Researchers tested the probe on prostate biopsy tissue from 152 men. It read 3,342 light signatures
from 836 samples. The software classified each signature as cancer or clear. A single model caught
80 cancers in every 100. Applying 5 filters in sequence raised that to 90. When the probe called a
sample clear, it was right 95 times in 100. The system prioritised catching cancer while clearing
62 in 100 samples without cancer. That cuts the samples needed for diagnosis by almost half.
Every reading was taken on tissue already removed from the patient.
An earlier study tested the same approach on 84 men. The probe read 2,395 light signatures from
599 biopsy samples. It found clinically significant cancer in 90 of every 100 cases. It correctly
cleared about 80 in 100 samples without significant cancer. It also graded the cancers it found.
It separated the more aggressive cancers from the less aggressive ones, correctly 90 times in 100.
These readings were taken at routine biopsy appointments, on tissue just removed.
