Utilization of artificial intelligence in minimally invasive right adrenalectomy: recognition of anatomical landmarks with deep learning
These videos have been matched automatically. Contact us if they are not relevant.
These videos have been matched automatically. Contact us if they are not relevant.
View abstract on PubMed
Summary
This summary is machine-generated.Artificial intelligence accurately identifies key anatomical structures during minimally invasive right adrenalectomy. This technology can enable real-time surgical navigation systems for improved patient outcomes.
Area Of Science
- Medical Imaging
- Surgical Technology
- Artificial Intelligence
Background
- Minimally invasive adrenalectomy is the standard for adrenal mass removal.
- Accurate identification of anatomical landmarks is crucial to minimize surgical complications.
- Artificial intelligence (AI) offers potential for real-time navigation systems in surgery.
Purpose Of The Study
- To develop deep learning models for recognizing critical anatomical structures during minimally invasive right adrenalectomy.
- To evaluate the performance of transformer and convolutional neural network (CNN)-based models in semantic segmentation of surgical வீடியோs.
Main Methods
- Intraoperative videos from 20 patients undergoing minimally invasive right adrenalectomy were analyzed.
- SwinUNETR (transformer) and MedNeXt (CNN) models were trained for semantic segmentation of the liver, inferior vena cava (IVC), and right adrenal gland.
- Models were trained using Dice-Cross Entropy and Dice-Focal Loss, with 5-fold cross-validation and strong data augmentation.
Main Results
- The SwinUNETR model achieved a mean Dice Similarity Coefficient (mDSC) of 78.37% with Dice-Cross Entropy Loss.
- The MedNeXt model achieved a mean Intersection over Union (mIoU) of 63.71%, outperforming SwinUNETR's 62.10% mIoU.
- Both models demonstrated high performance in pixelwise classification metrics like accuracy, sensitivity, and specificity.
Conclusions
- AI-based systems can effectively predict anatomical landmarks during minimally invasive right adrenalectomy.
- These AI tools hold promise for the future development of real-time intraoperative navigation systems.
- The study demonstrates the feasibility of using deep learning for enhanced surgical guidance.
These videos have been matched automatically. Contact us if they are not relevant.
These videos have been matched automatically. Contact us if they are not relevant.