Department of Electrical Engineering, Faculty of Physical and Mathematical Sciences, University of Chile, Santiago, Chile.
Millennium Institute for Intelligent Healthcare Engineering, Santiago, Chile.
Childs Nerv Syst. 2024 Aug;40(8):2295-2300. doi: 10.1007/s00381-024-06400-0. Epub 2024 Apr 22.
We studied a pediatric group of patients with sellar-suprasellar tumors, aiming to develop a convolutional deep learning algorithm for radiological assistance to classify them into their respective cohort.
T1w and T2w preoperative magnetic resonance images of 226 Chilean patients were collected at the Institute of Neurosurgery Dr. Alfonso Asenjo (INCA), which were divided into three classes: healthy control (68 subjects), craniopharyngioma (58 subjects) and differential sellar/suprasellar tumors (100 subjects).
The PPV among classes was 0.828±0.039, and the NPV was 0.919±0.063. Also explainable artificial intelligence (XAI) was used, finding that structures that are relevant during diagnosis and radiological evaluation highly influence the decision-making process of the machine.
This is the first experience of this kind of study in our institution, and it led to promising results on the task of radiological diagnostic support based on explainable artificial intelligence (AI) and deep learning models.
我们研究了一组鞍上-鞍旁肿瘤的儿科患者，旨在开发一种卷积深度学习算法，以协助放射学将其分类到各自的队列中。
我们收集了智利 226 名患者在神经外科研究所 Dr. Alfonso Asenjo（INCA）的术前 T1w 和 T2w 磁共振图像，并将其分为三组：健康对照组（68 例）、颅咽管瘤（58 例）和蝶鞍/鞍上差异肿瘤（100 例）。
各组间的阳性预测值为 0.828±0.039，阴性预测值为 0.919±0.063。此外，还使用了可解释的人工智能（XAI），发现对诊断和放射学评估相关的结构高度影响了机器的决策过程。
这是我们机构首次进行此类研究的经验，它在基于可解释人工智能（AI）和深度学习模型的放射学诊断支持任务中取得了有前景的结果。
