Neurodegeneration and digital pathology: Uniting human and artificial intelligence for better outcomes
Finding an easily accessible and reliable tool to diagnose neurodegenerative diseases collectively defined as 'synucleinopathies' is an urgent unmet priority. The synucleinopathies include Parkinson's disease, multiple system atrophy, pure autonomic failure and dementia with Lewy bodies, all progressive disorders leading to disability and death. Novel diagnostic advances involving skin-based pathology have revolutionized the clinical management of neurologic disease. Immunostaining for multiple proteins using immunofluorescent immunohistochemistry has resulted in unparalleled diagnostic sensitivity and specificity for these challenging neurologic disorders. However, interpretation of specimens remains a challenge for pathologists attempting to find proverbial 'needles in the haystack', resulting in low volume outputs, high expense and long test turn-around times.
We have initiated a program for whole slide digitization combined with artificial intelligence (AI) assisted detection of pathologic proteins. The application output provides heatmap assisted visualization of pathology with quantitative analysis of abnormal proteins by tissue region. Our initial study has assessed the impacts of AI assisted pathologic interpretation on cost, efficiency, sensitivity and specificity of testing. In a blinded parallel interpretation of samples we report equal or superior sensitivity and specificity, a 40% reduction in pathologic reading time, a 15% reduction in turn-around time, a decrease in cost of goods sold by 28% and increased pathologist job satisfaction. We anticipate further improvement in outcome metrics as familiarity with digital pathology increases.
We conclude that digital pathology with AI assisted pathologic pre-reading will dramatically alter the field of neurodegenerative disease and bring novel diagnostic advances into mainstream medical care.
Presented by:
Christopher Gibbons, MD, MMSc
Associate Professor of Neurology
Harvard Medical School
Christopher Gibbons is an Associate Professor of Neurology at Harvard Medical School. He is former Chair for the Autonomic Section of the American Academy of Neurology, Past-President of the American Autonomic Society and Past-Chair of the International Diabetes Neuropathy Consortium.
Dr. Gibbons is actively engaged in clinical/translational research. His current areas of research include: 1) developing novel biomarkers for neurodegenerative disorders and 2) defining the mechanisms and pathophysiology underlying the development of neuropathy. Dr. Gibbons has received research funding through the NIH (NINDS, NIA), the Goldberg Foundation, the Bion Family Foundation and the Juvenile Diabetes Research Foundation.