An accessible deep learning tool for voxel-wise classification of brain malignancies from perfusion MRI.

in Cell reports. Medicine by Alonso Garcia-Ruiz, Albert Pons-Escoda, Francesco Grussu, Pablo Naval-Baudin, Camilo Monreal-Aguero, Gretchen Hermann, Roshan Karunamuni, Marta Ligero, Antonio Lopez-Rueda, Laura Oleaga, M Álvaro Berbís, Alberto Cabrera-Zubizarreta, Teodoro Martin-Noguerol, Antonio Luna, Tyler M Seibert, Carlos Majos, Raquel Perez-Lopez

TLDR

  • The study uses deep learning to look at pictures of the brain and figure out if there's a tumor. The model is trained on pictures of tumors and it can tell if a new picture has a tumor or not. The model is very good at this and it can help doctors make better decisions about brain tumors.

Abstract

Noninvasive differential diagnosis of brain tumors is currently based on the assessment of magnetic resonance imaging (MRI) coupled with dynamic susceptibility contrast (DSC). However, a definitive diagnosis often requires neurosurgical interventions that compromise patients' quality of life. We apply deep learning on DSC images from histology-confirmed patients with glioblastoma, metastasis, or lymphoma. The convolutional neural network trained on ∼50,000 voxels from 40 patients provides intratumor probability maps that yield clinical-grade diagnosis. Performance is tested in 400 additional cases and an external validation cohort of 128 patients. The tool reaches a three-way accuracy of 0.78, superior to the conventional MRI metrics cerebral blood volume (0.55) and percentage of signal recovery (0.59), showing high value as a support diagnostic tool. Our open-access software, Diagnosis In Susceptibility Contrast Enhancing Regions for Neuro-oncology (DISCERN), demonstrates its potential in aiding medical decisions for brain tumor diagnosis using standard-of-care MRI.

Overview

  • The study aims to develop a deep learning model that can accurately diagnose brain tumors using dynamic susceptibility contrast (DSC) images from histology-confirmed patients with glioblastoma, metastasis, or lymphoma. The model is trained on ∼50,000 voxels from 40 patients and provides intratumor probability maps that yield clinical-grade diagnosis. The study tests the performance of the model in 400 additional cases and an external validation cohort of 128 patients. The primary objective of the study is to demonstrate the potential of the model as a support diagnostic tool for brain tumor diagnosis using standard-of-care MRI.

Comparative Analysis & Findings

  • The study compares the accuracy of the deep learning model with conventional MRI metrics, cerebral blood volume (0.55) and percentage of signal recovery (0.59). The model achieves a three-way accuracy of 0.78, which is superior to the conventional MRI metrics. The study also demonstrates the potential of the model as a support diagnostic tool for brain tumor diagnosis using standard-of-care MRI.

Implications and Future Directions

  • The study's findings suggest that deep learning models can accurately diagnose brain tumors using DSC images. The open-access software, Diagnosis In Susceptibility Contrast Enhancing Regions for Neuro-oncology (DISCERN), demonstrates the potential of deep learning models as a support diagnostic tool for brain tumor diagnosis using standard-of-care MRI. Future research could focus on improving the accuracy of the model and expanding its use to other types of brain tumors.
Read Full Article