bwboundaries but returns pixel edges rather than pixel centers

Context

I have a set of point coordinates (2D) and a set of different regions of interest (ROI, binary image). I want to classify the point coordinates according to in which ROI they are located.

Variables:

• ROIimg: binary image with ROI's
• spots: n-by-2 matrix where each row hold the x- and y-coordinates of a spot

My first try was to use bwboundaries() and then for each point use inpolygon():

[boundaries,LabeledROI,num] = bwboundaries(ROIimg);
spotLabels=nan(size(spots,1),1);
for roiIdx=1:num
  foundcell=inpolygon(spots(:,1),spots(:,2),boundaries{roiIdx,1}(:,2),boundaries{roiIdx,1}(:,1));
  spotLabels(foundcell)=roiIdx;
end

Problem

This approach works for most spots, however, the boundaries that are produced by bwboundaries consist of the pixel centers of the boundary pixels, not of their edges/boundaries. Consequently, for spots that lie in a pixel on the edge of the ROI, but on the outer halve of the pixel, the spot will not be labeled as inside the boundary of that ROI.

Illustration

figure, hold on,
imagesc([0 127],[0 127],ROIimg)
axis image
for roiIdx=1:num
    plot(boundaries{roiIdx,1}(:,2)-1,boundaries{roiIdx,1}(:,1)-1)
end
plot(spots(:,1),spots(:,2),'+')

As a temporary workaround, I am using https://nl.mathworks.com/matlabcentral/fileexchange/21925-accurate-polygon-extension This is a better approximation, but it does not work for pixel corners.