Sunday, November 26 | 1:50 p.m.-2:00 p.m. | S4-SSNR02-6 | Room E352
A systematic review has found that machine-learning and deep-learning techniques are highly sensitive and specific for detecting Parkinson’s disease on dopamine transporter (DaT) SPECT exams.
In an effort to assess the current state of AI techniques for diagnosis of Parkinson’s disease on DaT SPECT exams, the researchers led by Faranak Ebrahimian Sadabad, MD, of Yale University performed a systematic literature review that encompassed 27 studies and nearly 13,000 Parkinson’s disease patients and healthy controls.
Of these, 15 studies used traditional machine-learning techniques, 10 used deep-learning algorithms, and two combined transfer-learning methods with deep learning, according to the group.
The machine-learning and deep-learning algorithms yielded median sensitivity of 96.4%, median specificity of 95%, median accuracy of 95.2%, and a median area under the curve of 0.98.
“Overall, [machine learning] and [deep learning] have high sensitivity and specificity in the detection of Parkinson’s disease by DaT scans,” the researchers wrote.
What else did their analysis reveal? Take in this talk on Sunday afternoon to learn more.