Email this article 
CT angiography dataset with abdominal aorta segmentation
- Authors: Kodenko M.R.1,2, Vasilev Y.A.1, Solovev A.V.1,3, Gatin D.V.1, Yasakova E.P.1, Guseva A.V.1,2, Reshetnikov R.V.1
-
Affiliations:
- Research and Practical Clinical Center for Diagnostics and Telemedicine Technologies
- Bauman Moscow State Technical University
- Morozov Children's City Clinical Hospital
- Issue: Vol 6, No 1 (2025)
- Pages: 23-32
- Section: Datasets
- Submitted: 02.09.2024
- Accepted: 29.10.2024
- Published: 22.01.2025
- URL: https://jdigitaldiagnostics.com/DD/article/view/635589
- DOI: https://doi.org/10.17816/DD635589
- ID: 635589
Cite item
Abstract
BACKGROUND: Artificial intelligence algorithms are used to analyze images obtained through radiological diagnostic methods. The effectiveness of such algorithms depends on the availability of relevant and representative training datasets. The volume of such data in the public domain should be increased, particularly datasets containing abdominal aorta computed tomography angiography images, with pathology classification and vessel segmentation. The limitations of existing solutions include small sample sizes, restricted dataset specialization, and inconsistent dataset preparation methodologies.
Aim: To create an open dataset containing computed tomography angiography images of abdominal aorta segmentation for normal aorta, aneurysm, thrombosis, and calcification.
MATERIALS AND METHODS: A technical specification for dataset preparation was developed according to the methodology for testing artificial intelligence algorithms, the required sample size was calculated, and approval was obtained from an independent ethics committee. Regarding dataset creation, a previously developed original semiautomatic segmentation algorithm using Slicer 3D software was employed. The inclusion criteria were computed tomography angiography or abdominal computed tomography scans with contrast, arterial phase, and slice thickness ≤3 mm. Conversely, the exclusion criteria were presence of foreign bodies in the aorta lumen and aortic dissection. The algorithm was tested on patient data obtained from the Unified Radiological Information System. An expert evaluation was conducted to assess the compliance of obtained results with the established requirements and evaluate the time efficiency of using the developed segmentation algorithm.
RESULTS: The calculated sample size was 100 angiographic studies, including arterial phase scans with a slice thickness of ≤1.2 mm. Population data: number of unique patients, 100; percentage of female patients, 51%; and median age, 62 years (age range: 18–84 years). Pathology (including combined pathology) was detected in 61% of cases: 60 studies showed signs of calcification, 18 revealed aortic dilation, and 18 determined signs of thrombosed lumen. The average time to process one study (100 slices) using the developed segmentation algorithm was 0.8 hours.
CONCLUSIONS: A dataset containing 100 computed tomography angiography results with abdominal aorta segmentation for normal cases, aneurysm, thrombosis, and calcification was created. The dataset is publicly available and can be used for developing and testing artificial intelligence algorithms and for anthropomorphic modeling of the abdominal aorta.
Full Text
BACKGROUND
The abdominal aorta is a long blood vessel with patient-specific geometry. Its structure and lumen may be altered by various medical conditions [1]. Computed tomography (CT) angiography is considered the gold standard for diagnosing abdominal aortic conditions [2]. CT angiography data can be used for accurate diagnosis and surgical treatment planning [3]. The use of artificial intelligence (AI) to process CT angiography data automates routine medical tasks and optimizes diagnostic and treatment processes [4]. Representative datasets are required for training and testing AI algorithms [5]. Limited well-annotated, publicly accessible imaging data significantly hinders the development of clinically applicable algorithms. This study primarily focused on a CT angiography dataset with abdominal aortic segmentation for cases with and without radiological signs of atherosclerotic lesions, such as lumen dilatation, thrombosis, and wall calcification [6].
Review of publicly accessible, ready-made solutions revealed that a few heterogeneous datasets only partially address the challenges of training and testing AI algorithms. Current solutions do not provide detailed descriptions of methods for obtaining and annotating images acquired using diagnostic imaging1,2 [7–10]. In addition, some scans lack crucial clinical and demographic descriptions of patients whose findings are included in the dataset1 [7]. Notably, the single-center nature of the obtained data2 and narrow specialization of the datasets [8] limit their use for specific clinical tasks.
AIM
The study aimed to create an open dataset containing CT angiography images of abdominal aorta segmentation for normal aorta, aneurysm, thrombosis, and calcification.
METHODS
Study Design
The study was conducted in accordance with the MI-CLAIM checklist [11], which regulates the clinical use of AI technologies, based on the rules for testing AI algorithms in diagnostic imaging [12]. This study was a retrospective analysis of CT angiography data.
Eligibility Criteria
The following inclusion criteria were established:
- Contrast-enhanced abdominal CT angiography or CT
- CT angiography of the abdominal aorta and its branches
- Arterial phase scan
- Slice thickness ≤3 mm.
Exclusion criteria:
- Signs of vascular wall dissection
- Intravascular stents.
Study Setting
CT angiography data were obtained from the Unified Radiology Information Service (URIS) without restrictions for healthcare institutions. URIS is part of the Unified Medical Information and Analytical System of Moscow (EMIAS).3
Data were collected from March 24, 2017, to April 22, 2022. A radiologist with more than 3 years of experience manually selected the scans.
Data Processing Methods
Raw data are presented in Digital Imaging and Communications in Medicine (DICOM) format [13], and the segmentation results are saved in Neuroimaging Informatics Technology Initiative format [14]. The images were segmented using 3D Slicer (v.5.0.2)4 according to the original methodology, which is described in detail in our previous studies [15, 16]. The annotating scenario was as follows:
- Contouring of every fifth slice for areas of interest or every third slice for areas with significant vessel tortuosity;
- Approximation with the Fill Between Slices tool.
If necessary, the resulting mask was manually adjusted using the standard brush and eraser tools.
The images were segmented by a radiologist with 3 years of experience. Subsequently, a medical expert with 10 years of experience reviewed, verified, and adjusted the segmentation results, if required. The output data structure is represented by paired images and masks. Various segmentation masks were used to distinguish between different tissue types, such as the aorta, thrombus, and calcifications. The annotating time for each scan was recorded to determine the average processing time for one CT angiography image.
Ethics Approval
The Independent Ethics Committee of the Research and Practical Clinical Center for Diagnostics and Telemedicine Technologies (Moscow, Russia) permitted the use of patient data (protocol no. 5/2023, dated May 25, 2023). DICOM data downloaded from URIS were depersonalized; that is, personal data and unique scan identifiers were removed.
Statistical Analysis
Principles of sample size calculation: The dataset was intended for testing and training the AI algorithms. The sample size was estimated using the minimum size required to assess the diagnostic accuracy metrics of the AI algorithms [17]. The estimation parameters included the estimated area under receiver operating curve (AUROC), considering the acceptable confidence interval. The AUROC range was set to 10% [18], and the target AUROC was based on the confidence interval, ensuring that the lower limit exceeded the threshold established by the guidelines for testing AI algorithms [12] in the Moscow Experiment on Computer Vision [19]. Presize was used for sample size calculations.5 The estimated sample size required by the chosen methodology was approximately 80 scans, with a normal-to-abnormal ratio of 1:1. However, owing to the potential for variability in abnormal results, the sample size was increased to 100.
Statistical data analysis: Statistical data analysis was limited to descriptive statistics of the included scans and the structure of the dataset. These calculations were performed using MS Excel tools.6
RESULTS
Participants
Population data:
- Total number of patients: 100;
- Female patients: 51%;
- Median age: 62 years, ranging 18–84 years.
Primary Results
The final dataset included 100 annotated CT angiography scans, with a total size of 3.12 GB. Table 1 presents details on the dataset structure. In a sample of 80 scans, the normal-to-abnormal ratio was approximately 1:1, with an additional 20 scans containing abnormal patterns. Fig. 1 and 2 show examples of CT angiography data annotation and vessel lumen segmentation and various abnormalities, such as lumen, calcification, and thrombus abnormalities. Fig. 3 displays the dataset structure. The dataset is registered as a database under the state registration certificate number 2024621990 [20]. It is available for free download under Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported (CC BY-NC ND 3.0) license.7
Table 1. Dataset structure
Abnormal changes | Scan no. | ||||||||||||||||||||||||
1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | |
Abdominal aorta dilation | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
Maximum abdominal aorta diameter, mm | 18 | 16 | 16 | 17 | 18 | 18 | 18 | 18 | 17 | 19 | 21 | 16 | 17 | 23 | 18 | 16 | 17 | 15 | 17 | 16 | 21 | 17 | 17 | 17 | 16 |
Calcification | 0 | 0 | 0 | 1 | 0 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 0 | 1 | 0 | 1 | 0 | 1 | 1 | 1 | 0 | 1 |
Thrombosis | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 1 |
26 | 27 | 28 | 29 | 30 | 31 | 32 | 33 | 34 | 35 | 36 | 37 | 38 | 39 | 40 | 41 | 42 | 43 | 44 | 45 | 46 | 47 | 48 | 49 | 50 | |
Abdominal aorta dilation | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
Maximum abdominal aorta diameter, mm | 16 | 23 | 20 | 20 | 15 | 21 | 15 | 18 | 18 | 29 | 22 | 11 | 15 | 37 | 18 | 19 | 17 | 16 | 21 | 15 | 15 | 24 | 18 | 17 | 18 |
Calcification | 0 | 1 | 1 | 1 | 0 | 0 | 0 | 1 | 0 | 1 | 1 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 1 | 1 | 0 | 1 |
Thrombosis | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
51 | 52 | 53 | 54 | 55 | 56 | 57 | 58 | 59 | 60 | 61 | 62 | 63 | 64 | 65 | 66 | 67 | 68 | 69 | 70 | 71 | 72 | 73 | 74 | 75 | |
Abdominal aorta dilation | 0 | 0 | 0 | 1 | 1 | 1 | 1 | 0 | 1 | 0 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 0 | 1 | 0 | 0 | 1 |
Maximum abdominal aorta diameter, mm | 19 | 19 | 14 | 34 | 40 | 35 | 37 | 20 | 37 | 19 | 36 | 35 | 19 | 33 | 37 | 32 | 43 | 23 | 30 | 27 | 23 | 27 | 19 | 16 | 41 |
Calcification | 0 | 1 | 0 | 1 | 1 | 1 | 1 | 0 | 1 | 0 | 1 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
Thrombosis | 0 | 0 | 0 | 1 | 0 | 1 | 1 | 0 | 1 | 0 | 1 | 0 | 0 | 1 | 0 | 0 | 1 | 1 | 0 | 1 | 1 | 0 | 0 | 0 | 0 |
76 | 77 | 78 | 79 | 80 | 81 | 82 | 83 | 84 | 85 | 86 | 87 | 88 | 89 | 90 | 91 | 92 | 93 | 94 | 95 | 96 | 97 | 98 | 99 | 100 | |
Abdominal aorta dilation | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
Maximum abdominal aorta diameter, mm | 33 | 18 | 20 | 19 | 18 | 14 | 16 | 17 | 22 | 20 | 14 | 14 | 17 | 18 | 21 | 13 | 17 | 18 | 15 | 16 | 12 | 16 | 16 | 14 | 23 |
Calcification | 1 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 1 | 1 | 0 | 0 | 1 | 0 | 0 | 1 |
Thrombosis | 1 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
Note. 0, the feature is present; 1, the feature is absent. | |||||||||||||||||||||||||
Fig. 1. Abdominal aortic segmentation using 3D Slicer.
Fig. 2. Vessel annotation using 3D Slicer: red, vessel lumen; purple, thrombi; yellow, calcifications.
Fig. 3. Dataset fragment illustrating its organization. The classification.xlsx file contains a list of the included scans and classification results. The Studies directory contains scans and binary segmentation masks in Neuroimaging Informatics Technology Initiative (NIfTI) (.nii) format.
Secondary Results
Other study results include the average time required to annotate one slice using the proposed technique. This time ranged from 25 seconds to 2 minutes for complex cases, with an average processing time of 0.8 hours for 100 CT angiography slices. The estimated effort required to replicate or scale up this study is represented using the segmentation time. Object segmentation allows for using the study results for three-dimensional modeling of vessels. This type of modeling is necessary for developing medical training simulators and addressing research challenges in hemodynamics, as previously demonstrated [21, 22].
DISCUSSION
Summary of Primary Results
A dataset containing 100 CT angiography scans was created using a previously developed methodology for abdominal aortic segmentation for cases involving a normal aorta, lumen expansion, thrombosis, and wall calcification.
Discussion of Primary Results
A few heterogeneous results were revealed by searching well-known open-source datasets such as Google datasets,8 Kaggle datasets,9 Roboflow Universe,10 and Images.cv.11 Radl et al. [7] presented a dataset containing 56 aortic CT angiography scans (1 aneurysm, 5 dissections, and 50 normal cases), annotated with the aortic arch and its branches and the abdominal aorta and iliac arteries. The authors collected raw data from multiple sources. The method for image annotation using 3D Slicer was described. However, the number of annotators and their qualifications were not specified. Yaneth1 claimed that the proposed dataset resulted from a CT-based neural network segmentation of the abdominal aorta. It contains little supporting information. Notably, the author used machine learning to prepare and segment the images. The 5289 images were divided into 13 files, which included scan results and masks. Data on the normal-to-abnormal ratio of the dataset, order of image annotation, and requirements for annotators and population data are not available. The publicly available dataset new-workspace-xlgzg2 contains 936 JPEG images of the abdominal aorta. These include various CT slices; however, the parameters and annotation data are not specified. The results were supplemented by an additional search of Google Scholar publications referencing open-access datasets using an Abdominal Aorta Computed Tomographic Angiography Segmentation Dataset query. Imran et al. [8] annotated 59 CTA images of the abdominal aorta. These images had a resolution of 512 × 512 pixels, with a slice thickness of 0.8–2.0 mm. These images were obtained from a study of 734 patients with aortic dissection included in the authors’ database. Postgraduate students used 3D Slicer to semi-automatically segment the aorta and its branches. Information about the medical specialization of the annotators and experience was not provided. Fantanzinni et al. [9] prepared a dataset containing aortic and branch segmentation based on the results of 80 preoperative CT angiography scans of patients diagnosed with abdominal aortic aneurysm at the University Hospital San Martino, Italy. The dataset does not include patients with other aortic conditions. Expert radiologists were involved in aortic segmentation using ITK-SNAP.12 Moreover, the dataset does not include images of the normal aorta or other abnormalities besides aneurysms. In addition, the present study does not provide a direct link to dataset downloading. Jung et al. [10] provided a similar description. The authors comprehensively described their dataset, which contains 60 CT angiography scans of patients with abdominal aortic aneurysms from Gachon University Hospital in Korea. A radiologist with over 10 years of experience annotated the scans.
Comparison of the obtained dataset with similar, publicly accessible data revealed several advantages and disadvantages. The developed dataset has the following advantages:
- Volume exceeding that of known analogs (100 CTA scans);
- A unique, simultaneous presentation of cases involving the normal abdominal aorta, aneurysm, calcification, and thrombosis;
- Inclusion of multicenter scans;
- Expert verification of primary annotation;
- Manual pixel-by-pixel annotation of the area of interest;
- Open declaration of population data, the annotation scenario, and requirements for annotators.
Study Limitations
The small number of included abnormalities is a limitation of both the dataset and study. Despite its limitations, this dataset can be used for many purposes, including to classify different types of atherosclerotic abdominal aortic lesions and develop anthropomorphic aortic models with various wall abnormalities.
CONCLUSION
A dataset containing 100 CT angiography results with abdominal aorta segmentation for normal cases, aneurysm, thrombosis, and calcification was created using the open-source 3D Slicer software and a previously developed original segmentation method. The dataset is publicly available and can be used for developing and testing AI algorithms and anthropomorphic three-dimensional modeling of the abdominal aorta.
ADDITIONAL INFORMATION
Funding source. This article was prepared by a group of authors as a part of the research and development effort titled "Development of software for automated generation of data sets containing synthetic native-phase CT studies to train and validate AI algorithms", (USIS No. 123031500002-1) in accordance with the Order No. 1196 dated December 21, 2022 "On approval of state assignments funded by means of allocations from the budget of the city of Moscow to the state budgetary (autonomous) institutions subordinate to the Moscow Health Care Department, for 2023 and the planned period of 2024 and 2025" issued by the Moscow Health Care Department.
Disclosure of interests. The authors declare that they have no relationships, activities or interests (personal, professional or financial) with third parties (commercial, non-commercial, private) whose interests may be affected by the content of the article, as well as no other relationships, activities or interests over the past three years that must be reported.
Authors’ contribution. M.R. Kodenko: research methodology, literature review, data analysis, writing and editing the article; Yu.A. Vasilev: research planning and expert analysis of the article text; D.V. Gatin, A.V. Solovev, E.P. Yasakova: data segmentation and editing the article; A.V. Guseva editing the article; R.V. Reshetnikov: expert analysis and editing the article. Thereby, all authors provided approval of the version to be published and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
1 Aorta segmentation [Internet]. In: Kaggle; 2018–2024. Available at: https://www.kaggle.com/datasets/licethyaneth/aorta-segmentation
Accessed on May 22, 2024.
2 AAAtester computer vision project [Internet]. In: Roboflow; 2022–2024. Available at: https://universe.roboflow.com/new-workspace-xlgzg/aaatester Accessed on May 22, 2024.
3 Unified Radiological Information Service [Internet]. In: Unified Medical Information and Analytical System of Moscow; 2020–2023. Available at: https://telemedai.ru/proekty/edinyj-radiologicheskij-informacionnyj-servis_2020 Accessed on November 21, 2023.
4 3D Slicer image computing platform. В: 3D Slicer [Internet]. 2005–2022. Available at: https://slicer.org/ Accessed on September 11, 2022.
5 Presize: precision based sample size calculation [Internet]. In: The Swiss Clinical Trial Organisation; 2021–2022.
6 Descriptive Statistics in Excel; [≈14 pages]. In: Statistics By Jim [Internet]. 2021–2024. Available at: https://statisticsbyjim.com/basics/descriptive-statistics-excel/ Accessed on April 22, 2024.
7 Computed Tomography Angiography Dataset Containing Features of Calcification, Thrombosis, Dilation, and Aneurysms, with Abdominal Aortic Lumen and Wall Segmentation [Internet]. Moscow: Research and Practical Clinical Center for Diagnostics and Telemedicine Technologies of Moscow Department of Health; 2024–2024. Available at: https://mosmed.ai/datasets/nabor-dannih-kompyuterno-tomograficheskoi-angiografii-s-priznakami-kaltsinoza-tromboza-dilatatsii-i-anevrizmi-i-soderzhaschii-segmentatsiyu-prosveta-i-stenki-bryushnogo-otdela-aorti/ Accessed on April 20, 2024.
8 Dataset Search [Internet]. In: Google; 2018–2024. Available at: https://datasetsearch.research.google.com/ Accessed on March 22, 2024.
9 Datasets [Internet]. In: Kaggle; 2021–2024. Available at: https://www.kaggle.com/datasets Accessed on March 22, 2024.
10 Explore the Roboflow Universe [Internet]. In: Roboflow; 2021–2023. Available at: https://universe.roboflow.com/ Accessed on March 22, 2024.
11 Labeled image datasets for computer vision [Internet]. In: Images.cv; –2024. Available at: https://images.cv/ Accessed on: March 22, 2024.
12 ITK-SNAP Home [Internet]. In: ITK-SNAP; 2020–2024. Available at: http://www.itksnap.org/pmwiki/pmwiki.php Accessed on March 22, 2024.
About the authors
Maria R. Kodenko
Research and Practical Clinical Center for Diagnostics and Telemedicine Technologies; Bauman Moscow State Technical University
Author for correspondence.
Email: m.r.kodenko@yandex.ru
ORCID iD: 0000-0002-0166-3768
SPIN-code: 5789-0319
Cand. Sci. (Engineering)
Russian Federation, Moscow; MoscowYuriy A. Vasilev
Research and Practical Clinical Center for Diagnostics and Telemedicine Technologies
Email: VasilevYA1@zdrav.mos.ru
ORCID iD: 0000-0002-5283-5961
SPIN-code: 4458-5608
MD, Cand. Sci. (Medicine)
Russian Federation, MoscowAlexander V. Solovev
Research and Practical Clinical Center for Diagnostics and Telemedicine Technologies; Morozov Children's City Clinical Hospital
Email: SolovevAV10@zdrav.mos.ru
ORCID iD: 0000-0003-4485-2638
SPIN-code: 9654-4005
Russian Federation, Moscow; Moscow
Denis V. Gatin
Research and Practical Clinical Center for Diagnostics and Telemedicine Technologies
Email: GatinDV@zdrav.mos.ru
ORCID iD: 0000-0002-6218-3012
SPIN-code: 2256-3564
Russian Federation, Moscow
Elena P. Yasakova
Research and Practical Clinical Center for Diagnostics and Telemedicine Technologies
Email: YasakovaEP@zdrav.mos.ru
ORCID iD: 0000-0003-0315-5502
SPIN-code: 1047-4692
MD, Cand. Sci. (Medicine)
Russian Federation, MoscowAnastasia V. Guseva
Research and Practical Clinical Center for Diagnostics and Telemedicine Technologies; Bauman Moscow State Technical University
Email: GusevaAV13@zdrav.mos.ru
ORCID iD: 0009-0006-1787-4726
SPIN-code: 2778-3820
Russian Federation, Moscow; Moscow
Roman V. Reshetnikov
Research and Practical Clinical Center for Diagnostics and Telemedicine Technologies
Email: ReshetnikovRV1@zdrav.mos.ru
ORCID iD: 0000-0002-9661-0254
SPIN-code: 8592-0558
Cand. Sci. (Physics and Mathematics)
Russian Federation, MoscowReferences
- Kumar DS, Bhat V, Gadabanahalli K, Kalyanpur A. Spectrum of abdominal aortic disease in a tertiary health care setup: MDCT based observational study. J Clin Diagn Res. 2016;10(11):TC24–TC29. doi: 10.7860/JCDR/2016/21373.8928
- Russian Society of Angiologists and Vascular Surgeons. Abdominal aortic aneurysm: clinical guidelines [Internet]. Moscow: Russian Society of Angiologists and Vascular Surgeons; 2022 [cited 2022 Apr 7]. (In Russ.) Available from: https://angiolsurgery.org/library/recommendations/2022/aneurysm/recommendation.pdf
- Baliyan V, Shaqdan K, Hedgire S, Ghoshhajra B. Vascular computed tomography angiography technique and indications. Cardiovascular Diagnosis and Therapy. 2019;9(S1):S14S27. doi: 10.21037/CDT.2019.07.04 EDN: IPZHHC
- Alowais ShA, Alghamdi SS, Alsuhebany N, et al. Revolutionizing healthcare: the role of artificial intelligence in clinical practice. BMC Medical Education. 2023;23(1):689. doi: 10.1186/s12909-023-04698-z EDN: AJSDXW
- Ueda D, Kakinuma T, Fujita SH, et al. Fairness of artificial intelligence in healthcare: review and recommendations. Japanese Journal of Radiology. 2023;42(1):3–15. doi: 10.1007/s11604-023-01474-3 EDN: WQQDIA
- Shchupakova AN, Litvyakov AM. Characteristics of atherosclerotic lesion of the abdominal aorta and its unpaired visceral branches in patients with chronic abdominal ischemia. Terapevticheskii arkhiv. 2004;79(6):70–74. EDN: OJZUCJ
- Radl L, Jin YU, Pepe A, et al. AVT: Multicenter aortic vessel tree CTA dataset collection with ground truth segmentation masks. Data in Brief. 2022;40:107801. doi: 10.1016/j.dib.2022.107801 EDN: PEOYKJ
- Imran M, Kreds JR, Gopu VRR, et al. CIS-UNet: Multi-class segmentation of the aorta in computed tomography angiography via context-aware shifted window self-attention. Computerized Medical Imaging and Graphics. 2024;118:102470. doi: 10.1016/j.compmedimag.2024.102470
- Fantazzini A, Esposito M, Finotello A, et al. 3D automatic segmentation of aortic computed tomography angiography combining multi-view 2D convolutional neural networks. Cardiovascular Engineering and Technology. 2020;11(5):576–586. doi: 10.1007/s13239-020-00481-z EDN: FHKUXK
- Jung Y, Kim S, Kim J, et al. Abdominal aortic thrombus segmentation in postoperative computed tomography angiography images using Bi-directional cnvolutional long short-term memory architecture. Sensors. 2022;23(1):175. doi: 10.3390/s23010175 EDN: SGCHXK
- Norgeot B, Quer G, Beaulieu-Jones BK, et al. Minimum information about clinical artificial intelligence modeling: the MI-CLAIM checklist. Nature Medicine. 2020;26(9):1320–1324. doi: 10.1038/s41591-020-1041-y EDN: NRQASJ
- Vasilev YuA, Arzamasov KM, Vladzymyrskyy AV, et al. Preparing a dataset for training and testing software based on artificial intelligence technology: a training manual. Moscow: Moscow Center for Diagnostics and Telemedicine; 2023. (In Russ.) EDN: OGKFGM
- Tymkovich MYu, Avruninn OG, Semenets VV. Using DICOM images in medical systems. Technical Electrodynamics. 2012;(thematic issue):178–183. (In Russ.)
- Li X, Morgan P, Ashburner J, et al. The first step for neuroimaging data analysis: DICOM to NIfTI conversion. Journal of Neuroscience Methods. 2016;264:47–56. doi: 10.1016/j.jneumeth.2016.03.001
- Kodenko MR, Vasilev YuA, Vladzymyrskyy AV. Segmentation of arterial vessels based on CT angiography data using 3D Slicer software: Guidelines. Moscow: Moscow Center for Diagnostics and Telemedicine; 2024. (In Russ.) EDN: CYLZQL
- Kodenko MR, Makarova TA. Preparation of abdominal computed tomography data set for patients with abdominal aortic aneurysm. Digital Diagnostics. 2023;4(1S):90–92. doi: 10.17816/DD430355 EDN: SIUWRL
- Riley RD, Snell KIE, Archer L, et al. Evaluation of clinical prediction models (part 3): calculating the sample size required for an external validation study. BMJ. 2024;384:e074821. doi: 10.1136/bmj-2023-074821
- Hazra A. Using the confidence interval confidently. Journal of Thoracic Disease. 2017;9(10):4124–4129. doi: 10.21037/jtd.2017.09.14
- Vasilev YuA, Vladzymyrskyy AV, Arzamasov KM, et al. Computer vision in radiation diagnostics: the first stage of the Moscow experiment. [Internet]. Moscow: Izdatel'skie resheniya; 2022 [cited 2024 Apr 22]. Available from: https://telemedai.ru/biblioteka-dokumentov/kompyuternoe-zrenie-v-luchevoj-diagnostike-pervyj-etap-moskovskogo-eksperimenta
- Certificate of state registration of the database N 2024621990/ 08.05.2024. Byul. N 5. Vasilev YuА, Kodenko МR, Solovev AV, et al. A computed tomography angiography dataset showing calcification, thrombosis, dilation, and aneurysm, and containing segmentation of the lumen and wall of the abdominal aorta. Available from: https://elibrary.ru/download/elibrary_67262583_69739205.PDF (In Russ.) EDN: FOPTBQ
- Guseva AV, Kodenko MR. Anthropomorphic abdominal aortic phantoms for computed tomography angiography. Digital Diagnostics. 2024;5(1S):27–29. doi: 10.17816/DD626820 EDN: BMDJUN
- Lesage D, Angelini ED, Bloch I, Funka-Lea G. A review of 3D vessel lumen segmentation techniques: models, features and extraction schemes. Medical image analysis. 2009;13(6):819–845. doi: 10.1016/j.media.2009.07.011
Supplementary files
