Related Experiment Video
Updated: Sep 28, 2025
02:42
Analysis of Craniomaxillofacial Malformations in Mice Using Three-dimensional Microcomputed Tomography
Published on: January 17, 2025
475
Rapid Automated Analysis of Skull Base Tumor Specimens Using Intraoperative Optical Imaging and Artificial
Cheng Jiang1, Abhishek Bhattacharya2, Joseph R Linzey3
1Department of Computational Medicine and Bioinformatics, University of Michigan, Ann Arbor, Michigan, USA.
Neurosurgery
|March 28, 2022
Summary
Artificial intelligence and stimulated Raman histology (SRH) enable rapid, accurate intraoperative analysis of skull base tumors. This AI-powered approach aids surgeons in making critical decisions during procedures.
Area of Science:
- Oncology
- Medical Imaging
- Artificial Intelligence
Background:
- Accurate intraoperative analysis of skull base tumors is crucial for personalized surgical treatment.
- Interpreting skull base pathologies during surgery is challenging due to complexity and limited resources.
Purpose of the Study:
- Develop an independent, parallel intraoperative workflow for rapid and accurate skull base tumor analysis.
- Utilize label-free optical imaging and artificial intelligence for specimen evaluation.
Main Methods:
- Employed stimulated Raman histology (SRH), a label-free, high-resolution microscopy technique (<60 seconds per 1 × 1 mm2).
- Trained convolutional neural network (CNN) models using cross-entropy, self-supervised contrastive learning, and supervised contrastive learning on SRH images.
- Validated CNN models on a held-out, multicenter SRH dataset.
Main Results:
- SRH successfully imaged diagnostic features of benign and malignant skull base tumors.
- Supervised contrastive learning demonstrated superior performance, achieving 96.6% diagnostic accuracy.
- The AI model accurately segmented tumor-normal margins and detected microscopic tumor infiltration.
Conclusions:
- AI models combined with SRH offer rapid and precise intraoperative analysis of skull base tumor specimens.
- This technology can significantly inform surgical decision-making in real-time.
