![]() To present SPECTR, we proceed through the different steps described next: ( a) breaking down the analysis workflow performed by human experts into simple short tasks ( Fig. We herein present the development of SPECTR (serum protein electrophoresis computer-assisted recognition), an AI-based tool that performs a complete interpretation of SPE, from raw curves associated with clinically relevant data (sex, age, serum total protein concentration) into a text comments output. Several machine learning-based algorithms have been used to analyze SPE curves ( 20–23) however, to our knowledge, no method has achieved a complete automation of the whole SPE analysis up to medical interpretation. In the past decade, major advances in deep learning-based algorithms ( 13) have supported the development of artificial intelligence (AI) in the medical field, allowing the achievement of human expert-level in multiple medical disciplines ( 14–17), but there are still very few applications in medical biology ( 18, 19). This heavy reliance on an experienced operator limits the throughput of the SPE in clinical laboratories and hampers interoperator and interlaboratory harmonization and reproducibility of the techniques ( 10–12). Indeed, the interpretation of the curves may be affected by both endogenous and exogenous potential interfering products and multiple physiological or pathological conditions ( 6–9) and is subject to a thorough understanding of the clinical context, requiring an experienced operator ( 1, 2). ![]() SPE is divided into 2 steps: ( a) first, a technical step consisting in the acquisition of a curve representing the migration of serum proteins in an electrical field according to their charge, mass, and shape, historically using gel and more recently capillary electrophoresis ( b) second, the interpretation of this curve to highlight clinically relevant abnormalities, such as a spike caused by the presence of a monoclonal immunoglobulin (M-spike) ( 2).Īlthough the first step is now usually fully automatized using high-throughput devices capable of analyzing up to 100 sera per hour ( 5), the second step still requires human intervention. ![]() The main indication for SPE is the diagnosis and follow-up of monoclonal gammopathies such as myeloma, Waldenström disease, and monoclonal gammopathy of undetermined significance ( 2), as well as many clinical conditions that also cause more or less specific changes in the electropherogram of serum proteins ( 3, 4). Serum protein electrophoresis (SPE) is a widely used biochemical test, representing a major part of the biological tests performed in clinical laboratories ( 1). Artificial intelligence, deep learning, serum protein electrophoresis, myeloma, monoclonal gammopathy, neural network Introduction ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |