Researchers have developed a new method that automates the classification of progressive diabetic kidney disease, reducing variability and boosting precision, and may be applied to diagnose more severe diseases.
The new method uses a kind of “digital pipeline” to extract and classify important tissue structures in renal biopsies, thus help automated diagnosis.
The digital pipeline can literally diagnose with “one click,” according to senior author of the study, Dr. Pinaki Sarder. By analyzing the amount of disease-related changes observed in a renal biopsy, the diagnosis is made by the digital pipeline.
The focus is on a structure called the glomerulus, a sac-like bundle of capillaries that do first-line filtration of blood in the kidneys. Glomeruli in diabetic patients form deposition of scar tissue, which eventually prevents the glomerulus normal functions to filter the blood.
The researchers compared results from the digital pipeline and three renal pathologists, one of whom considered as a “gold standard” pathologist. The automated method agrees with the gold standard pathologist about 50% of the time, suggesting that is has learned how to diagnose the cases but also learned to develop its own opinion.
A 100% same results between the automated method and human pathologist is bad, because that means the method is learning too specifically to be like the specific doctor, with possible bias.
Mechanism of the method
The digital pipeline method is based on fixed mathematical measurements, which cannot be influenced by the person doing the analysis. In other words, the diagnosis of the automated method is independent from the pathologist who runs the algorithm.
"The algorithm is also flexible enough to accept any number of new or different features, but the features I selected are specifically targeted to quantify important characteristics of diabetic disease, such as scarring of the glomeruli, which is reported in past clinical literature.
"Having a set of features hand-crafted by a human is beneficial because it increases the ability of a clinician to interpret why the algorithm made a particular decision," said Brandon Ginley, the paper’s first author.
The information gathered from renal biopsies using this method could eventually allow clinicians to predict which diabetic patients are likely to develop more severe renal disease.