Related Experiment Video
Updated: Aug 5, 2025
Role of Diffusion MRI Tractography in Endoscopic Endonasal Skull Base Surgery
Published on: July 5, 2021
Uncertainty-Aware Deep Learning Classification of Adamantinomatous Craniopharyngioma from Preoperative MRI
Eric W Prince1,2,3, Debashis Ghosh2, Carsten Görg2
1Department of Neurosurgery, University of Colorado School of Medicine, Aurora, CO 80045, USA.
Abstract:
Diagnosis of adamantinomatous craniopharyngioma (ACP) is predominantly determined through invasive pathological examination of a neurosurgical biopsy specimen. Clinical experts can distinguish ACP from Magnetic Resonance Imaging (MRI) with an accuracy of 86%, and 9% of ACP cases are diagnosed this way. Classification using deep learning (DL) provides a solution to support a non-invasive diagnosis of ACP through neuroimaging, but it is still limited in implementation, a major reason being the lack of predictive uncertainty representation. We trained and tested a DL classifier on preoperative MRI from 86 suprasellar tumor patients across multiple institutions. We then applied a Bayesian DL approach to calibrate our previously published ACP classifier, extending beyond point-estimate predictions to predictive distributions. Our original classifier outperforms random forest and XGBoost models in classifying ACP. The calibrated classifier underperformed our previously published results, indicating that the original model was overfit. Mean values of the predictive distributions were not informative regarding model uncertainty. However, the variance of predictive distributions was indicative of predictive uncertainty. We developed an algorithm to incorporate predicted values and the associated uncertainty to create a classification abstention mechanism. Our model accuracy improved from 80.8% to 95.5%, with a 34.2% abstention rate. We demonstrated that calibration of DL models can be used to estimate predictive uncertainty, which may enable clinical translation of artificial intelligence to support non-invasive diagnosis of brain tumors in the future.
Insights
Deep learning models can now estimate uncertainty for non-invasive adamantinomatous craniopharyngioma (ACP) diagnosis using MRI. This approach improves accuracy by allowing the model to abstain from uncertain classifications.
Area of Science:
- Neuroimaging
- Artificial Intelligence
- Oncology
Background:
- Adamantinomatous craniopharyngioma (ACP) diagnosis typically requires invasive neurosurgical biopsies.
- Current deep learning (DL) models for ACP diagnosis lack predictive uncertainty representation, limiting clinical implementation.
- Radiologists achieve 86% accuracy in distinguishing ACP from other tumors via MRI, with 9% diagnosed non-invasively.
Purpose of the Study:
- To develop and validate a Bayesian deep learning approach for non-invasive ACP diagnosis using MRI.
- To incorporate predictive uncertainty estimation into DL classification for improved diagnostic reliability.
- To establish a classification abstention mechanism based on model uncertainty.
Main Methods:
- Trained and tested a DL classifier on preoperative MRI scans from 86 suprasellar tumor patients.
- Applied a Bayesian DL approach to calibrate the existing ACP classifier, focusing on predictive distributions.
- Developed an algorithm integrating predicted values and uncertainty for a classification abstention mechanism.
Main Results:
- The calibrated DL model achieved 95.5% accuracy with a 34.2% abstention rate, significantly improving upon the initial 80.8% accuracy.
- Predictive distribution variance, not mean values, effectively indicated model uncertainty.
- The abstention mechanism enhanced diagnostic performance by deferring uncertain cases.
Conclusions:
- Bayesian DL calibration effectively estimates predictive uncertainty in neuroimaging AI models.
- Incorporating uncertainty estimation and abstention mechanisms can significantly improve AI diagnostic accuracy for brain tumors.
- This approach holds promise for the clinical translation of AI in non-invasive tumor diagnosis.
More Related Videos
13:12Translational Brain Mapping at the University of Rochester Medical Center: Preserving the Mind Through Personalized Brain Mapping
Published on: August 12, 2019
04:25Author Spotlight: Bridging Gaps in Anatomy and Establishing a Foundation for Algorithmic Studies
Published on: December 15, 2023
