Additional material and experiments:
Test data Mammography:
learningImage=
,
,
,
,
,
;
Check image dimensions
In[]:=
ImageDimensions/@learningImage
Out[]=
{{583,914},{268,839},{396,836},{495,838},{750,783},{406,776}}
Check the first image properties
In[]:=
Information[First[learningImage]]
Out[]=
Image
Color Space
RGB
Image Dimensions
{583,914}
Image Resolution
{96.012,96.012}
Channels
4
Interleaving
True
Transparency
True
Data Type
Byte
Function to detect white spots, using image processing computation:
In[]:=
detectWhiteSpots[image_]:=Module[{threshold,binarized,components,spots,boundingBox},​​threshold=0.7;​​binarized=Binarize[image,threshold];​​components=MorphologicalComponents[binarized];​​spots=SelectComponents[components,"FilledCircularity",2];​​Image[spots]​​]
View original image and the mask created by the function, side by side, it works in most cases...
In[]:=
#->detectWhiteSpots[#]&/@learningImage//Column
Out[]=
Using morphological components to get a square around the white spot area, according the color brightness:
In[]:=
#->Image[MorphologicalComponents[Binarize[Blur[#,10],.7],Method->"BoundingBox"]]&/@learningImage//Column
Out[]=
Again, but now view the image mask, the mammography high white image spots, sometimes it is not what we are looking for:
Exploring function by function:
Another way to do it... we can play around the parameters threshold for refinement, the appropriate values depends at each image data:
Another version with basically the same functions and computation sequence:
New we can extract the centroid with component measurements:
A function to get a Rectangle from the centroid provided:
Dirt cleaning...
Combine for a better visualization with a Rectangle over the white spot area:
Now using the initial function "detectWhiteSpots":
There are false positives and false negatives, pointing to the wrong spot because of areas brightness.
To fix it define a calculation for appropriate threshold functions (parameters) for each image bright data...
We are not using AI or ML here, just image processing traditional matrix computation (pure linear algebra)

References

CBIS-DDSM: Breast Cancer Image Dataset
Curated Breast Imaging Subset DDSM Dataset (Mammography)
​https://www.kaggle.com/datasets/awsaf49/cbis-ddsm-breast-cancer-image-dataset​
5.3GB Zip
Mammogram Mastery: A Robust Dataset for Breast Cancer Detection and Medical Education
​https://data.mendeley.com/datasets/fvjhtskg93/1​
978GB Zip
Mammography Screening for Breast Cancer: Decoding Its Effectiveness and Safety
​https://integrative-cancer-care.org/mammography-screening-for-breast-cancer-decoding-its-effectiveness-and-safety/

CITE THIS NOTEBOOK

Image processing for automated detection of high-brightness spots in digital mammography exams​
by Daniel Carvalho​
Wolfram Community, STAFF PICKS, October 28, 2025
​https://community.wolfram.com/groups/-/m/t/3566959