Please use this identifier to cite or link to this item:
http://hdl.handle.net/2445/180158
Title: | ARTERIAL an AI framework for mechanical thrombectomy Planning through the automated characterization of vascular tortuosity |
Author: | Balocco, Simone Díaz, Oliver Canals Canals, Pere Ribó Jacobi, Marc |
Keywords: | Processament d'imatges Malalties cerebrovasculars Programari Image processing Cerebrovascular disease Computer software |
Issue Date: | 21-Sep-2021 |
Publisher: | Universitat de Barcelona |
Abstract: | [cat] Arterial és un programari d’anàlisi d’imatges mèdiques per a cirurgia de l’ictus, mitjançant l'avaluació quantitativa de la tortuositat vascular, derivada de la imatge mèdica abans de la intervenció en el context de l’ictus. L’objectiu final de l’Arterial és optimitzar el tractament endovascular de l 'ictus isquèmic agut, mitjançant prediccions rellevants abans de la intervenció, de manera que els neurointervencionista pot prendre decisions sobre les possibilitats de tractament de cada individu. [eng] Arterial framework is an AI-powered medical image analysis framework for stroke surgical planning through the quantitative assessment of vascular tortuosity, derived from medical imaging prior to intervention in the stroke context. The ultimate goal of Arterial is to optimize endovascular treatment for acute ischemic stroke by delivering relevant predictions regarding the difficulty of endovascular treatment in stroke patients prior to intervention, such that neurointerventionalists can make supported decisions regarding the treatment possibilities for each individual patient, and advanced towards a more personalized medicine through the power of AI. |
Note: | System requirements: For deployment, Arterial has been partly tested on Ubuntu 18.04 and 20.04. Check for Cuda compatibility (cuDNN, cudatoolkit) between your Nvidia GPU and the PyTorch version installed through the nnU-Net package5 . Your GPU
should have at least 3GB of VRAM to correctly perform inference with the segmentation models (otherwise, you could run into memory problems).
For development, Arterial, has been tested on MacOS 10.15 or newer, excluding the segmentation module due to the lack of Cuda support. L'accés als materials dipositats no estarà disponible fins la fi de la data d'embargament. Si esteu interessats a accedir-hi, contacteu amb idea@fbg.ub.edu |
URI: | http://hdl.handle.net/2445/180158 |
Appears in Collections: | Programari - Recerca Programari (Matemàtiques i Informàtica) |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
arterial_code.tar.gz | Code | 1.36 GB | Unknown | View/Open Request a copy |
User_manual_Arterial.pdf | Manual | 2.15 MB | Adobe PDF | View/Open Request a copy |
README.md | 5.76 kB | Unknown | View/Open Request a copy | |
requirements.txt | 91 B | Text | View/Open Request a copy |
Document embargat fins el 21-9-2026
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.