A research group, from Boston University School of Medicine, have applied machine learning frameworks — convolutional neural networks — for object recognition tasks in the medical setting. This has been applied first to the assessment of damaged kidneys.
Detecting kidney damage is of great importance: getting this wrong can lead to life-threatening conditions. The primary way by which medics assess kidney disease is through image analysis.
Images are taken following a visit of a patient to a hospital and where a biopsy is taken. With a renal biopsy a small piece of kidney is removed from the body for examination, under a microscope. This specialist procedure is called nephropathology. The microscopic examination of the tissue provides the information needed to diagnose, monitor or treat problems of the kidney.
Computer model based on artificial intelligence
This long-standing practice can now be augmented by the use of a computer model based on artificial intelligence (AI). The use of machine intelligence can help to improve the detection rate and provide data to assist medics in making point-of-care and clinical decisions.
As well as supporting qualified medics, some parts of the world do not have qualified nephropathologists in place. The use of computer assisted diagnostics therefore helps to boost local medical services.
As well as biopsy samples, the artificial intelligence can be applied to the analysis of radiology images. In trials, the research group have studied kidney biopsy sections with different degrees of kidney fibrosis. Renal fibrosis is characterized as a progressive detrimental connective tissue deposition on the kidney. This can lead to renal function deterioration.
Digitized images for medical practice
The new technology uses pixel density of digitized images, which are analysed by the convolutional neural networks. The platform has learned to assess the extent of a disease based in the degree of fibrosis assessed.
Discussing the technology with International Hospital magazine, lead researcher Dr. Vijaya B. Kolachalama said: “While the trained eyes of expert pathologists are able to gauge the severity of disease and detect nuances of kidney damage with remarkable accuracy, such expertise is not available in all locations, especially at a global level.”
The researcher is hopeful that the new technology will enable allow pathologists to detect things early and obtain insights that were not hitherto available.
The research has been published in the journal Kidney International Reports, with the paper titled “Association of Pathological Fibrosis With Renal Survival Using Deep Neural Networks.”
In related medical technology news, Royal Phillips and Hologic have entered into a partnership designed to develop technology that fits with the emerging precision (or personalized) medicine paradigm. See the Digital Journal article “Philips and Hologic provide imaging systems for women’s health.”