies using a probe cover surrounding a catheter, showed an increase of 54.3% (silicone rubber) and 119.9% (PVA-hydrogel) of the filling percentage measured. Indentation testing was confirmed as a possible alternative to tensile testing for the verification of mechanical properties. In this study the production process of a thin-walled double-layered phantom of the vaginal cavity has been developed and tested. The developed phantoms have been used to test strategies for sealing the vaginal cavity. The concept using a catheter surrounded by a probe cover, is considered to improve the sealing of the vaginal cavity.

Radiofrequency ablation (RFA) is a localized thermalintervention technique with the purpose to destroy small tumours (< 3 cm) byheating the tumour tissue. Due to various challenges with targeting andvisualisation during RFA, the clinical need of steerable needles is present forradiologists. Helwig, B.P. developed a new omnidirectional steerableneedle for this purpose. However, this needle is not yet ready forimplementation in the operation room (OR) because no clinical testing isperformed in vivo. Therefore, the aim of this thesis is to develop a novelneedle guide tool for percutaneous interventions with an omnidirectionalsteerable needle during real-time US-guidance. In order to examine the visibility of the steerable needle,an experimental study has been conducted. The steerable needle has a promisingvisibility in Polyvinyl alcohol and is expected to be well visible duringultrasound guided interventions. The developed needle guide tool is calledUShift and provides both the benefits of needle guided puncturing and free-handpuncturing. Which are for free-hand puncturing: freedom of movement of boththe probe and needle during puncturing and for needle guided puncturing:  a shorter procedure time, reduced needle manipulation and improvedneedle visualisation. A user test is performed to verify how the needle guide isperforming compared to the currently used needle guide while hypothetical usingthe Omnidirectional steerable needle. The difference between the two puncturetechniques is 0,9 seconds, the UShift is 23% faster. In conclusion, thisproject resulted in a novel needle guide tool that can be used in combinationwith the omnidirectional steerable needle during percutaneous interventions.The UShift has great benefits compared to currently used needle guides and istherefore stated as innovative. More gain is to be expected through future researchconcerning production methods, material options, the cost-effectiveness andother application fields of the UShift.

In recent years, more and more medical operations are done minimally invasive. Intervention radiology is a medical specialty which uses minimally invasive techniques to diagnose, or treat diseases. Instruments like needles and catheters are used by radiologists to enter the network of veins and arteries guided by image modalities. A complex treatment in the interventional radiology is the Transjugular Intrahepatic Portosystemic Shunt (TIPS) procedure. This treatment is developed for people who suffer from liver cirrhosis which are not eligible for liver transplantation. The problem which arise with a liver affected by cirrhosis is that it can not transmit enough blood. Without transplantation or a treatment this will eventually lead to death. During the TIPS procedure a connection is made between the right hepatic and the portal vein by a shunt. As a result, blood pressure reduction in the portal system since the blood can flow back to the right atrium of the heart, bypassing the liver.

The hardest part in a TIPS procedure is the intrahepatic puncture between the hepatic and portal vein. The interventional radiologist tries to enter the portal vein by puncturing a small stylet from the hepatic vein through the liver tissue. Due to cirrhosis the liver tissue is very stiff and stylet deflection will occur. To reduce the uncertainty of entering the portal vein, this thesis is focused on designing a stylet that is more stiff and able to steer. It is expected that the complexity of the procedure will be reduced and a higher hit rate to enter the portal vein will be achieved.\\
\\
The prototype of the steerable stylet has been evaluated through various experiments, a visibility test and with procedures in a test liver model. During these experiments the steering characteristics, the stiffness of the stylet, the maximum lateral forces exerted by the tip while steering, the influence of the stylet orientation and the visibility are obtained. Afterwards, an evaluation was done in a liver model, made of PVA, to determine whether the stylet is capable to reduce the complexity of the intrahepatic puncture step in the TIPS procedure.

With the prototype made in this graduation project, based on a steerable ablation needle, the complexity of the TIPS procedure is not reduced yet. The steerable stylet was not able to enter the portal vein. It was already hard to enter the right hepatic vein since the pre-bent stiffening cannula was adapted with a smaller angle which was necessary since the steerable stylet was too stiff to push through the pre-bent angle. According to this prototype, possibilities are shown to use a mechanical steering mechanism in instruments with a long thin shaft. The transmission in combination with the joint mechanisms fits within 1.3mm diameter, was able to bridge 60cm from distal end to proximal end and had only 4 components, the stylet, the rigid cannula, the key to fix the stylet to the rigid cannula and the transmission. By translating the stylet in a push or pull direction relative to the rigid cannula, steering angles could be achieved. With further research and development this steering mechanism must be able to steer the required amount of degrees without any extra components, is well visible with ultrasound, and is good resistant against lateral forces.

Medical Asset Management (MAM), track and tracing of medical instruments throughout a hospital, is very comprehensive and complicated. To implement such a system on individual instrument level, a technique that can create Unique Device Identification (UDI) codes is needed. Current UDI systems, such as RFID and barcode scanning, do not suffice. Adding a RFID tag to a large instrument is not a problem, but adding the tag to a small scissor can be problematic. Adding a barcode to an instrument can be done in multiple differ- ent ways such as; laser engraving, adding a sticker or adding a tag with a barcode on it. The tags tend to be removed by medical professionals and stickers can get loose. Laser engraving seems to be a good alternative, if no human material would stay behind in the slots after sterilizing.
Both techniques discussed above are not specific designed for Medical Asset Management. The goal of this thesis is to research if a different technique can be used for MAM; medical instruments identification via lu- minescent micro-particles coating. The main question is:
Is it possible to coat individual medical instruments with a coating via electroless nickel plating, with embed- ded luminescent micro-particles and identify these instruments?
To answer this question, this thesis is divided into 4 parts; Electroless nickel plating of stainless steel; Physic- ochemical characterization of the coatings; Identification of the substrates and Implementation. During the 8 experiments, 316L medical grade stainless steel substrates are coated and characterized. The identification part consists out of building a setup for a UV-C light and creating a Matlab file for automatically identification of the substrates. Medical instrument experts are consulted for the implementation part.
With the use of a nickel strike as a pre-treatment, a successful coating was embedded into the substrates. The Barium Magnesium Aluminate (BAM:Eu) particles were visible via the SEM analysis and was identified via a EDS analysis. Exciting the substrate with the UV-C light resulted in a clear blue substrate. Trying the same experiments with different particles, Yttrium-Oxide (YO:Eu), was not successful. The dispersion of the YO:Eu particles was different in comparison to the BAM:Eu particles. The YO:Eu particles eventually sank to the bottom or agglomerated on the top of the solution. The created Matlab script was successful in identifying both colors. During the discussion of the technique with medical instruments experts, questions arose about the change in material properties. Some properties do change, but the impact of this needs to be further researched.
This research proved the concept of identifying individual medical instruments via a luminescent micro par- ticle coating. It is possible to coat and identify a substrate with this technique.