Sürdürülebilir Kalkınma Amaçları
Yayınlar
- MakaleAçık Erişim2025Scopus SJRQ1Comparison of surgical approaches to the hippocampal formation with artificial intelligence
Dundar, Tolga Turan; Pehlivanoğlu, Meltem Kurt; Eker, Ayşe Gül; Albayrak, Nur Banu; Mutluer, Ahmet Serdar; Yurtsever, İsmail; Doğan, İhsan; Duru, Nevcihan; Türe, Uğur
The relatively complex functional anatomy of the mediobasal temporal region makes surgical approaches to this area challenging. Several studies describe various surgical approaches, along with their combinations and modifications, to reach lesions of this region. Some of these surgical approaches have been compared using artificial intelligence-based approaches that can be predicted, classified, and analyzed for complex data. Several surgical approaches, such as anterior transsylvian, trans-superior temporal sulcus, trans-middle temporal gyrus, subtemporal–transparahippocampal, presigmoid-retrolabyrinthine, supratentorial-infraoccipital, and paramedian supracerebellar-transtentorial, were selected for comparison. Magnetic resonance images (MRIs) were taken according to the criteria specified by the Radiology Department. With an open-source software tool, volumetric data from cranial MRIs were segmented and anatomical structures in the main regions were reconstructed. The Q-learning algorithm was used to find pathways similar to these standard surgical pathways. The Q-learning scores among the selected pathways are as follows: anterior transsylvian (Q_A) = 31.01, trans-superior temporal sulcus (Q_B) = 25.00, trans-middle temporal gyrus (Q_C) = 28.92, subtemporal-transparahippocampal (Q_D) = 23.51, presigmoid- retrolabyrinthine (Q_E) = 27.54, supratentorial-infraoccipital (Q_F) = 27.2, and paramedian supracerebellar-transtentorial (Q _G) = 21.04. The Q-value score for the supracerebellar transtentorial approach was the highest among the examined approaches and therefore optimal. A difference was also found between the total risk score of all points with pathways drawn by clinicians and the total risk scores of the pathways formed and followed by Q-learning. Artificial intelligence-based approaches may significantly contribute to the success of the surgical approaches examined. Furthermore, artificial intelligence can contribute to clinical outcomes in both preoperative surgical planning and intraoperative technical equipment-assisted neurosurgery. However, further studies with more detailed data are needed for more sensitive results.
AtıfScopus2WoS2OpenAlex2
İncele - MakaleAçık Erişim2024Scopus SJRQ1BrainPixGAN: Generating intraoperative MRI images with mask-based generative networks
Eker, Ayşe Gül; Pehlivanoğlu, Meltem Kurt; Duru, Nevcihan; Dündar, Tolga Turan
In recent years, efforts to enhance precision in brain tumor surgeries have focused on using artificial intelligence (AI) with medical imaging. This involves integrating AI with medical imaging. This study aimed to generate a tumor-free MRI by using Generative Adversarial Networks (GANs) to establish a relationship between preoperative magnetic resonance imaging (MRI) and resection cavity segmentation masks obtained from intraoperative ultrasound (IOUS) data. For cavity segmentation, U-Net and U-Net with transfer learning were used, with the U-Net + EfficientNetB7 model achieving a high dice score of 97.82. The resection cavity mask was applied to preoperative MRI images using Pix2Pix, SPADE GAN, and BrainPixGAN. BrainPixGAN, incorporating transfer learning, outperformed the others, achieving SSIM 0.87, PSNR 35.89, and LPIPS 0.0037. This innovative approach represents a pioneering effort in generating GAN models for intraoperative MRI (iMRI) images using IOUS data, despite the challenges in setup and cost associated with iMRI imaging.
AtıfScopus13WoS10OpenAlex8İncele