Artificial Intelligence Networks Combining Histopathology and Machine Learning Can Extract Axon Pathology in Autism Spectrum Disorder.
Arash Yazdanbakhsh, Kim T M Dang, Kelvin Kuang and 4 others
PMID 41178535WHAT IT FOUND
An AI trained on brain tissue images from 12 autopsies identified autism with 90.88% accuracy on electron microscopy and 97.68% on light microscopy.
It failed to distinguish white matter depth. This is a proof-of-concept for research tools, not a clinical diagnostic test.
Key findings
01The network achieved high accuracy in distinguishing autism from controls, reaching 90.88% for electron microscopy images and 97.68% for optical microscopy images.
02Classification of white matter depth (superficial versus deep) was significantly less accurate, with 57.21% for electron microscopy and 69.13% for optical microscopy.
03The AI consistently misclassified deep white matter in autism as superficial, suggesting it learned superficial white matter features as the primary marker for the condition.
STILL TO COME
How it was doneWhat they found
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The sample size was extremely small (12 cases total), limiting generalizability. The study used postmortem tissue, so findings cannot be applied to living patients for diagnosis. The model failed to accurately distinguish between superficial and deep white matter layers. The study is a technical proof-of-concept for an AI tool, not a clinical trial of a treatment or diagnostic intervention. No clinical outcomes or patient behaviors were measured; the study focused solely on image classification accuracy.
Declared interests
The study was funded by the National Institutes of Health (specifically NIMH and NIA) and the Simons Foundation. The authors declare no competing interests.
The easy way to misread this
Do not interpret this as a diagnostic tool for living patients. The AI was trained on only 12 postmortem brains and cannot distinguish white matter depth reliably. It is a research prototype for analyzing tissue structure, not a clinical test.
Summarised by AI from the full paper, without a clinician reviewing it. Check it against the source before it changes what you do. Read it on PubMed →