No description available.

It is an important step in clinical practice to discriminate real diseased patients from healthy persons. It would be great to get such discrimination from some common information like personal information, life style, and the contact with diseased patient. In this study, a score is calculated for each patient based on a survey through generalized linear model, and then the diseased status is decided according to previous sexually transmitted diseases (STDs) records. This study will facilitate clinics in grouping patients into real diseased or healthy, which in turn will affect the method clinics take to screen patients: complete screening for possible diseased patient and some common screening for potentially healthy persons.

The Top Scoring Pair (TSP) classifier, based on the notion of relative ranking reversals in the expressions of two marker genes, has been proposed as a simple, accurate, and easily interpretable decision rule for classification and class prediction of gene expression profiles. We introduce the AUC-based TSP classifier, which is based on the Area Under the ROC (Receiver Operating Characteristic) Curve. The AUCTSP classifier works according to the same principle as TSP but differs from the latter in that the probabilities that determine the top scoring pair are computed based on the relative rankings of the two marker genes across all subjects as opposed to for each individual subject. Although the classification is still done on an individual subject basis, the generalization that the AUC-based probabilities provide during training yield an overall better and more stable classifier. Through extensive simulation results and case studies involving classification in ovarian, leukemia, colon, and breast and prostate cancers and diffuse large b-cell lymphoma, we show the superiority of the proposed approach in terms of improving classification accuracy, avoiding overfitting and being less prone to selecting non-informative pivot genes. The proposed AUCTSP is a simple yet reliable and robust rank-based classifier for gene expression classification. While the AUCTSP works by the same principle as TSP, its ability to determine the top scoring gene pair based on the relative rankings of two marker genes across {\em all} subjects as opposed to each individual subject results in significant performance gains in classification accuracy. In addition, the proposed method tends to avoid selection of non-informative (pivot) genes as members of the top-scoring pair.\\ We have also proposed the use of the AUC test statistic in order to reduce the computational cost of the TSP in selecting the most informative pair of genes for diagnosing a specific disease. We have proven the efficacy of our proposed method through case studies in ovarian, colon, leukemia, breast and prostate cancers and diffuse large b-cell lymphoma in selecting informative genes. We have compared the selected pairs, computational cost and running time and classification performance of a subset of differentially expressed genes selected based on the AUC probability with the original TSP in the aforementioned datasets. The reduce sized TSP has proven to dramatically reduce the computational cost and time complexity of selecting the top scoring pair of genes in comparison to the original TSP in all of the case studies without degrading the performance of the classifier. Using the AUC probability, we were able to reduce the computational cost and CPU running time of the TSP by 79\% and 84\% respectively on average in the tested case studies. In addition, the use of the AUC probability prior to applying the TSP tends to avoid the selection of genes that are not expressed (``pivot'' genes) due to the imposed condition. We have demonstrated through LOOCV and 5-fold cross validation that the reduce sized TSP and TSP have shown to perform approximately the same in terms of classification accuracy for smaller threshold values. In conclusion, we suggest the use of the AUC test statistic in reducing the size of the dataset for the extensions of the TSP method, e.g. the k-TSP and TST, in order to make these methods feasible and cost effective.

This research project aimed to develop a condition monitoring system for the final production quality of pharmaceutical tablets and detection capping and lamination during powder compression process using the acoustic emission (AE) method. Pharmaceutical tablet manufacturers obliged by regulatory bodies to test the tablet's physical properties such as hardness, dissolution and disintegration before the tablets are released to the market. Most of the existing methods and techniques for testing and monitoring these tablet's properties are performed at the tablet post-compression stage. Furthermore, these tests are destructive in nature. Early experimental investigations revealed that the AE energy that is generated during powder compression is directly proportional to the peak force that is required to crush the tablet, i. e. crushing strength. Further laboratory and industrial experimental investigation have been conducted to study the relationship between the AE signals and the compression conditions. Traditional AE signal features such as energy, count, peak amplitude, average signal level, event duration and rise time were recorded. AE data analysis with the aid of advanced classification algorithm, fuzzy C-mean clustering showed that the AE energy is a very useful parameter in tablet condition monitoring. It was found that the AE energy that is generated during powder compression is sensitive to the process and is directly proportional to the compression speed, particle size, homogeneity of mixture and the amount of material present. Also this AE signal is dependent upon the type of material used as the tablet filler. Acoustic emission has been shown to be a useful technique for characterising some of the complex physical changes which occur during tabletting. Capping and lamination are serious problems that are encountered during tabletting. A capped or laminated tablet is one which no longer retains its mechanical integrity and exhibit low strength characteristics. Capping and lamination can be caused by a number of factors such as excessive pressure, insufficient binder in the granules and poor material flowabilities. However, capping and lamination can also occur randomly and they are also dependent upon the material used in tabletting. It was possible to identify a capped or laminated tablet by monitoring the AE energy level during continuous on-line monitoring of tabletting. Capped tablets indicated by low level of AE energy. The proposed condition monitoring system aimed to set the AE energy threshold that could discriminate between capped and non-capped tablets. This was based upon statistical distributions of the AE energy values for both the capped and non-capped tablets. The system aims to minimise the rate of false alarms (indication of capping when in reality capping has not occurred) and the rate of missed detection (an indication of non capping, when in reality capping has occurred). A novel approach that employs both the AE method and the receiver operating characteristic (ROC) curve was proposed for the on-line detection of capping and lamination during tabletting. The proposed system employs AE energy as the discriminating parameter to detect between capped and non-capped tablets. The ROC curve was constructed from the area under the two distributions of both capped and non-capped tablet. This curve shows a trade-off between the probabilities of true detection rate and false alarm rate for capped and non-capped tablet. A two-graph receiver operating characteristic (ROC) curve was presented as a modification of the original ROC curve to enable an operator to directly select the desired energy threshold for tablet monitoring. This plot shows the ROC co-ordinate as a function of the threshold value over the entire threshold (AE energy) range for all test outcomes. An alternative way of deciding a threshold based on the slope of the ROC curve was also developed. The slope of the ROC curve represents the optimal operating point on the curve. It depends upon the penalties cost of capping and the prevalence of capping. Sets of guidelines have been outlined for decision making i.e. threshold setting. These guidelines take into account both the prevalence of capping in manufacturing and the cost associated with various outcomes of tablet formation. The proposed condition monitoring system also relates AE monitoring to non-AE measurement as it enable an operator predicting tablet hardness and disintegration form the AE energy, a relationship which was established in this research.

This is a foundational study that applies Receiver Operating Characteristic (ROC) analysis to the evaluation of a chronic disease model that utilizes Remote Monitoring (RM) devices to identify clinical deterioration in a Chronic Obstructive Pulmonary Disease (COPD) population. Background: RM programmes in Disease Management (DM) are proliferating as one strategy to address management of chronic disease. The need to validate and quantify evidence-based value is acute. There is a need to apply new methods to better evaluate automated RM systems. ROC analysis is an engineering approach that has been widely applied to medical programmes but has not been applied to RM systems. Evaluation of classifiers, determination of thresholds and predictive accuracy for RM systems have not been evaluated using ROC analysis. Objectives: (1) apply ROC analysis to evaluation of a RM system; (2) analyse the performance of the model when applied to patient outcomes for a COPD population; (3) identify predictive classifier(s); (4) identify optimal threshold(s) and the predictive capacity of the classifiers. Methods: Parametric and non-parametric methods are utilized to determine accuracy, sensitivity, specificity and predictive capacity of classifiers Saturated Peripheral Oxygen (SpO2), Blood Pressure (BP), Pulse Rate (PR) based on event-based patient outcomes that include hospitalisation (IP), accident & emergency (A&E) and home visits (HH). Population: Patients identified with a primary diagnosis of COPD, monitored for a minimum of 183 days with at least one episode of in-patient (IP) hospitalisation for COPD in the 12 months preceding the monitoring period. Data Source: A subset of retrospective de-identified patient data from an NHS Direct evaluation of a COPD RM programme. Subsets utilized include classifiers, biometric readings, alerts generated by the system and resource utilisation. Contribution: Validates ROC methodology, identifies classifier performance and optimal threshold settings for the classifier, while making design recommendations and putting forth the next steps for research. The question answered by this research is that ROC analysis can provide additional information on the predictive capacity of RM systems. Justification of benefit: The results can be applied when evaluating health services and planning decisions on the costs and benefits. Methods can be applied to system design, protocol development, work flows and commissioning decisions based on value and benefit. Conclusion: Results validate the use of ROC analysis as a robust methodology for DM programmes that use RM devices to evaluate classifiers, thresholds and identification of the predictive capacity as well as identify areas where additional design may improve the predictive capacity of the model.

Debris-slides are fast-moving landslides that occur in the Appalachian region including the Great Smoky Mountains National Park (GRSM). Various knowledge and data-driven approaches using spatial distribution of the past slides and associated factors could be used to estimate the region’s debris-slide susceptibility. This study developed two debris-slide susceptibility models for GRSM using knowledge-driven and data-driven methods in GIS. Six debris-slide causing factors (slope curvature, elevation, soil texture, land cover, annual rainfall, and bedrock discontinuity), and 256 known debris-slide locations were used in the analysis. Knowledge-driven weighted overlay and data-driven bivariate frequency ratio analyses were performed. Both models are helpful; however, each come with a set of advantages and disadvantages regarding degree of complexity, time-dependency, and experience of the analyst. The susceptibility maps are useful to the planners, developers, and engineers for maintaining the park’s infrastructures and delineating zones for further detailed geotechnical investigation.

Physiological signals, which are controlled by the autonomic nervous system (ANS), could be used to detect the affective state of computer users and therefore find applications in medicine and engineering. The Pupil Diameter (PD) seems to provide a strong indication of the affective state, as found by previous research, but it has not been investigated fully yet. In this study, new approaches based on monitoring and processing the PD signal for off-line and on-line affective assessment (“relaxation” vs. “stress”) are proposed. Wavelet denoising and Kalman filtering methods are first used to remove abrupt changes in the raw Pupil Diameter (PD) signal. Then three features (PDmean, PDmax and PDWalsh) are extracted from the preprocessed PD signal for the affective state classification. In order to select more relevant and reliable physiological data for further analysis, two types of data selection methods are applied, which are based on the paired t-test and subject self-evaluation, respectively. In addition, five different kinds of the classifiers are implemented on the selected data, which achieve average accuracies up to 86.43% and 87.20%, respectively. Finally, the receiver operating characteristic (ROC) curve is utilized to investigate the discriminating potential of each individual feature by evaluation of the area under the ROC curve, which reaches values above 0.90. For the on-line affective assessment, a hard threshold is implemented first in order to remove the eye blinks from the PD signal and then a moving average window is utilized to obtain the representative value PDr for every one-second time interval of PD. There are three main steps for the on-line affective assessment algorithm, which are preparation, feature-based decision voting and affective determination. The final results show that the accuracies are 72.30% and 73.55% for the data subsets, which were respectively chosen using two types of data selection methods (paired t-test and subject self-evaluation). In order to further analyze the efficiency of affective recognition through the PD signal, the Galvanic Skin Response (GSR) was also monitored and processed. The highest affective assessment classification rate obtained from GSR processing is only 63.57% (based on the off-line processing algorithm). The overall results confirm that the PD signal should be considered as one of the most powerful physiological signals to involve in future automated real-time affective recognition systems, especially for detecting the “relaxation” vs. “stress” states.

Artificial Neural Networks (ANNs) have been comprehensively researched, both from a computer scientific perspective and with regard to their use for predictive modelling in a wide variety of applications including hydrology and the environment. Yet their adoption for live, real-time systems remains on the whole sporadic and experimental. A plausible hypothesis is that this may be at least in part due to their treatment heretofore as “black boxes” that implicitly contain something that is unknown, or even unknowable. It is understandable that many of those responsible for delivering Early Warning Systems (EWS) might not wish to take the risk of implementing solutions perceived as containing unknown elements, despite the computational advantages that ANNs offer. This thesis therefore builds on existing efforts to open the box and develop tools and techniques that visualise, analyse and use ANN weights and biases especially from the viewpoint of neural pathways from inputs to outputs of feedforward networks. In so doing, it aims to demonstrate novel approaches to self-improving predictive model construction for both regression and classification problems. This includes Neural Pathway Strength Feature Selection (NPSFS), which uses ensembles of ANNs trained on differing subsets of data and analysis of the learnt weights to infer degrees of relevance of the input features and so build simplified models with reduced input feature sets. Case studies are carried out for prediction of flooding at multiple nodes in urban drainage networks located in three urban catchments in the UK, which demonstrate rapid, accurate prediction of flooding both for regression and classification. Predictive skill is shown to reduce beyond the time of concentration of each sewer node, when actual rainfall is used as input to the models. Further case studies model and predict statutory bacteria count exceedances for bathing water quality compliance at 5 beaches in Southwest England. An illustrative case study using a forest fires dataset from the UCI machine learning repository is also included. Results from these model ensembles generally exhibit improved performance, when compared with single ANN models. Also ensembles with reduced input feature sets, using NPSFS, demonstrate as good or improved performance when compared with the full feature set models. Conclusions are drawn about a new set of tools and techniques, including NPSFS and visualisation techniques for inspection of ANN weights, the adoption of which it is hoped may lead to improved confidence in the use of ANN for live real-time EWS applications.

Breast cancer is a major global health problem and the most common form of cancer among women. Major advances in the technologies of imaging provide improved detection and sensitivity with fewer unnecessary biopsies. Commonly used imaging modalities include mammography, ultrasonography, magnetic resonance imaging (MRI), scintimammography, single photon emission computed tomography (SPECT) and positron emission tomography (PET). The current study is focused on breast MRI imaging, especially one of the most promising recent techniques, i.e. the Diffusion Weighted Imaging breast MRI (DWI). DWI is an unenhanced MRI technique, based on volume sequences on various b values (the b value identifies the measurement's sensitivity to diffusion and determines the strength and duration of the diffusion gradients) measuring the mobility of water molecules (Brownian motion) in vivo (in tissues) and provides different and potentially complementary information to Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) technique. As DWI based on the diffusive properties of water molecules, reflects their random motion resulting from thermal agitation. Water diffusion on breast can be quantified by measuring the mean diffusivity, which is the average of Apparent Diffusion Coefficient (ADC). The ADC can be calculated by making measurements at a low b factor, b1, and a higher b factor, b2. DWI allows the mapping of the diffusion process of molecules by the ADC map. ADC maps are calculated by collecting images with at least 2 different values, b1 and b2, of the b factor. The ADC map is a parametric image whose color scale or gray scale represents the ADC values of the voxels and is usually generated by proprietary or in house software. DWI apart from the 3D anatomical information, provides a noninvasive investigation of tissue vascularity, a novel contrast mechanism in MRI and has a high sensitivity in the detection of changes in the local biologic environment due to a pathologic process. Therefore, in addition to contrast enhancement-based characterization (DCE-MRI), measurement of the motion of water molecules in DWI provides an additional feature for lesion characterization that may further increase the specificity of MRI for classifying breast lesions. The diagnostic task that the current study deals with, accounts for the diagnosis of mass-like lesions in Diffusion Weighed Magnetic Resonance Imaging, based on low ADC values compared to high once in case of benign versus normal tissue. The hypothesis is that diffusivity of water molecules is restricted in environments of high cellularity, intracellular and extracellular edema, high viscosity, and fibrosis, such as malignant tumors, because these conditions become barriers to the movement of water molecules. Therefore, most of breast cancers show low ADC values compared with benign and normal tissue. Many studies have revealed the usefulness of ADC values in the differential diagnosis of breast lesions; however, the clinical effect remains limited because of the substantial overlap between benign and malignant lesions, which presents challenges for implementing a useful diagnostic ADC threshold. The majority of studies, similar to the current study, determined optimal cutoff levels of the ADC value between malignant and benign lesions by using ROC analysis, and ranged from 0.90 to 1.76 × 10-3 mm2/s while the sensitivity and specificity ranged from 63% to 100% and 46% to 97%, respectively. In addition, the methods for measuring ADC differ among reported studies, with the most representative method being the mean value of ADC (mean ± standard deviation) over a Region Of Interest representative of the breast lesion. The purpose of this study was to investigate the ability of histogram characteristics of Apparent Diffusion Coefficient (Apparent Diffusion Coefficient-ADC) to differentiate malignant from benign breast lesions in breast DWI. To this end the ADC maps of representative lesion ROIs were subjected to first order statistics analysis by calculating five first order textural features: Mean value, Standard Deviation, Kurtosis, Skewness and Entropy. This approach is intended to offer a more complete assessment of tumor texture and heterogeneity. The dataset analyzed is comprised of 92 histologically verified breast lesions, originating from 69 women with mammographically and/or ultrasonographically detected or palpable findings. Histology revealed 53 malignant lesions originating from 45 women and 39 benign lesions originating from 26 women. All of the breast MR examinations were performed with a 3T MR scanner, for b= 0, 900 s/mm2. Diagnostic performances of these parameters were compared by receiver operating characteristic (ROC) curve analysis. The mean of ADC of benign lesions [(1.470 ± 0.342) × 10-3 mm2/s] was found to be significantly higher than that of malignant tumours, [(0.965 ± 0.268) × 10-3 mm2/s, (p<0.00001)]. The standard deviation of ADC of benign lesions [(0.184 ± 0.999) × 10-3 mm2/s] was not significantly different from that of malignant tumours, [(0.192 ± 0.151) × 10-3 mm2/s, (p=0.6581)]. The skewness of ADC of benign [-0.303 ± 0.584] was significantly different than that of malignant tumours, 0.210 ± 0.725. (p = 0.0008)]. The kurtosis of ADC of benign [3.003 ± 1.065] was not significantly different from that of malignant tumours, [3.337 ± 1.334. (p=0.0987)]. The entropy of ADC of benign [4.794 ± 0.665] was significantly lower than that of malignant tumours, [5.569 ± 0.649, (p<0.00001)] The corresponding area under the empirical receiver operating characteristic curve was: 0.862 ± 0.042 (95% confidence interval: 0.754, 0.925) for mean of ADC, 0.705 ± 0.054 (95% confidence interval: 0.589, 0.800) for skeweness of ADC, 0.800 ± 0.046 (95% confidence interval: 0.691, 0.874) for entropy of ADC, resulting a good diagnostic performance of DWI for these parameters. On the other hand, an AUC of 0.527 ± 0.063 (95% confidence interval: 0.393, 0.640) and 0.601 ± 0.061 (95% confidence interval: 0.470, 0.707) for Standard deviation and kurtosis respectively, suggests a degree of overlap in ADC values between benign and malignant tumors. In an effort to identify optimal threshold values for differentiating benign versus malignant lesions these were selected to correspond to the points of highest accuracy of the ROC curves. In our study, we obtained two threshold values of mean ADC, both with an accuracy of 83.15%: 1.21 x 10-3 mm2/s with a sensitivity of 86.27% and specificity of 78.95%; and 1.32 x 10-3 mm2/s with a sensitivity of 92.16% and specificity of 71.05%. The threshold value of skeweness was -0.06 with an accuracy of 68.54%, a sensitivity of 66.03% and specificity of 66.67%. Finally, we found two threshold values of entropy, both with an accuracy of 76.40%: 5.17 with a sensitivity of 75.47% and specificity of 71.80%; and 5.21 with a sensitivity of 73.59% and specificity of 74.36%. In conclusion, results of the current study suggest the contribution of texture analysis methods in Diffusion-weighted MRI breast imaging for the quantification of tissue heterogeneity, providing important information for breast cancer diagnosis. Histogram analysis of ADC values in breast cancer has potential for differentiating benign and malignant tumors, providing information about the entire tumor. The mean, skewness and entropy of ADC are valuable parameters that are correlated with pathologic characterization of breast tumors. These 3 ADC parameters significantly elevated the quantitative diagnostic performance of breast DWI and would be effective parameters in distinguishing between malignant and benign breast lesions. Finally, future efforts will also focus on investigating the correlation of extracted texture features with histopathological findings, in order to verify the potential of the proposed texture analysis of ADC map in providing non-invasive prognostic factors of breast cancer. / Ο καρκίνος του μαστού είναι ένα σημαντικό παγκόσμιο πρόβλημα υγείας και η πιο διαδεδομένη μορφή καρκίνου στον γυναικείο πληθυσμό. Η ολοένα και πιο έγκαιρη διάγνωση του καρκίνου του μαστού έχει οδηγήσει σε σημαντική βελτίωση του ρυθμό θεραπείας της νόσου. Σημαντικές πρόοδοι στην τεχνολογία της απεικόνισης παρέχουν τη βελτιωμένη ανίχνευση και ευαισθησία του καρκίνου και οδηγούν σε όλο ένα και λιγότερες περιττές βιοψίες. Οι πιο συνηθισμένες μέθοδοι απεικόνισης, που χρησιμοποιούνται, περιλαμβάνουν την Μαστογραφία, την Υπερηχογραφία, την Μαγνητική Τομογραφία (MRI), την σπινθηρομαστογραφία, την Τομογραφία Εκπομπής Φωτονίων (SPECT) και την Τομογραφία Εκπομπής Ποζιτρονίων (PET). Η παρούσα μελέτη επικεντρώνεται στην τεχνολογία της Απεικόνισης Μαγνητικού Συντονισμού ειδικά σε μία πρόσφατη και ελπιδοφόρα τεχνική απεικόνισης του καρκίνου του μαστού, που ονομάζεται Απεικόνιση Σταθμισμένης Διάχυσης στη Τομογραφία Πυρηνικού Μαγνητικού Συντονισμού (Diffusion Weighted Imaging breast MRI (DWI)). Η DWI είναι μια MRI ακολουθία χωρίς χρήση σκιαγραφικής ουσίας, η οποία βασίζεται σε αλληλουχίες για διάφορες τιμές του παράγοντα διάχυσης b (η τιμή b προσδιορίζει την διαχυτότητα και καθορίζει την ένταση και τη διάρκεια των βαθμωτών πεδίων διάχυσης). Η DWI ποσοστικοποιεί την κινητικότητα των μορίων του νερού (Brownian κίνηση) in vivo (σε ιστούς) και παρέχει διαφορετικές και ενδεχομένως συμπληρωματικές πληροφορίες στην μαστογραφία μαγνητικής τομογραφίας με χρήση σκιαγραφικού (Dynamic Contrast-Enhanced Magnetic Resonance Imaging: DCE-MRI). Η DWI με βάση τις ιδιότητες διάχυσης των μορίων του νερού, αντανακλά την τυχαία κίνησής τους λόγω της θερμικής τους ενέργειας. Η διάχυση του νερού στον μαστό μπορεί να ποσοτικοποιηθεί με τη μέτρηση της μέσης διαχυτότητας, η οποία αναφέρεται ως Φαινόμενος Συντελεστής Διάχυσης (Apparent Diffusion Coefficient-ΑDC). Ο ADC υπολογίζεται ύστερα από μετρήσεις για δυο b τιμές, μια χαμηλή b1 και μια υψηλότερη b2 τιμή. Η DWI επιτρέπει την χαρτογράφηση της διάχυσης των μορίων του νερού μέσω του ADC χάρτη. Ο ADC χάρτης είναι μια παραμετρική εικόνα της οποίας η κλίμακα χρωμάτων ή κλίμακα των τόνων του γκρι, αντιπροσωπεύει τις ADC τιμές των voxels και συνήθως παράγεται από λογισμικό. Οι παραμετρικοί ADC χάρτες απεικόνισης DWI αναπαριστούν τη μικροδομή των ιστών για διάφορους συνδυασμούς τιμών της παραμέτρου b. Η DWI εκτός από 3D ανατομική πληροφορία, παρέχει μια μη επεμβατική διερεύνηση της αγγειοβρίθειας του ιστού, έναν νέο μηχανισμό αντίθεσης στην MRI, και χαρακτηρίζεται από υψηλή ευαισθησία στην ανίχνευση ενδεχόμενων αλλαγών στο τοπικό βιολογικό περιβάλλον, οι οποίες οφείλονται σε παθολογία. Ως εκ τούτου, εκτός από τον χαρακτηρισμό αλλοιώσεων βάση σκιαγραφικής ενίσχυσης (DCE - MRI), η ποσοτικοποίηση της κίνησης των μορίων του νερού στην DWI παρέχει επιπλέον στοιχεία για τον χαρακτηρισμό της αλλοίωσης, κάτι το οποίο μπορεί να αυξήσει περαιτέρω την ειδικότητα της MRI για την ταξινόμηση των αλλοιώσεων του μαστού. Το διαγνωστικό πρόβλημα το οποίο αντιμετώπισε/εστίασε η παρούσα διπλωματική εργασία, αφορά στο χαρακτηρισμό/διάγνωση χωροκατακτητικών αλλοιώσεων (mass-like) του μαστού στην Απεικονιση Μαγνητικου Συντονισμου Σταθμισμενης Διαχυσης (DWI) και την ποσοτικη αναλυση του Φαινομενου Συντελεστη Διαχυσης (ADC) για διαγνωση καρκινου του μαστου. Η ικανότητα διάχυσης των μορίων του νερού περιορίζεται σε περιβάλλον υψηλής κυτταροβρίθιας, ενδοκυττάριων και εξωκυττάριων οιδημάτων, υψηλού ιξώδους και ίνωσης, όπως συμβαίνει στους κακοήθεις όγκους, διότι οι παράγοντες αυτοί εμποδίζουν την κυκλοφορία των μορίων του νερού. Αποτέλεσμα αυτού είναι οι περισσότεροι καρκίνοι του μαστού να παρουσιάζουν χαμηλές ADC τιμές σε σύγκριση με τους καλοήθεις όγκους ή τον φυσιολογικό ιστό. Πολλές μελέτες έχουν δείξει τη χρησιμότητα των ADC τιμών στη διαφορική διάγνωση των αλλοιώσεων του μαστού. Εν τούτοις, το κλινικό αποτέλεσμα παραμένει περιορισμένο λόγω της σημαντικής επικάλυψης καλοήθων και κακοήθων αλλοιώσεων, γεγονός που αποτελεί πρόκληση για την εφαρμογή ενός χρήσιμου διαγνωστικού ορίου της μέσης ADC. Στη πλειοψηφία των μελετών, όπως και στη παρούσα μελέτη, τα βέλτιστα επίπεδα αποκοπής της ADC μεταξύ κακοήθων και καλοήθων αλλοιώσεων προσδιορίστηκαν με τη χρήση ROC ανάλυσης. Στις μέχρι τώρα μελέτες τα διαγνωστικά όρια της μέσης ADC κυμαίνονται από 0.90 έως 1.76 × 10-3 mm2 / s, με ευαισθησία και ειδικότητα να κυμαίνονται από 63% έως 100% και 46% έως 97%, αντίστοιχα. Γεγονός αποτελεί, επίσης, η διαφορετική μέθοδος υπολογισμού της ADC που ακολουθεί η κάθε μελέτη, με πιο συχνή μέθοδο, ο υπολογισμός της μέσης τιμής της ADC (μέση τιμή ± τυπική απόκλιση) σε μια περιοχή ενδιαφέροντος (ROI) μιας αλλοίωσης του μαστού. Σκοπός της παρούσας μεταπτυχιακής διπλωματικής εργασίας ήταν να διερευνηθεί η ικανότητα των χαρακτηριστικών ιστογράμματος του Φαινόμενου Συντελεστή Διάχυσης (Apparent Diffusion Coefficient-ADC) να διαφοροποιούν κακοήθεις από καλοήθεις αλλοιώσεις του μαστού στην Απεικόνιση Μαγνητικού Συντονισμού Σταθμισμένης Διάχυσης (Diffusion Weighted MRI-DWI). Για το σκοπό αυτό, δημιουργήθηκε ο ADC παραμετρικός χάρτης ο οποίος αποτέλεσε τη βάση για την εφαρμογή μεθόδου ανάλυσης υφής εικόνας, και τον υπολογισμό πέντε χαρακτηριστικών υφής πρώτης τάξης: την μέση τιμή, την τυπική απόκλιση, την κύρτωση, την λοξότητα και την εντροπία. Η προσέγγιση αυτή θεωρήθηκε ότι θα προσφέρει μια πιο ολοκληρωμένη αξιολόγηση της υφής του όγκου και της ετερογένειας. Η προσέγγιση εφαρμόσθηκε σε κλινικό δείγμα 92 ιστολογικά αποδεδειγμένων αλλοιώσεων του μαστού, οι οποίες προέρχονται από 69 γυναίκες οι οποίες είχαν νωρίτερα ανιχνευθεί μέσω μαστογραφίας ή/και υπερηχογραφίας ή από ψηλαφητά ευρήματα. Η ιστολογική εξέταση αποκάλυψε 53 κακοήθεις αλλοιώσεις που προέρχονταν από 45 γυναίκες και 39 καλοήθεις αλλοιώσεις από 26 γυναίκες. Όλες οι εξετάσεις μαγνητικής τομογραφίας του μαστού έγιναν με σύστημα MRI 3T και για b=0 και 900 s/mm2. Η διαγνωστική απόδοση/επίδοση των παραμέτρων αυτών συγκρίθηκε με την ανάλυση λειτουργικού χαρακτηριστικού δέκτη (ROC analysis). Τα αποτελέσματα υποδεικνύουν τον σημαντικό ρόλο της ανάλυσης ADC ιστογράμματος χρησιμοποιώντας τα 5 παραπάνω χαρακτηριστικά υφής για την ταυτοποίηση των αλλοιώσεων του μαστού. Οι μετρήσεις της μέσης τιμής, της λοξότητας και της εντροπία της ADC των καλοήθων και κακοήθων αλλοιώσεων του μαστού είχαν στατιστικώς σημαντική διαφορά. Ειδικότερα, η μέση ADC τιμή των καλοήθων όγκων [(1.470 ± 0.342) × 10-3 mm2/s] ήταν σημαντικά υψηλότερη από εκείνη των κακοήθων, [(0.965 ± 0.268) × 10-3 mm2/s, (ρ < 0.00001)]. Η λοξότητα της ADC των καλοήθων όγκων [-0.303 ± 0.584], διέφερε σημαντικά από εκείνη των κακοήθων, [0.210 ± 0.725. (ρ= 0.0008)]. Και η εντροπία της ADC των καλοήθων όγκων [4.794 ± 0.665], ήταν σημαντικά χαμηλότερη από εκείνη των κακοήθων, [5.569 ± 0.649, (ρ < 0.00001)]. Ωστόσο, η τυπική απόκλιση και η κύρτωση της ADC των καλοήθων και κακοήθων αλλοιώσεων του μαστού δεν είχαν στατιστικώς σημαντική διαφορά. Συγκεκριμένα, η τυπική απόκλιση της ADC των καλοήθων όγκων ήταν [(0.184 ± 0.999) × 10-3 mm2/s] ενώ των κακοήθων ήταν [(0.192 ± 0.151) × 10-3 mm2/s, (ρ = 0.6581)] και η κύρτωση της ADC των καλοήθων όγκων ήταν [3.003 ± 1.065] ενώ των κακοήθων ήταν [3.337 ± 1.334, (ρ = 0.0987)]. Η περιοχή κάτω από την ROC καμπύλη (AUC) για τη μέση ADC τιμή ήταν 0.862 ± 0.042 (95% διάστημα εμπιστοσύνης: 0.754, 0.925), για την λοξότητα της ADC ήταν 0.705 ± 0.054 (95% διάστημα εμπιστοσύνης: 0.589, 0.800) και για η εντροπία της ADC ήταν 0.800 ± 0.046 (95% διάστημα εμπιστοσύνης: 0.691, 0.874), και είχαν ως αποτέλεσμα μια καλή διαγνωστική απόδοση/ επίδοση της DWI για τις παραμέτρους αυτές. Από την άλλη πλευρά, η AUC με 0.527 ± 0.063 (95% διάστημα εμπιστοσύνης: 0.393, 0.640) και με 0.601 ± 0.061 (95% διάστημα εμπιστοσύνης: 0.470, 0.707) για την τυπική απόκλιση και την κύρτωση, αντίστοιχα, υποδηλώνει ένα βαθμό επικάλυψης στις ADC τιμές μεταξύ καλοήθων και κακοήθων όγκων. Τα βέλτιστα κατώφλια αποκοπής για διαφοροποίηση καλοήθων έναντι κακοήθων αλλοιώσεων καθορίστηκαν με τον εντοπισμό των σημείων όπου η ακρίβεια ήταν μέγιστη στις καμπύλες ROC. Από τη συγκεκριμένη μελέτη, προέκυψαν δύο τιμές κατωφλίου αποκοπής της μέσης ADC, με την ίδια ακρίβεια 83.15%. Το πρώτο κατώφλι με τιμή 1.21 x 10-3 mm2/s χαρακτηρίζεται με ευαισθησία 86.27% και ειδικότητα 78.95%. Και το δεύτερο κατώφλι με τιμή 1.32 x 10-3 mm2/s χαρακτηρίζεται με ευαισθησία 92.16% και ειδικότητα 71.05%. Το κατώφλι για την λοξότητα της ADC ήταν στα -0.06 με ακρίβεια 68.54%, ευαισθησία 66.03% και ειδικότητα 66.67%. Τέλος, προέκυψαν δύο τιμές κατωφλίου αποκοπής της εντροπίας της ADC με την ίδια ακρίβεια ακρίβεια 76.40%. Το πρώτο κατώφλι με τιμή 5.17 χαρακτηρίζεται με ευαισθησία 75.47% και ειδικότητα 71.80%. Και το δεύτερο κατώφλι με τιμή 5.21 χαρακτηρίζεται με ευαισθησία 73.59% και ειδικότητα 74.36%. Συμπερασματικά, τα αποτελέσματα της παρούσας μελέτης δείχνουν τη συνεισφορά των μεθόδων ανάλυσης υφής εικόνας στη Απεικονιση Μαγνητικου Συντονισμου Σταθμισμενης Διαχυσης (DWI) του μαστού για την ποσοτικοποίηση της ετερογένειας του ιστού, παρέχοντας σημαντικές πληροφορίες για τη διάγνωση του καρκίνου του μαστού. Η ανάλυση ιστογράμματος των ADC τιμών στον καρκίνο του μαστού έχει τη δυνατότητα διαφοροποίησης καλοήθων και κακοήθων όγκων, παρέχοντας πληροφορίες για το σύνολο του όγκου. Η μέση τιμή, η λοξότητα και η εντροπία του ADC είναι πολύτιμες παράμετροι που συσχετίζονται με παθολογικό χαρακτηρισμό των όγκων του μαστού. Αυτές οι 3 ADC παράμετροι αύξησαν σημαντικά την ποσοτική διαγνωστική απόδοση της DWI του μαστού και πιθανότατα να είναι αποτελεσματικές παράμετροι όσον αφορά τη διάκριση μεταξύ καλοήθων και κακοήθων αλλοιώσεων του μαστού Μελλοντικές προσπάθειες πρόκειται να εστιάσουν στη διερεύνηση της συσχέτισης των εξαχθέντων χαρακτηριστικών υφής με ιστοπαθολογικούς δείκτες, με σκοπό την περαιτέρω επιβεβαίωση των προτεινόμενων προσεγγίσεων και ενδεχομένως την χρήση συγκεκριμένων χαρακτηριστικών υφής ως μη επεμβατικών προγνωστικών δεικτών καρκίνου του μαστού.

Le sepsis tardif, défini comme une infection systémique chez les nouveaux nés âgés de plus de 3 jours, survient chez environ 7% à 10% de tous les nouveau-nés et chez plus de 25% des nouveau-nés de très faible poids de naissance qui sont hospitalisés dans les unités de soins intensifs néonatals (USIN). Les apnées et bradycardies (AB) spontanées récurrentes et graves sont parmi les principaux indicateurs précoces cliniques de l'infection systémique chez les prématurés. L'objectif de cette thèse est de déterminer si la variabilité du rythme cardiaque (VRC), la respiration et l'analyse de leurs relations aident au diagnostic de l'infection chez les nouveaux nés prématurés par des moyens non invasifs en USIN. Par conséquent, on a effectué l'analyse Mono-Voie (MV) et Bi-Voies (BV) sur deux groupes sélectionnés de nouveau-nés prématurés: sepsis (S) vs. non-sepsis (NS). (1) Tout d'abord, on a étudié la série RR non seulement par des méthodes de distribution (moy, varn, skew, kurt, med, SpAs), par les méthodes linéaire: le domaine temporel (SD, RMSSD) et dans le domaine fréquentiel (p_VLF, p_LF, p_HF), mais aussi par les méthodes non–linéaires: la théorie du chaos (alphas, alphaF) et la théorie de l'information (AppEn, SamEn, PermEn, Regul). Pour chaque méthode, nous étudions trois tailles de fenêtre 1024/2048/4096, puis nous comparons ces méthodes afin de trouver les meilleures façons de distinguer S de NS. Les résultats montrent que les indices alphaS, alphaF et SamEn sont les paramètres optimaux pour séparer les deux populations. (2) Ensuite, la question du couplage fonctionnel entre la VRC et la respiration nasale est adressée. Des relations linéaires et non-linéaires ont été explorées. Les indices linéaires sont la corrélation (r²), l'indice de la fonction de cohérence (Cohere) et la corrélation temps-fréquence (r2t,f) , tandis que le coefficient de régression non-linéaire (h²) a été utilisé pour analyser des relations non-linéaires. Nous avons calculé les deux directions de couplage pendant l'évaluation de l'indice h2 de régression non-linéaire. Enfin, à partir de l'ensemble du processus d'analyse, il est évident que les trois indices (r2tf_rn_raw_0p2_0p4, h2_rn_raw et h2_nr_raw) sont des moyens complémentaires pour le diagnostic du sepsis de façon non-invasive chez ces patients fragiles. (3) Après, l'étude de faisabilité de la détection du sepsis en USIN est réalisée sur la base des paramètres retenus lors des études MV et BV. Nous avons montré que le test proposé, basé sur la fusion optimale des six indices ci-dessus, conduit à de bonnes performances statistiques. En conclusion, les mesures choisies lors de l'analyse des signaux en MV et BV ont une bonne répétabilité et permettent de mettre en place un test en vue du diagnostic non invasif et précoce du sepsis. Le test proposé peut être utilisé pour fournir une alarme fiable lors de la survenue d'un épisode d'AB tout en exploitant les systèmes de monitoring actuels en USIN. / Late-onset sepsis, defined as a systemic infection in neonates older than 3 days, occurs in approximately 10% of all neonates and in more than 25% of very low birth weight infants who are hospitalized in Neonatal Intensive Care Units (NICU). Recurrent and severe spontaneous apneas and bradycardias (AB) is one of the major clinical early indicators of systemic infection in the premature infant. Various hematological and biochemical markers have been evaluated for this indication but they are invasive procedures that cannot be repeated several times. The objective of this Ph.D dissertation was to determine if heart rate variability (HRV), respiration and the analysis of their relationships help to the diagnosis of infection in premature infants via non-invasive ways in NICU. Therefore, we carried out Mono-Channel (MC) and Bi-Channel (BC) Analysis in two selected groups of premature infants: sepsis (S) vs. non-sepsis (NS). (1) Firstly, we studied the RR series not only by distribution methods (moy, varn, skew, kurt, med, SpAs), by linear methods: time domain (SD, RMSSD) and frequency domain (p_VLF, p_LF, p_HF), but also by non-linear methods: chaos theory (alphaS, alphaF) and information theory (AppEn, SamEn, PermEn, Regul). For each method, we attempt three sizes of window 1024/2048/4096, and then compare these methods in order to find the optimal ways to distinguish S from NS. The results show that alphaS, alphaF and SamEn are optimal parameters to recognize sepsis from the diagnosis of late neonatal infection in premature infants with unusual and recurrent AB. (2) The question about the functional coupling of HRV and nasal respiration is addressed. Linear and non-linear relationships have been explored. Linear indexes were correlation (r²), coherence function (Cohere) and time-frequency index (r2t,f), while a non-linear regression coefficient (h²) was used to analyze non-linear relationships. We calculated two directions during evaluate the index h2 of non-linear regression. Finally, from the entire analysis process, it is obvious that the three indexes (r2tf_rn_raw_0p2_0p4, h2_rn_raw and h2_nr_raw) were complementary ways to diagnosticate sepsis in a non-invasive way, in such delicate patients.(3) Furthermore, feasibility study is carried out on the candidate parameters selected from MC and BC respectively. We discovered that the proposed test based on optimal fusion of 6 features shows good performance with the largest Area Under Curves (AUC) and the least Probability of False Alarm (PFA). As a conclusion, we believe that the selected measures from MC and BC signal analysis have a good repeatability and accuracy to test for the diagnosis of sepsis via non-invasive NICU monitoring system, which can reliably confirm or refute the diagnosis of infection at an early stage.