Machine learning differentiates brain tumors

Tuesday, November 28 | 3:10 p.m.-3:20 p.m. | SSJ19-02 | Room N228
Machine learning can distinguish between glioblastoma multiforme and primary central nervous system lymphoma on multiparametric MRI, according to researchers from Japan.

While the standard treatment for glioblastoma multiforme (GBM) is resection, resection for primary central nervous system lymphoma (PCNSL) has no survival benefits and is associated with an increase in postoperative deficits, said presenter Dr. Takeshi Nakaura of Kumamoto University. As a result, differential diagnosis of these tumors is important for treatment planning.

"However, MR imaging features of these tumors are highly variable and overlapping by the variety of tumor homogeneity, and previous reports suggested that the [differential] diagnosis of GBM and PCNSL is difficult in some cases," Nakaura said. "Therefore, I think it is well-suited [for] machine learning based on texture analysis."

The researchers developed a machine-learning model for differentiating between GBM and PCNSL based on texture features on multiparametric MRI. They compared the results with the performance of three experienced radiologists.

"The performance of machine learning based on texture futures in multiparametric MRI was superior to that of the conventional cutoff method and board-certified radiologists in differentiating GBM and PCNSL," Nakaura told AuntMinnie.com.

Check out this Tuesday afternoon talk for all the details.

Page 1 of 605
Next Page