Malaria remains a leading cause of mortality, particularly in regions with limited access to healthcare. Despite the availability of diagnostic methods such as microscopy, Rapid Diagnostic Tests (RDT), and Polymerase Chain Reaction (PCR), these methods face challenges in accuracy and practicality, especially in field conditions. This thesis investigates the use of automated, Computer-aided diagnostics (CAD) to improve in-the-field malaria diagnosis. While CAD shows promise, significant challenges remain in dealing with image imperfections and data variability. The study focuses on the use of blob detectors and Zernike decomposition to improve the precision of the CAD system. Additionally, data quality is investigated to evaluate the reliability of malaria diagnostic pipeline. The results suggest that Zernike decomposition is effective in identifying shapes most likely to correspond to parasites, enhancing the blob detector’s precision. Furthermore, the use of data stratification is shown to reduce the variance between models during training, which indicates that it can generalise better to unseen data. This work contributes to the ongoing effort to develop a robust, field-deploy-able malaria diagnostic tool by incorporating prior knowledge of the malaria parasite its shape to improve precision and ensuring reliability by evaluating data quality.

Fast and accurate diagnosis of illnesses or other health complications is not accessible in many locations around the world. Due to this, illnesses are unnecessarily left untreated. Therefore, AiDx Medical has developed a portable automated diagnostic microscope with re liable and rapid AI-assisted detection specifically for low-resource settings.

Scan speed is a key issue for the popularisation of whole slide imaging systems [1]. To address this issue a recent paper by [29] proposed a Kalman filter-based scanning algorithm. This approach eliminates the necessity for focus map generation prior to scanning and in doing so, reduces the scan time. Importantly, the proposed approach requires no additional hardware, and is more robust to noise.

In this thesis, two modifications to this work are proposed. Firstly, higher order process models are used to generate more precise estimates of the best-in-focus positions. Secondly, a two-dimensional Non-Symmetric Half Plane Kalman filter is developed to incorporate neighbouring state estimates in the prediction – an approach previously thought inapplicable for this purpose [24]. In a simulation, the new scanning algorithms are applied to scan thin smear malaria specimens and compared to state-of-the-art focus map surveying procedures.

There is an ever-increasing demand for high data transfer rates. Laser satellite communication is a promising technology which offers high data rates and more secure links than the mature radio frequency (RF) communication. To make a laser communication terminal market viable it must be low size, weight and power (SWaP). The telescope drives the volume of the terminal. Making the telescope more compact with high magnification imposes stricter tolerances. A solution to alleviate these tolerances is to use a refocusing mechanism (RFM) to align the telescope in space after launch. In this project, a refocusing system is designed. This comprises of making system level choices like where to place the mechanism, how to close the loop and how to use the onboard detectors. All these choices lead to the requirements for the RFM. Further, a mechanism is designed that fulfills the requirements. In addition, mechanical and thermal FEM simulations are carried out to show that the RFM survives launch and can work in the challenging thermal environment.

Phase Retrieval Problem is very common and has a lot of applications in fields such as microscope, astronomy, crystallography, optical imaging etc.. In this problem, PSF and the pupil function are given, and the goal is to reconstruct the phase of the object. For camera which has 8 bits color depth, the maximum number of colors that can be displayed at any one time is 2^8 , i.e. 256. When the minimal exposure time of the camera is longer than "correct" exposure time or the intensity is changing, e.g. dynamic focusing, too much intensity is measured by the camera and causes saturation, i.e. over-exposure. Over-exposure causes distortion of the image, thereby losing the information. However, on the other hand, the overall brightness of the PSF image will be enhanced due to over-exposure. It highlight structure and increase the number of informative pixels of the image. We can use this to enhance the information in the PSF image to offset the negative effects of noise or background in the image.
Gerchberg-Saxton algorithm is a solution to phase retrieval problem and is a nice example of projection based algorithm. The algorithm has a perfect performance in the case that the PSF is the modules square of the object, i.e. without noise or other preconditioners effect. It is demonstrated in [Nishizaki, 2019] that an over-exposure preconditioner can improve the performance for phase restoration of a machine-learning model based algorithm. In this master thesis, we propose modification of the model of PSF amplitude in Gerchbrg-Saxton algorithm to make the algorithm be able to cope with over-exposed PSF image. This approach is tested both by numerical simulation and with experimental data.

In the field of Adaptive Optics (AO) a new development has been introduced recently: the Wavefront Sensorless Adaptive Optics (WFSless AO) approach, which only uses camera measurements to optimize image quality and does not use a Shack-Hartman (SH) Wavefront Sensor (WFS) [Booth, 2006, Débarre et al., 2009, Hinnen et al., 2008]. In order to optimize the image quality of a point source or an extended object, the residual wavefront should be minimised as much as possible. The Model Based Approach [Booth, 2007, Linhai and Rao, 2011] is a very promising approach which can be used for this optimisation, resulting in a significant reduction in the amount of measurements. This approach uses an initial measurement and then excites all the modes that need to be optimised separately and takes a measurement of the intensity distribution after each excitation. Then, the Second-Moment (SM)’s of the measured intensity distributions are computed and used for the computation of the optimal control input, which is the coefficient vector in a Zonal representation of the wavefront. The control vector can be used to minimise the residual wavefront and thus optimise the image quality. However, in this approach it is assumed that the incoming wavefront does not change between the first and last measurement. This assumption is called the frozen window time and it is not valid for real-time problems. In [Lianghua et al., 2017] a solution was proposed for this problem, by using a precomputed independent set of modes, which allows a decoupling of the modes and computations and reduces the frozen window time. Unfortunately, this research introduces more measurements which leads to more delay. In this thesis a new approach will be presented, the Continuous Model Based Approach. This approach tries to solve both problems simultaneously: a reduction of the number of measurements and a reduction of the frozen window time. The number of measurements will be reduced by combining the excitation and optimisation step and the frozen window time will be reduced by using a two step optimisation, based on [Lianghua et al., 2017]. First, the validity of this approach will be proven mathematically and with simulations, then it will be tested in experiments. In order to make the step to real-time experiments, also an online calibration method will be introduced, proven and tested. This is also a new technique, based on the Model Based Approach [Linhai and Rao, 2011] and can be used to obtain the correlation matrix for any set of modes, focussed on actuator responses.

Ground based telescope imaging suffers from interference from the earth’s atmosphere. Fluctuations in the refractive index of the air delay incoming light randomly, resulting in blurred images. A deconvolution from wavefront sensing system is an adaptive optics system that measures the modes in which the light is corrupted (i.e. the wavefront) and corrects it using a process called deconvolution. The wavefront is measured using a wavefront sensor, which consists of an array of microlenses combined with an imaging sensor. Each microlens casts an image of the object unto the imaging sensor, resulting in a collection of images that are differently aberrated depending on their location on the sensor. Conventionally, the wavefront is calculated by measuring the shifts of each microlens image and integrating these shifts over the aperture. This method, however, discards information about the higher order deformations of the microlens images.

In this thesis, a novel method of wavefront reconstruction has been developed which makes use of artificial neural networks in order to extract this higher order information. In order to do this, the images produced by the microlenses are normalized, which is done using a modified version of the blind deconvolution algorithm called TIP. After the normalization, the microlens images are reduced to what they would look like if a point source was observed, instead of the object. With the influence of the object removed, an artificial neural network is used for the estimation of the wavefront.

By using this method, the wavefront can be reconstructed with twice the turbulence strength compared to what is possible with conventional methods. Combining this method with an image deconvolution step results in a real-time image correction system that works up to 10Hz on the tested system, consisting of a desktop PC with an Intel Xeon E5-2630 DUAL CPU and a NVIDIA GeForce GTX 970 GPU.

Parasitic diseases such as malaria remain a mayor burden on global health. One of the biggest challenges still to be overcome is that of inadequate diagnoses. This research explores the opportunities that Hyperspectral Imaging yields in this field. The first goal is to estimate the spectral signature of Malaria parasites in non-stained or Giemsa-stained thin smear blood samples and of Schistosoma parasite eggs in urine samples. For this different endmember extraction algorithms are combined with various methods of pre-processing and dimensionallity reduction. The used endmember extraction methods are pure pixel index (PPI), NFINDR, Statistics Based and simplex identification via split augmented Lagrangian (SISAL). For denoising Savitzky Golay and 3 dimensional gaussian filtering is used and the dimensionallity reduction is done with PCA, ICA or HySime. The resulting spectral signatures of the algorithms are validated by inspecting the endmember locations, spectra and abundance maps. They have furthermore been compared by the classification performance where the spectral signatures are used in the feature derivation. This is done by deriving a detection map using OSP or CEM detection and then using the SVM or random forest classifiers to classify cells as being infected or not. These performances are furthermore compared to RGB image based classification.

In case of the stained Malaria sample the four endmember extraction methods are shown to be applicable to various degrees. Firstly, the PPI method is shown to be inconsistent, resulting in different spectra each run. Secondly, the statistics based method unable to separate the spectral signatures of the red blood cells and thirdly the background. Thirdly, The NFINDR method seems to work well considering the endmember locations, spectra and abundance maps, but leads to a low classification performance. The research concludes that Sisal made the most accurate estimations of the spectral signature of the parasite. The results from all the validation methods are in line with expectations. Furthermore, the use of this spectral signature in the feature derivation process results in the highest classification performance. This performance is also shown to be significantly higher compared to using either the first principal component of the full hyperspectral data or the RGB images. Applying the same methods to the Schistosoma sample it is found that some of the methods, though interestingly not Sisal, are able to to create an abundance map in which the egg is separated from the background. However, none of them are able to separate the egg and the white blood cell and detection maps using these signatures did not show the egg more clearly than the first principal component did. None of the methods are found to be able to extract the spectral signature of the unstained Malaria parasite.

Finally, a hypothetical multispectral microscope is proposed which images at the wavelengths where the spectral signature of the parasite in a stained sample has the biggest difference in light transmittance to the other endmembers. This setup is simulated from the available hyperspectral data and its classification performance is compared to classification performance using the full hyperspectral data and using the RGB images which are simulated from the same data. The classification using the discriminative wavelengths is found to outperform both in terms of sensitivity and specificity. This implies that the images at these specific wavelengths provide more discriminative power and such a multispectral setup could provide a significant advantage over RGB imaging.

To meet the demanding requirements on the environmental impact of aircraft, radically new aircraft concepts need to be developed. Within the NOVAIR project, Royal Netherlands Aerospace Centre (NLR) tests these new concepts on a Scaled flight demonstrator (SFD). Using an SFD allows for testing of dynamic and flight physical behavior of new aircraft concepts, which is difficult with more theoretical methods. Furthermore, by using an SFD, the risks associated with full-scale testing in terms of cost and time are minimized. One of the new concepts developed for the SFD is Distributed electric propulsion (DEP). Here, the two jet engines are replaced by six electric propellers. As these can be used actively for control, this results in an over-actuated system. These propellers interact with the aerodynamics of the wing resulting in Propulsion airframe interaction (PAI) effects. Although using the PAI effects for control has the potential to improve capabilities, efficiency and robustness of the aircraft, research into controllers using these effects actively is limited.

This thesis, therefore, presents a new control method including control allocation for the DEP-SFD aircraft, based on the nonlinear Incremental nonlinear dynamic inversion (INDI) controller. INDI enables controlling the nonlinear dynamics of the DEP aircraft over the complete flight envelope with one controller. By feeding back real-time sensor measurements, robustness to modeling errors and external disturbances is increased. Using the Incremental nonlinear control allocation (INCA) method, the full control authority of the DEP can be used. This technique enables taking into account nonlinear allocation relations and control effector interactions, while solving the control allocation real-time, which is key in actively using the PAI effects for control. The INCA method is used for two performance improvements: tracking performance and propeller power efficiency. To compensate for actuator dynamics, this thesis implements an Model predictive control (MPC) controller which results in improved tracking and higher efficiency. The performance of the controller was analyzed in simulation, where a reference square signal input on the roll angle was applied, while minimizing the sideslip angle and maintaining a steady altitude and velocity. The INCA controller with MPC is compared to a conventional INDI controller, showing a significant decrease in rise time from 2.46 s to 0.703 s with minimal tracking error. Furthermore, the effective bandwidth of the system was increased from 0.186 Hz to 0.663 Hz and the power consumption reduced by 6.3%. Modeling uncertainties, external disturbances and a propeller fault were introduced to verify the robustness of the controller. Finally, the reference altitude and velocity were varied, demonstrating controller performance over a large part of the flight envelope.

Large scale, highly sensitive and specific diagnostic tools are needed to eliminate the transmission of schistosomiasis. This report expands upon the novel approach from M. Hoeboer and P.M. Nijman. Their Smart Optical Diagnostic Of Schistosomiasis device (SODOS), combining the principles of flow cytometry and digital holography, was able to detect schistosomiasis haematobium eggs in a 10 ml urine sample. However the mechanical complexity made it difficult to control the flow and more than 650 frames needed to be processed for each sample. The reason for this was the small flow cell volume examined within each frame. This report shows it is possible to increase the volume while maintaining the required resolution for accurate diagnosis. The size of the volume is limited by the sensor’s pixel pitch, the sensor area, the shutter time of the sensor, the thickness of the volume, the density of particles and to a lesser extent the distance of the volume to the sensor and the wavelength of the source. The density of particles in combination with the thickness of the volume proved to be
the main limiting factors. With to many particles the scattered light causes a cloudy, speckle pattern on the sensor, which does not give the desired resolution upon reconstruction. By using a 24MP APS­C CMOS colour sensor, present in the Canon EOS M50 camera, in combination with a larger volume it was possible to reconstruct eggs throughout 1.5 ml of urine using a single frame, decreasing the total required number of frames to 7. The volume thickness was 5 mm. Increasing the thickness to analyse the required 10 ml of urine in a single frame was not feasible due to scattering. However when
immersing the eggs in a Phosphate­buffered saline solution a 14 ml volume, with a thickness of 43 mm, could be analysed successfully, showing the possibilities of implementing the technique in other fields as well. The Fraunhofer particle field holography theory proved to be a valid model when making predictions regarding the resolution throughout the volume as well as providing information on the axial position of the particles.

Schistosomiasis is a neglected tropical disease (NTD) that affects around a quarter-billion people worldwide. Most of the infected people live in tropical regions, especially in sub-Saharan Africa, where there is limited access to diagnostic and other relevant medical facilities. The current reference standard diagnostic procedure - conventional microscopy - is a relatively inexpensive procedure to use on a large scale, but it still requires trained operators and an initial financial investment which are hard to procure and maintain in remote areas with inadequate facilities. Moreover, the diagnostic sensitivity of the procedure is modest and varies over a wide range. In this work, we present the development of an inexpensive diagnostic instrument for urinary schistosomiasis that is automated to scan, analyse and diagnose the disease, and quantify the level of infection. We explore the design of the device with an open-source philosophy in mind, to enable makerspaces and other interested parties to reproduce the device locally. The device is manufacturable for as little as €200. It adheres to the standard sample preparation and diagnosis procedure established by the World Health Organisation (WHO), and images the relevant biomarkers to an adequate resolution for automated detection, diagnosis, and quantification for epidemiological surveillance. We believe this device can be an essential means for point-of-care diagnosis in resource-limited settings.

Reliable malaria diagnosis techniques that are suitable for point-of-care testing in high burden areas, are vital for effective treatment and monitoring of the disease. Identification of malaria parasites in Giemsa stained blood slides is currently the most widely accepted technique, but its availability is limited by the need for highly trained experts to interpret the data.
In this work, a two stage automated image classification strategy is proposed, to eliminate this dependency on human expertise. Blood slides that were photographed at 20 X magnification were used in our experiments, allowing for a larger Field of View than regular thin film microscopy at 100 X.
Erythrocytes are first localised and segmented by a Convolutional Neural Network, the architecture of which is based on U-Net, with some adaptations and improvements made for our purposes. The sensitivity and positive predictive value of the localisation were both 0.998, resulting in accurate cell counts.
A transfer learning strategy, in which the existing VGG-16 network is used as a feature extractor and combined with a new fully connected layer to predict correct activations for our classification, is then used to classify the segmented erythrocytes as either infected with Plasmodium Falciparum parasite or healthy. Sensitivity and specificity of the predicted classification were 0.795 and 0.915 respectively.
It is concluded that, although this method may not fully eliminate the need for trained experts, the algorithms proposed can be of great assistance in aiding the diagnostic decision making process.

Proper diagnostics are essential in the combat against severe diseases which mainly have big impacts in remote areas in poor countries. A focus direction within the NC4I group at DCSC, Delft University of Technology, is the development of new imaging modalities and the design and implementation of smarter algorithms for improved detection of parasitic diseases. The first part of my research exploits hyperspectral imagery (HI) as new potential imaging modality of thin blood smears that could highly improve on preparation time, labor intensiveness and use of materials. HI retrieves both spatial and spectral information of the observed objects simultaneously, thus providing the ability to discriminate near similar constituents within the blood smear. In doing so, it enables the possibility of label-free detection. In this thesis, the development and building of such a system is addressed and carried out. In the context of malaria, it is shown that HI is promising and lays a profound foundation for further exploration. The design and evaluation of improved generalizing neural networks characterize the essence of the second and larger part of the research. Several group-equivariant networks are evaluated and compared with conventional convolutional networks which shows that efficient and redefined integration of weights can help build smarter and more robust classifiers for the detection of parasites. In group-equivariant networks, re-interpreting the way feature maps are connected to one another manifests in the development of convolutional stages that equivary under an increased amount of transformations besides merely translations. It is shown that enlarging the heuristic of that transformation group (the extra amount of transformations the operations are equivariant under) significantly contributes to better performance without necessarily increasing the size or changing the architecture of the networks. Compared to the aforementioned baseline (conventional convolutional stages), the best network (being equivariant under 16 equidistant rotations and mirror reflections) improves approximately 2-fold on all relevant performance metrics, among which are accuracy, sensitivity, specificity, precision, and the F1-score which are common measures in the classification of malaria. The networks were tested on the Rajaraman database. Furthermore, the pre-trained models are used as classifiers for a different database extracted from the microscope build by AiDx medical. At least for this specific database, it is shown that the more realistic transformations the pre-trained networks equivary under, the more robust they are.

This thesis discusses developments towards a low-cost Fourier Ptychographic microscope for label free imaging of malaria parasites.
A review of the morphology and life cycle of malaria and the main diagnostic methods for its detection is followed by an introduction to Fourier Ptychography with emphasis on the underlying imaging principles and phase retrieval algorithms which are at the core of the algorithms used.
The practical realization of the Fourier Ptychographic setup with the required resolution has proven to be very challenging due to its susceptibility to errors when operating the system at its theoretical limits.
Insights from in-depth analyses of the effects of quantization noise, intensity drop-off due to angled illumination, and partial coherence are presented. These insights rule in- or out these potential error sources and help identify potential mitigations in the design.
In the final chapters the realization of the setup is described, and the results with real blood smear samples are used to illustrate the interference of the error sources. The thesis concludes with considerations for further research and recommendations for international collaboration.

Hyperspectral imaging (HSI) is a promising imaging modality in medical applications, especially for non-invasive and non-contact disease diagnosis and image-guided surgery. Encoding both spatial and spectral information, it can detect subtle changes in the biochemical and morphological properties of a tissue, revealing the early progression of a pathological condition like cancer. Previous medical hyperspectral image analysis approaches depended on handcrafted features or feature extraction principle, requiring considerable domain expertise. To overcome this, automatic feature learning approaches like convolutional neural networks (CNN), previously used in tasks like classification, detection and segmentation in medical images were applied to hyperspectral data, although in a limited number of research studies. This thesis was proposed to review the state-of-the-art in medical hyperspectral image analysis, identify the limitations in current methods, and present a proof-of-concept for using limited hyperspectral image data in CNN-driven tissue characterization.



The goal of this thesis is to characterize, using CNNs, ex vivo head and neck (tongue) tissue of patients affected by tumors. While previous work in this field implemented patch-based classification of tissue, in this thesis, a pixel-wise classification approach was proposed to obtain a smooth and continuous segmentation of hyperspectral images. To this end, two types of CNN models were trained from scratch using limited labelled training data, one to automatically learn the spectral features present in the hyperspectral data and the other to learn the combined spectral-spatial features from the hyperspectral data. Performance of four different trained models was evaluated by using a leave-one-out testing scheme, with the spectral-spatial learning approach with larger input spatial dimensions outperforming the other considered approaches.

Schistosomiasis is an intravascular infection with major public health consequences in developing countries. It is one of the major Neglected Tropical Diseases with more than 240 million people infected and 800 million people at risk in 2015, mostly in sub-Saharan Africa. It is caused by trematode parasites of the genus Schistosoma, in this report the focus was on Schistosomiasis Haematobium since it is the most prevalent form of the disease. One of the limiting factors of the control program is the standard diagnostic procedure set by World Health Organization, which is based on counting the parasite's eggs in a person's urine. Examination by microscopy requires the use of expensive microscopes, is prone to human errors and inconsistency, is time consuming, and uses filters which are often not available. The research objective was identified from these shortcomings: "Develop a low-cost, smart diagnostic method for Schistosomiasis Haematobium based on detecting eggs in urine by combining lensless imaging and flow cytometry, and developing Artificial Intelligence models for automated detection." In-line planar wavefront digital holography was identified as the most suitable lensless imaging method. A sample will be analyzed by the following repetitive procedure: (1)mechanically press the piston of a syringe by a small volume, (2)wait for the flow to stop, (3)record a hologram, (4)detect eggs. The implemented egg detection procedure consisted of a series of image processing algorithms: (1)apply Foreground detection, (2)localize the moving objects with a Blob detector, (3)locally reconstruct the hologram at the found locations, (4)classify the reconstruction as egg or not egg. The imaging method provided accurate reconstructions of eggs and the object detection algorithm was able to locate moving objects with sufficient accuracy and computational time. On the other hand, the lab and field tests showed that the data set of the classifier did not contain enough images to train a generalized model and that the local reconstruction and classification takes increasingly more time during analysis. As of now the method is an order of magnitude slower than an expert microscopist. The diagnostic method is not yet able fulfill the research objective. However, there are some promising aspects such as the low-cost imaging method, fast object detection algorithm, and absence of sample preparation which makes further research worthwhile.

When acquiring information using imaging techniques, one wants to achieve the best possible resolution in order to obtain valuable results. In the ideal scenario the size of the smallest resolved object features is limited by the wave properties of light, with a number of techniques developed to overcome this limit. In real life measurements the image quality is additionally deteriorated by the non-uniformities in refractive index along the beam propagation path, called aberrations. Adaptive Optics, introduced in 1950’s independently by Horace Babcock and Vladimir Linnik is a set of methods used to overcome the effect of aberrations and thus improve the image quality. Initially used in military applications for tracking satellites, it later found civilian applications as advances in computing made it practical.
Adaptive Optics use principles of feedback control to correct aberrations, and thus a measurement of controlled variable is required. The most commonly measured quantity is first derivative of wavefront provided by the Shack-Hartmann sensor. Another possibility is to measure wavefront Laplacian, i.e. curvature, using the sensor proposed by Roddier in 1988. With the correct use of this sensor the deformable mirror actuators are decoupled from each other, enabling fast and accurate control of high order systems and potentially analog feedback loop.
In this thesis, the Roddier curvature wavefront sensor is studied. A number of modifications in the original sensor geometry and operating principles are made in order to make the system more robust and easy to implement in laboratory environment. Performance of the modified curvature wavefront sensor is evaluated in both simulation environment and physical setup.
Finally, application of the curvature sensor to measure wind velocity is demonstrated.