<p dir="ltr">Progress in deep learning methodologies has transformed the landscape of medical image analysis, opening fresh pathways for precise and effective diagnostics. Currently, multi-phase liver CT scans follow a four-stage process, commencing with an initial scan carried out before the administration of <a href="" target="_blank">intravenous (IV) contrast-enhancing material</a>. Subsequently, three additional scans are performed following the contrast injection. The primary objective of this research is to automate the analysis and prediction of 50% of liver CT scans. It concentrates on discerning patterns of intensity change during the second, third, and fourth phases concerning the initial phase. The thesis comprises two key sections. The first section employs the non-contrast phase (first scan), late hepatic arterial phase (second scan), and portal venous phase (third scan) to predict the delayed phase (fourth scan). In the second section, the non-contrast phase and late hepatic arterial phase are utilized to predict both the portal venous and delayed phases. The study evaluates the performance of two deep learning models, U-Net and U²-Net. The process involves preprocessing steps like subtraction and normalization to compute contrast difference images, followed by post-processing techniques to generate the predicted 2D CT scans. Post-processing steps have similar techniques as in preprocessing but are performed in reverse order. Four fundamental evaluation metrics, including <a href="" target="_blank">Mean Absolute Error (MAE), Signal-to-Reconstruction Error Ratio (SRE), Peak Signal-to-Noise Ratio (PSNR), and Structural Similarity Index Measure (SSIM), </a>are employed for assessment. Based on these evaluation metrics, U²-Net performed better than U-Net for the prediction of both portal venous (third) and delayed (fourth) phases. Specifically, U²-Net exhibited superior MAE and PSNR results for the predicted third and fourth scans. However, U-Net did show slightly better SRE and SSIM performance in the predicted scans. On the other hand, for the exclusive prediction of the fourth scan, U-Net outperforms U²-Net in all four evaluation metrics. This implementation shows promising results which will eliminate the need for additional CT scans and reduce patients’ exposure to harmful radiation. Predicting 50% of liver CT volumes will reduce exposure to harmful radiation by half. The proposed method is not limited to liver CT scans and can be applied to various other multi-phase medical imaging techniques, including multi-phase CT angiography, multi-phase renal CT, contrast-enhanced breast MRI, and more.</p>

Abstract Biology and medicine rely heavily on images to understand how the body functions, for diagnosing diseases and to test the effects of treatments. In recent decades, microscopy has experienced rapid improvements, enabling imaging of fixed and living cells at higher resolutions and frame rates, and deeper inside the biological samples. This has led to rapid growth in the image data. Automated methods are needed to quantitatively analyze these huge datasets and find statistically valid patterns. Cell segmentation and tracking is critical for automated analysis, yet it is a challenging problem due to large variations in cell shapes and appearances caused by various factors, including cell type, sample preparation and imaging setup. This thesis proposes novel methods for segmentation and tracking of cells, which rely on machine learning based approaches to improve the performance, generalization and reusability of automated methods. Cell proposals are used to efficiently exploit spatial and temporal context for resolving detection ambiguities in high-cell-density regions, caused by weak boundaries and deformable shapes of cells. This thesis presents two cell proposal methods: the first method uses multiple blob-like filter banks for detecting candidates for round cells, while the second method, Cell Proposal Network (CPN), uses convolutional neural networks to learn the cell shapes and appearances, and can propose candidates for cells in a wide variety of microscopy images. CPN first regresses cell candidate bounding boxes and their scores, then, it segments the regions inside the top ranked boxes to obtain cell candidate masks. CPN can be used as a general cell detector, as is demonstrated by training a single model to segment images from histology, fluorescence and phase-contrast microscopy. This work poses segmentation and tracking as proposal selection problems, which are solved optimally using integer linear programming or approximately using iterative shortest cost path search and non-maximum suppression. Additionally, this thesis presents a method which utilizes graph-cuts and an off-the-shelf edge detector to accurately segment highly deformable cells. The main contribution of this thesis is a cell tracking method which uses CPN to propose cell candidates, represents alternative tracking hypotheses using a graphical model, and selects the globally optimal sub-graph providing cell tracks. It achieves state-of-the-art tracking performance on multiple public benchmark datasets from both phase-contrast and fluorescence microscopy containing cells of various shapes and appearances. / Tiivistelmä Biologia ja lääketiede nojaavat vahvasti kuvatietoon solujen ja kehon toimintojen ymmärtämiseksi sairauksien diagnostiikassa ja hoitojen vaikutusten seuraamisessa. Viime vuosikymmeninä mikroskopiassa on tapahtunut nopeaa teknistä kehitystä, mikä on mahdollistanut elävien solujen kuvantamisen tarkemmin, nopeammin sekä syvemmältä automatisoidusti useasta näytteestä. Tämä taas on johtanut kuvadatan nopeaan kasvuun ja suurempaan määrään biologisia kysymyksiä, joihin voidaan vastata. Kuvadatan räjähdysmäisen kasvun vuoksi kaikkia tuloksia ei voida enää tulkita pelkästään ihmistyövoimaa käyttämällä, mikä on johtanut tarpeeseen kehittää automaattisia menetelmiä analysoimaan kvantitatiivisesti suuria datajoukkoja ja löytämään tilastollisesti kelvollisia malleja. Solujen erottaminen niiden ympäristöstä ja toisista soluista (segmentointi) ja solujen seuranta ovat kriittisiä alkuvaiheen osia onnistuneessa automaattisessa analyysissä. Automaattisten menetelmien kehittämisessä solusegmentointi on kuitenkin osoittautunut hyvin haastavaksi ongelmaksi solujen muodon ja ulkonäön suurten muutosten vuoksi solutyypistä, näytteen valmistelusta ja kuvantamisjärjestelmästä johtuen. Tämä väitöskirja esittää uusia menetelmiä solujen segmentointiin ja seurantaan keskittyen koneoppimiseen perustuviin lähestymistapoihin, jotka parantavat automaattisten menetelmien suorituskykyä ja uudelleenkäytettävyyttä. Spatiaalista ja ajallista kontekstia tehokkaasti hyödyntäviä soluehdotelmia käytetään ratkaisemaan solujen heikosti erottuvista reunoista ja joustavista muodoista johtuvaa solujen muodon monitulkintaisuutta erityisesti silloin kun tutkittava solutiheys on suuri. Tämä väitöskirja esittää kaksi menetelmää soluehdotelmille; ensimmäinen menetelmä käyttää useita läikkätyyppisiä suodatinpankkeja ilmaisemaan kandidaatteja pyöreänmuotoisille soluille, kun taas toinen menetelmä nimeltään soluehdotelmaverkko (Cell Proposal Network, CPN) käyttää konvoluutionaalisia neuroverkkoja oppiakseen tunnistamaan solut niiden muodon sekä ulkonäön perusteella erityyppisissä mikroskooppikuvissa. CPN regressoi ensin solukandidaatteja ympäröivät suorakaiteet ja niiden pistemäärän, jonka jälkeen se segmentoi alueet parhaiten sijoittuneiden suorakaiteiden joukosta tuottaen solukandidaattimaskit. CPN:ää voidaan mahdollisesti käyttää yleisenä soluilmaisimena erityyppisilla kuvantamistekniikoilla tuotetuissa kuvissa mukaan lukien histologisen valo-, fluoresenssi- ja vaihekontrastimikroskooppian. Väitöskirja esittää solujen segmentoinnin ja seurannan soluehdotelmien valintaongelmina, mitkä ratkaistaan joko optimaalisesti käyttämällä kokonaislukuoptimointia tai likimääräisesti käyttämällä iteratiivista lyhimmän kustannuspolun hakua sekä ei-maksimien vaimennusta. Tämä väitöskirja esittää myös verkon leikkaukseen (graph cut) perustuvan menetelmän, mikä hyödyntää valmiiksi saatavilla olevaa reunanilmaisinta segmentoimaan tarkasti muotoaan voimakkaasti muuttavia soluja. Väitöskirjatutkimuksen keskeinen tulos on uusi solujen seurantamenetelmä, mikä käyttää CPN:ää solukandidaattien ehdottamiseen, esittää vaihtoehtoiset seurantahypoteesit verkkomallia hyödyntämällä, ja valitsee globaalisti optimaalisen aliverkon solujen kulkemille reitille. Verrattuna useisiin muihin julkisesti saatavilla oleviin kuva-analyysiohjelmistoihin tässä väitöskirjassa kehitetyt menetelmät olivat suorituskyvyltään parhaita vaihekontrasti- ja fluoresenssimikroskopialla tuotettujen kuva-aineistojen analyyseissa, joissa solujen ulkomuoto oli hyvin vaihteleva.

In ischaemic stroke a disruption of cerebral blood flow leads to impaired metabolism and the formation of an ischaemic penumbra in which tissue at risk of infarction is sought for clinical intervention. In stroke trials, therapeutic intervention has largely been based on perfusion-weighted measures, but these have not been shown to be good predictors of tissue outcome. The aim of this thesis was to develop analysis techniques for magnetic resonance imaging (MRI) of chemical exchange saturation transfer (CEST) in order to quantify metabolic signals associated with tissue fate in patients with acute ischaemic stroke. This included addressing robustness for clinical application, and developing quantitative tools that allow exploration of the in-vivo complexity. Tissue-level analyses were performed on a dataset of 12 patients who had been admitted to the John Radcliffe Hospital in Oxford with acute ischaemic stroke and recruited into a clinical imaging study. Further characterisation of signals was performed on stroke models and tissue phantoms. A comparative study of CEST analysis techniques established a model-based approach, Bloch-McConnell model analysis, as the most robust for measuring pH-weighted signals in a clinical setting. Repeatability was improved by isolating non-CEST effects which attenuate signals of interest. The Bloch-McConnell model was developed further to explore whether more biologically-precise quantification of CEST effects was both possible and necessary. The additional model complexity, whilst more reflective of tissue biology, diminished contrast that distinguishes tissue fate, implying the biology is more complex than pH alone. The same model complexity could be used reveal signal patterns associated with tissue outcome that were otherwise obscured by competing CEST processes when observed through simpler models. Improved quantification techniques were demonstrated which were sufficiently robust to be used on clinical data, but also provided insight into the different biological processes at work in ischaemic tissue in the early stages of the disease. The complex array of competing processes in pathological tissue has underscored a need for analysis tools adequate for investigating these effects in the context of human imaging. The trends herein identified at the tissue level support the use of quantitative CEST MRI analysis as a clinical metabolic imaging tool in the investigation of ischaemic stroke.

This thesis addresses three problems related to image registration, prediction and tracking, applied to Angiography and Oncology. For image analysis, various probabilistic models have been employed to characterize the image deformations, target motions and state estimations. (i) In Digital Subtraction Angiography (DSA), having a high quality visualization of the blood motion in the vessels is essential both in diagnostic and interventional applications. In order to reduce the inherent movement artifacts in DSA, non-rigid image registration is used before subtracting the mask from the contrast image. DSA image registration is a challenging problem, as it requires non-rigid matching across spatially non-uniform control points, at high speed. We model the problem of sub-pixel matching, as a labeling problem on a non-uniform Markov Random Field (MRF). We use quad-trees in a novel way to generate the non uniform grid structure and optimize the registration cost using graph-cuts technique. The MRF formulation produces a smooth displacement field which results in better artifact reduction than with the conventional approach of independently registering the control points. The above approach is further improved using two models. First, we introduce the concept of pivotal and non-pivotal control points. `Pivotal control points' are nodes in the Markov network that are close to the edges in the mask image, while 'non-pivotal control points' are identified in soft tissue regions. This model leads to a novel MRF framework and energy formulation. Next, we propose a Gaussian MRF model and solve the energy minimization problem for sub-pixel DSA registration using Random Walker (RW). An incremental registration approach is developed using quad-tree based MRF structure and RW, wherein the density of control points is hierarchically increased at each level M depending of the features to be used and the required accuracy. A novel numbering scheme of the control points allows us to reuse the computations done at level M in M + 1. Both the models result in an accelerated performance without compromising on the artifact reduction. We have also provided a CUDA based design of the algorithm, and shown performance acceleration on a GPU. We have tested the approach using 25 clinical data sets, and have presented the results of quantitative analysis and clinical assessment. (ii) In External Beam Radiation Therapy (EBRT), in order to monitor the intra fraction motion of thoracic and abdominal tumors, the lung diaphragm apex can be used as an internal marker. However, tracking the position of the apex from image based observations is a challenging problem, as it undergoes both position and shape variation. We propose a novel approach for tracking the ipsilateral hemidiaphragm apex (IHDA) position on CBCT projection images. We model the diaphragm state as a spatiotemporal MRF, and obtain the trace of the apex by solving an energy minimization problem through graph-cuts. We have tested the approach using 15 clinical data sets and found that this approach outperforms the conventional full search method in terms of accuracy. We have provided a GPU based heterogeneous implementation of the algorithm using CUDA to increase the viability of the approach for clinical use. (iii) In an adaptive radiotherapy system, irrespective of the methods used for target observations there is an inherent latency in the beam control as they involve mechanical movement and processing delays. Hence predicting the target position during `beam on target' is essential to increase the control precision. We propose a novel prediction model (called o set sine model) for the breathing pattern. We use IHDA positions (from CBCT images) as measurements and an Unscented Kalman Filter (UKF) for state estimation. The results based on 15 clinical datasets show that, o set sine model outperforms the state of the art LCM model in terms of prediction accuracy.

The analysis of wide-angle cellular light scattering patterns is a challenging problem. Small changes to the organization, orientation, shape, and optical properties of scatterers and scattering populations can significantly alter their complex two-dimensional scattering signatures. Because of this, it is difficult to find methods that can identify medically relevant cellular properties while remaining robust to experimental noise and sample-to-sample differences. It is an important problem. Recent work has shown that changes to the internal structure of cells---specifically, the distribution and aggregation of organelles---can indicate the progression of a number of common disorders, ranging from cancer to neurodegenerative disease, and can also predict a patient's response to treatments like chemotherapy. However, there is no direct analytical solution to the inverse wide-angle cellular light scattering problem, and available simulation and interpretation methods either rely on restrictive cell models, or are too computationally demanding for routine use. This dissertation addresses these challenges from a computational vantage point. First, it explores the theoretical limits and optical basis for wide-angle scattering pattern analysis. The result is a rapid new simulation method to generate realistic organelle scattering patterns without the need for computationally challenging or restrictive routines. Pattern analysis, image segmentation, machine learning, and iterative pattern classification methods are then used to identify novel relationships between wide-angle scattering patterns and the distribution of organelles (in this case mitochondria) within a cell. Importantly, this work shows that by parameterizing a scattering image it is possible to extract vital information about cell structure while remaining robust to changes in organelle concentration, effective size, and random placement. The result is a powerful collection of methods to simulate and interpret experimental light scattering signatures. This gives new insight into the theoretical basis for wide-angle cellular light scattering, and facilitates advances in real-time patient care, cell structure prediction, and cell morphology research.

No description available.