Email this article Development of a three-dimensional convolutional neural network with attention for aneurysm detection
- Authors: Sinitsa S.G.1, Zyablova E.I.2, Kardailskaya D.O.2, Zayats I.A.1, Khalafyan A.A.1, Ishchenko A.V.3
-
Affiliations:
- Kuban State University
- Scientific Research Institute – S.V. Ochapovsky Regional Clinical Hospital № 1
- Limited Liability Company «KUB»
- Issue: No 2 (2024)
- Pages: 116-122
- Section: Machine Learning, Neural Networks
- URL: https://journal-vniispk.ru/2071-8594/article/view/265525
- DOI: https://doi.org/10.14357/20718594240209
- EDN: https://elibrary.ru/WGLAYC
- ID: 265525
Cite item
Full Text
Abstract
The paper considers a prototype of a three-dimensional convolutional neural network with an attention block detecting the probability of intracranial cerebral aneurysms in a single contrast computed tomography-angiography study. DICOM contrast computed tomography-angiography data with and without intracranial aneurysms were used to train the network. Metadata from the studies were not used. The data were divided into training and validation subsets in the proportion of 65% and 35%, respectively. Using Keras and Tensorflow libraries in the Python programming environment, a 192x192x128 three-dimensional convolutional neural network model with 4 convolutional layers, a kernel of dimension 3 and self-attention block was developed. The accuracy, precision and recall of classification on test samples reached 96%, 99% and 93% respectively that exceeded the performance of previously known neural networks.
About the authors
Sergey G. Sinitsa
Kuban State University
Author for correspondence.
Email: sin@kubsu.ru
Candidate of technical sciences. Assistant Professor, Information Technology Department
Russian Federation, KrasnodarElena I. Zyablova
Scientific Research Institute – S.V. Ochapovsky Regional Clinical Hospital № 1
Email: elenazyablova@inbox.ru
Candidate of medical sciences, doсent, Head of the Radiology Department
Russian Federation, KrasnodarDaria O. Kardailskaya
Scientific Research Institute – S.V. Ochapovsky Regional Clinical Hospital № 1
Email: k.daria2702@gmail.com
Radiologist, Assistant, Diagnostic Radiology Department No 2, Faculty of Continuing Professional Development and Retraining, Kuban State Medical University
Russian Federation, KrasnodarIlya A. Zayats
Kuban State University
Email: zayatsman@gmail.com
Student, Computer Technologies and Applied Mathematics Faculty
Russian Federation, KrasnodarAlexan A. Khalafyan
Kuban State University
Email: statlab@kubsu.ru
Doctor of technical sciences, doсent, Professor, Data Analysis and Artificial Intelligence Department, honorary educator of the Russian Federation
Russian Federation, KrasnodarAnton V. Ishchenko
Limited Liability Company «KUB»
Email: kub@kub.ru
Director, Head of the Pharmacy Products Search Service Apteki.su
Russian Federation, KrasnodarReferences
- Goodfellow J., Bengio I., Courville A. Deep Learning. MIT Press. 2016.
- Nikolenko S. Kadurin A., Arkhangelskaya E. Glubokoe obuchenie. Seriya Biblioteka programmista [Deep Learning. Programmer's Library series]. SPb: Peter. 2018.
- Prikhodko I.V., Erogodsky E.A., Kuzmina D.V., Sinitsa S.G., Zyablova E.I., Khalafyan A.A. Algoritmy poiska prostranstvennogo raspolozheniya intrakranial'nyh anevrizm [Algorithms for search of spatial location of intracranial aneurysms] // Materialy III vserossijskoj nauchno-prakticheskoj konferencii molodyh uchenyh [Proceedings of the III All-Russian Scientific and Practical Conference of Young Scientists]. Krasnodar, 2021. P. 313-320.
- Allison Park, Chris Chute, Pranav Rajpurkar, Joe Lou, Robyn L. Ball, Deep Learning–Assisted Diagnosis of Cerebral Aneurysms Using the HeadXNet Model // JAMA Network Open. URL: https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2735471.
- Konstantinos Kamnitsas, Christian Ledig, Virginia F.J. Newcombe, Joanna P. Simpson, Andrew D. Kane, David K. Menon, Daniel Rueckert, and Ben Glocker. Efficient Multi-Scale 3D CNN with Fully Connected CRF for Accurate Brain Lesion Segmentation // Medical Image Analysis. 2016. No 36.
- Sichtermann T, Faron A, Sijben R, Teichert N, Freiherr J, Wiesmann M. Deep Learning-Based Detection of Intracranial Aneurysms in 3D TOF-MRA. AJNR American Journal of Neuroradiology. 2019. No 40 (1). P. 25-32. doi: 10.3174/ajnr.A5911. Epub 2018 Dec 20. PMID: 30573461; PMCID: PMC7048599.
- Zunair Hasib, Rahman Aimon, Mohammed Nabeel, Cohen Joseph Paul. Uniformizing Techniques to Process CT Scans with 3D CNNs for Tuberculosis Prediction // International Workshop on PRedictive Intelligence In MEdicine. Springer. 2020. P. 156-168.
- Xin Zhang, Liangxiu Han, Wenyong Zhu, Liang Sun, Daoqiang Zhang. An Explainable 3D Residual Self-Attention Deep Neural Network For Joint Atrophy Localization and Alzheimer’s Disease Diagnosis using Structural MRI // IEEE Journal of Biomedical and Health Informatics. 2021. V. 26. No 11. P. 5289-5297.
- Zyablova E.I., Sinitsa S.G., Zayac I.A., Halafyan A.A., Kardail'skaya D.O., Porhanov V.A. Ispol'zovanie trekhmernyh svertochnyh nejronnyh setej dlya vyyavleniya intrakranial'nyh anevrizm po dannym kt-angiografii brahiocefal'nyh arterij [Use of three-dimensional convolutional neural networks for intracranial aneurysms detection based on brachiocephalic arteries CT-angiography] // Innovacionnaya medicina Kubani [Innovative Medicine of Kuban]. 2023. V. 8. No 2. P. 21-27.
- Vaswani A., Shazeer N., Parmar N., Uszkoreit J., Jones L., Gomez A. N., Kaiser L., Polosukhin I. Attention is all you need // Advances in Neural Information Processing Systems. MIT Press. 2017. No 30. P. 5998–6008.
Supplementary files
