Search

Ziel Individuelle Risikoevaluierung von COVID-19-Patienten anhand einer bei stationärer Aufnahme initial angefertigten CT-Untersuchung des Thorax zur Prognose einer notwendigen intensivmedizinischen Versorgung im weiteren klinischen Verlauf. Material und Methoden Das CT-Thorax von 34 symptomatischen SARS-CoV-2-positiven Patienten (58,2 ± 11,8 Jahre) wurde mittels Künstlicher-Intelligenz-Algorithmus (KI-Algorithmus) [CT Pneumonia Analysis, Siemens Healthineers] analysiert. Neben der vollautomatischen Quantifizierung des pneumonischen Lungenbefalls (Opacity) wurden die Vitalparameter SpO2 und Atemfrequenz erfasst und deren Einfluss auf eine potenzielle intensivmedizinische Behandlung analysiert. Die anschließend erfolgte intensivmedizinische bzw. normalstationäre Versorgung sowie eine Lungenverdichtung < 10 % (Opacitylow) bzw. ≥ 10 % (Opacityhigh) wurden als Subkollektive definiert. Ergebnisse Patienten, welche drei Tage nach der CT intensivstationär versorgt werden mussten, wiesen eine höhere initiale Infiltration von im Mdn = 34,57 % (IQR = 59,76 %) im Vergleich zu normalstationär versorgten Patienten mit im Mdn = 5,53 % (IQR = 4,79%) (z = 3,599, p < 0,001, r = 0,617) auf. Für den 7. und 14. Tag post-CT ergaben sich vergleichbare Ergebnisse mit 28,45 % gegenüber 5,62 % initialer Infiltration (z = -3,289 p = 0,001, r = 0,564). Bei Opacityhigh- (7/13) war die Notwendigkeit einer intensivmedizinischen Behandlung häufiger als bei Opacitylow-Patienten (0/21) (p < 0,001). Die Kombination aus klinischen und computertomographischen Parametern mit Schwellenwerten von SpO2 ≤ 95 %, Atemfrequenz ≥ 20 Atemzüge/min und erhöhter Lungeninfiltration (≥ 10 %) ergab ein kombiniertes relatives Risiko für eine intensivmedizinische Versorgung von 9,75 95 % CI [2,43, 39,16] 7 und 14 Tage nach der initialen CT-Untersuchung. Schlussfolgerung Die KI-Algorithmus-basierte Auswertung des CT-Thorax ermöglicht eine automatisierte und untersucherunabhängige Quantifizierung der pulmonalen Infiltration bei COVID-19-Patienten. Mit der CT-Quantifizierung allein und in Kombination mit den klinischen Vitalparametern SpO2 und Atemfrequenz wird eine frühzeitige und individuelle prognostische Risikostratifizierung erreicht, welche für eine frühzeitige Therapieeskalation oder als Grundlage für die Planung intensivmedizinischer Kapazitäten im Rahmen der COVID-19-Pandemie herangezogen werden kann., Objective Individual risk evaluation of COVID-19 patients based on a chest CT scan performed at the time of hospital admission in order to predict the need for intensive care treatment in the further clinical curse. Material and methods The thoracic CT scans of 34 symptomatic SARS-CoV-2 positive patients (58.2 ± 11.8 years) were analyzed using an artificial intelligence algorithm (AI algorithm) [CT Pneumonia Analysis, Siemens Healthineers]. In addition to the fully automated quantification of pneumonic infiltrates (Opacity), the vital parameters SpO2 and respiratory rate were recorded, and their influence on a potential intensive care treatment was analyzed. The subsequent intensive care and normal inpatient treatment as well as a pulmonary involvement < 10 % (Opacitylow) and ≥ 10 % (Opacityhigh) were defined as subcollectives. Results Patients who received intensive care treatment three days after admission showed a higher pulmonary involvement in the initial CT scan with a Mdn = 34.57 % (IQR = 59.76 %), compared to patients who received normal care with a Mdn = 5.53 % (IQR = 4.79 %) (z = -3.599, p < 0.001, r = 0.617). For the 7th and 14th day post-CT, similar results were obtained, with 28.45 % compared to 5.62 % initial involvement (z = -3.289 p = 0.001, r = 0.564). Opacityhigh patients (7/13) required intensive care treatment more often than Opacitylow patients (0/21) (p < 0.001). The combination of vital parameters and CT scan results with thresholds of SpO2 ≤ 95 %, respiratory rate ≥ 20 breaths/min and increased pulmonary involvement (≥ 10 %) resulted in a combined relative risk for intensive care treatment of 9.75 95 % CI [2.43, 39.16] 7 and 14 days after initial CT examination. Conclusion The AI algorithm-based evaluation of chest CT scans allows an automated and observer-independent quantification of pneumonic infiltrates in COVID-19 patients. This method may – by itself as well as in combination with the vital parameters SpO2 and respiratory rate – enable an early and individual prognostic risk stratification, which can be used for an early treatment escalation or as a basis for managing intensive care capacities in the context of the COVID-19 pandemic.

Roboticists have shown a great interest in designing and building robotic hands, mainly because of the complex structure and high range of functionalities that the human hand has. Even though many hand models have been produced, being already used in many different applications, they are still not precise and skilled enough to be considered a proper replica of the human hand, due the size and exibility issues. In order to build an accurate representation of the human hand kinematics, in vivo 3D MRI (Magnetic Resonance Imaging) scans of the hand in different postures are used, here for the investigation of bone poses and finger movements. This thesis deals with an important step in deriving an accurate human hand model using MRI imaging: the precise segmentation of bone structures. In this case, the task of segmenting multiple bone structures is of great diffculty, as it has to take into consideration the characteristic problems encountered with MR imaging techniques. The segmentation algorithm focuses on the boundary between the bone marrow (yielding high gray level intensity values) and the cortical bone (represented by dark regions). Therefore, methods for reducing noise and enhancing edges are considered, and the extraction of the structures of interest is accomplished by using implicit active contours. After initialization is performed using Fuzzy C-means or thresholding techniques, a hybrid level set method, which includes region-based and edge-based information, is used for the segmentation of the bone volumes. The whole algorithm is implemented in MATLAB and comes in the form of a Graphical User Interface (GUI) which makes the interaction with the user much easier. For validation, the segmentation results are compared with manually acquired ground truth, providing an average sensitivity of 93% and a Dice similarity coefficient of 95%. The segmented bone structures are exported in 3D coordinates with respect to a fixed reference system, as it is requested for the further steps in the human hand modelling: pose estimation and registration of different postures.