Breast　cancer　is　the　most　deadly　cancer　in　females　globally.　Architectural　distortion　is　the　third　most　often　reported　irregularity　on　digital　mammograms　among　the　masses　and　microcalcification.　Physically　identifying　architectural　distortion　for　radiologists　is　problematic　because　of　its　subtle　appearance　on　the　dense　breast.　Automatic　early　identification　of　breast　cancer　using　computer　algorithms　from　a　mammogram　may　assist　doctors　in　eliminating　unwanted　biopsies.　This　research　presents　a　novel　diagnostic　method　to　identify　AD　ROIs　from　mammograms　using　computer　vision-based　depth-wise　CNN.　The　proposed　methodology　is　examined　on　private　PINUM　2885　and　public　DDSM　3568　images　and　achieved　a　0.99　and　0.95　sensitivity,　respectively.　The　experimental　findings　revealed　that　the　proposed　scheme　outperformed　SVM,　KNN,　and　previous　studies.　©　2022,　The　Author(s),　under　exclusive　license　to　Springer　Nature　Singapore　Pte　Ltd.