Artificial intelligence can distinguish between hematologic neoplasms and solid tumors with high accuracy, according to a new study published in The American Journal of Pathology.

Geometric mean naïve Bayesian algorithm showed correct first-choice diagnosis in the majority of acute lymphoblastic leukemia, acute myeloid leukemia, diffuse large B-cell lymphoma, colorectal cancer, lung cancer, chronic lymphocytic leukemia, and follicular lymphoma cases.

“We conclude that targeted transcriptome combined with artificial intelligence are highly useful for diagnosis and classification of various cancers,” the study authors wrote. “Mutation profiles and clinical information can improve these algorithms and minimize errors in diagnoses.”

Read more about the diagnosis of diffuse large B-cell lymphoma

The diagnosis and classification of tumors depend on biomarkers, and RNA expression profiles can provide reliable and reproducible information about the biology of the tumors.

In the present study, a team of researchers, led by Maher Albitar, MD, from the Genomic Testing Cooperative in Irvine, California, and Hackensack Meridian School of Medicine, in Nutley, New Jersey, assessed the potential of targeted transcriptome and artificial intelligence in differentially diagnosing hematologic and solid tumors. 

The team sequenced RNA samples obtained from 2606 hematologic neoplasms, 2038 solid tumors, 782 bone marrow without cancer, and 24 lymph node controls using a targeted 1408-gene panel.

They showed that machine learning was able to distinguish between hematologic neoplasms and solid tumors with an area under the curve between 1 and 0.841.

The algorithm showed the correct first-choice diagnosis in all cases of acute lymphoblastic leukemia. It also showed correct first-choice diagnosis in 88% of acute myeloid leukemia and lung cancer cases, 82% of colorectal cancer cases, and 72% of chronic lymphocytic leukemia and follicular lymphoma cases. The algorithm was able to correctly identify 85% of diffuse large B-cell lymphoma cases as the first-choice diagnosis.

Reference

Zhang H, Qureshi MA, Wahid M, et al. Differential diagnosis of hematologic and solid tumors using targeted transcriptome and artificial intelligence. Am J Pathol. 2022;13:S0002-9440(22)00313-3. doi:10.1016/j.ajpath.2022.09.006