Table 2. Results of statistical tests comparing inter-grader IOU scores and IOU scores between CNN and graders.

Two-sided t tests were performed when data was approximately normally distributed. Non-parametric Man-Whitney U-tests were performed, when data was not normally distributed.

Group	Compartment	Test	p-value
L	Vitreous	Two-sided t test	1.151e-06*
L	Retina	Two-sided t test	1.95e-09*
L	Choroid	Man-Whitney U-test	0.02895
L	Sclera	Man-Whitney U-test	0.1701
EO	Vitreous	Two-sided t test	1.369e-06*
EO	Retina	Two-sided t test	1.219e-10*
EO	Choroid	Man-Whitney U-test	0.004588*
EO	Sclera	Man-Whitney U-test	0.02977
RC	Vitreous	Two-sided t test	0.000669*
RC	Retina	Two-sided t test	5.056e-07*
RC	Choroid	Man-Whitney U-test	0.65
RC	Sclera	Man-Whitney U-test	0.8525

Cross symbols * indicate statistically significant results at the 1% confidence level.