About CogniScan

Dataset, methodology, and clinical context

Overview

CogniScan predicts dementia risk from neuroimaging and cognitive assessment variables using a Random Forest classifier trained on the OASIS Longitudinal MRI Dataset. Predictions are explained using SHAP values.

Not a medical device. Portfolio demonstration only.

Feature Glossary

MMSEMini-Mental State Examination (0-30). Scores below 24 suggest possible impairment.

nWBVNormalized Whole Brain Volume. Decreases with neurodegeneration.

eTIVEstimated Total Intracranial Volume. Used for brain size normalization.

ASFAtlas Scaling Factor. MRI-derived scalar for volume normalization.

EDUCYears of formal education. Higher education = greater cognitive reserve.

SESSocioeconomic status (Hollingshead Index, 1=highest to 5=lowest).

Tech Stack

Python 3.11TensorFlow 2.15CNN (Keras)FastAPIPillowNext.js 14Tailwind CSSDocker