You can get a binary "map" of where the image is lower than some amount, like
darkParts = grayImage < 0.005;
but then we need to know exactly what you want to know. What shape descriptors would be useful to you? The area? The perimeter? The circularity = Perimeter^2/(4*pi*area)?
Perhaps this would be useful for you: http://www.mathworks.com/matlabcentral/fileexchange/35452-finddepressions