Logo: to the web site of Uppsala University

Computed Tomography Pulmonary Angiography (CTPA) is an essential imaging modality for diagnosing pulmonary embolism (PE). Although its primary focus is on the pulmonary arteries, the detailed images it provides can also offer valuable insights into cardiovascular structures. This thesis focuses on developing automated systems to detect, segment, and measure mediastinal structures and PE in CTPA scans, addressing challenges such as increasing radiology workloads and the need for high diagnostic accuracy.

Paper I presents a deterministic algorithm that automates segmentation and measurement of major thoracic vessels, including the ascending aorta (AAo), descending aorta (DAo), and pulmonary trunk (PT). This method eliminates the need for manual annotations, achieving segmentation success rates of up to 100% and demonstrating high correlations with radiologist measurements (Pearson’s r = 0.68–0.99) across varying image qualities. External validation further confirmed its robust performance, with a Dice score of 0.92, highlighting its clinical applicability.

Paper II builds on this by training a deep learning model (nnU-Net) using segmentation masks derived from the deterministic approach. This model achieved improved segmentation accuracy (Dice score: 0.95) and higher success rates for detecting AAo, DAo, and PT compared to traditional methods. Validation on external datasets further confirmed its reliability and potential for integration into clinical workflows.

Paper III focuses on detecting pulmonary embolism (PE) using a 3D U-Net generated by the nnU-Net framework, which was trained on an internal dataset of 149 CTPA scans containing PE. The model achieved exceptional classification performance, with sensitivity and specificity exceeding 96% in internal and external validations. Advanced post-processing strategies improved specificity and reduced false positives, outperforming state-of-the-art methods in sensitivity, specificity, and overall diagnostic accuracy.

Future work includes integrating these models into Picture Archiving and Communication Systems (PACS) for seamless clinical deployment, refining algorithms for PT measurement and right/left ventricle analysis, and exploring advanced architectures like vision transformers to further enhance performance. In conclusion, the proposed developments aim to elevate diagnostic precision and clinical outcomes, paving the way for routine deployment of automated CTPA analysis systems in healthcare.

PURPOSE: To develop a deep learning-based algorithm that automatically and accurately classifies patients as either having pulmonary emboli or not in CT pulmonary angiography (CTPA) examinations.

MATERIALS AND METHODS: For model development, 700 CTPA examinations from 652 patients performed at a single institution were used, of which 149 examinations contained 1497 PE traced by radiologists. The nnU-Net deep learning-based segmentation framework was trained using 5-fold cross-validation. To enhance classification, we applied logical rules based on PE volume and probability thresholds. External model evaluation was performed in 770 and 34 CTPAs from two independent datasets.

RESULTS: A total of 1483 CTPA examinations were evaluated. In internal cross-validation and test set, the trained model correctly classified 123 of 128 examinations as positive for PE (sensitivity 96.1 %; 95 % C.I. 91-98 %; P < .05) and 521 of 551 as negative (specificity 94.6 %; 95 % C.I. 92-96 %; P < .05), achieving an area under the receiver operating characteristic (AUROC) of 96.4 % (95 % C.I. 79-99 %; P < .05). In the first external test dataset, the trained model correctly classified 31 of 32 examinations as positive (sensitivity 96.9 %; 95 % C.I. 84-99 %; P < .05) and 2 of 2 as negative (specificity 100 %; 95 % C.I. 34-100 %; P < .05), achieving an AUROC of 98.6 % (95 % C.I. 83-100 %; P < .05). In the second external test dataset, the trained model correctly classified 379 of 385 examinations as positive (sensitivity 98.4 %; 95 % C.I. 97-99 %; P < .05) and 346 of 385 as negative (specificity 89.9 %; 95 % C.I. 86-93 %; P < .05), achieving an AUROC of 98.5 % (95 % C.I. 83-100 %; P < .05).

CONCLUSION: Our automatic pipeline achieved beyond state-of-the-art diagnostic performance of PE in CTPA using nnU-Net for segmentation and volume- and probability-based post-processing for classification.

Mediastinal structure measurements are important for the radiologist's review of computed tomography pulmonary angiography (CTPA) examinations. In the reporting process, radiologists make measurements of diameters, volumes, and organ densities for image quality assessment and risk stratification. However, manual measurement of these features is time consuming. Here, we sought to develop a time-saving automated algorithm that can accurately detect, segment and measure mediastinal structures in routine clinical CTPA examinations. In this study, 700 CTPA examinations collected and annotated. Of these, a training set of 180 examinations were used to develop a fully automated deterministic algorithm. On the test set of 520 examinations, two radiologists validated the detection and segmentation performance quantitatively, and ground truth was annotated to validate the measurement performance. External validation was performed in 47 CTPAs from two independent datasets. The system had 86-100% detection and segmentation accuracy in the different tasks. The automatic measurements correlated well to those of the radiologist (Pearson's r 0.68-0.99). Taken together, the fully automated algorithm accurately detected, segmented, and measured mediastinal structures in routine CTPA examinations having an adequate representation of common artifacts and medical conditions.