Deep learning models can classify brain tumors from a single MRI scan

Oak Creek, sick

Satrajit Chakrabarty ms, a doctoral student under the supervision of Dr.

The six most common types of intracranial tumors are high-grade glioma, low-grade glioma, brain metastases, meningioma, pituitary adenoma, and acoustic neuroma

Chakrabarty said that machines and deep learning methods using MRI data may automatically detect and classify brain tumors

"Non-invasive MRI can be used as a supplement, or in some cases, as an alternative to histopathological examination," he said

To build their machine learning model (called a convolutional neural network), Chakrabarty and researchers from the Mallinckrodt Institute of Radiology developed a large, multi-institutional, intracranial 3D MRI scan data set from four public sources

The researchers divided a total of 2,105 scans into three data subsets: 1396 for training, 361 for internal testing, and 348 for external testing

Using internal detection data, the model achieved an accuracy of 93.

For an external test data set containing only two tumor types (high-grade glioma and low-grade glioma), the accuracy of the model was 91.

Chakrabarty said: "These results show that deep learning is a very promising method to automatically classify and evaluate brain tumors

Chakrabarty said that by improving the existing 2D methods, the 3D deep learning model is closer to the goal of achieving end-to-end automated workflows

Dr.

Chakrabarty added: “This network is the first step in the development of an AI-enhanced radiology workflow that can support image interpretation by providing quantitative information and statistics

"MRI-based Identification and Classification of Major Intracranial Tumor Types Using a 3D Convolutional Neural Network: A Retrospective Multi-Institutional Analysis.
" Collaborating with Satrajit Chakrabarty and Drs.
Sotiras and Marcus were Mikhail Milchenko, Ph.
D.
, Pamela LaMontagne, Ph .
D.
, and Michael Hileman, BS