Classification accuracy of a hierarchical molecular inference-based deep-learning system for CNS tumour diagnosis: a multi-institutional, retrospective study

Key Points

• Classification accuracy of the deep-learning system meets high standards, improving CNS tumour diagnosis.
• The system achieved over 85% accuracy in distinguishing between different types of CNS tumours observed across multiple institutions.
• Retrospective study design utilized data from several institutions to validate the use of molecular inference with deep learning.
• This warrants further exploration in larger datasets and diverse clinical settings to enhance diagnostic reliability.