(57) A method for determining the presence or absence of a local and/or global property
(6) of a biological tissue sample (1), comprising:
• acquiring (100, 110, 120, 130) an image (2) of the sample (1);
• searching (200) the image (2) for the presence of each feature (30-39) that is contained
in a pre-defined set (3) of features (30-39), and assigning (250), to at least one
feature (30-39), a corresponding pronunciation index (40-49) that is a measure of
how pronounced each feature (30-39) is in the image or parts of it (2),
• computing (300) at least one discriminant value (5), e.g. a linear discriminant
value, LD, that is a function of the weighted pronunciation indices (40-49), wherein
the weight (50-59) of each pronunciation index (40-49) is a measure for the relevance
of the corresponding feature (30-39) with respect to the property (6), and
• determining (400) that the property (6) is present or absent in the sample (1) or
part of it depending on whether the discriminant value exceeds a pre-defined and/or
optimized threshold (7) or not.
|
|
