The　degree　of　Ca2+　leakage　from　the　sarcoplasmic　reticulum　(SR)　in　cardiomyocytes　can　lead　to　dysfunction　in　both　contraction　and　relaxation,　potentially　triggering　arrhythmias.　Therefore,　it　is　crucial　to　develop　simple　and　effective　methods　for　quantitatively　detectinga　Ca2+　leakage.　In　this　study,　we　employed　an　asymmetric　Gaussian　function　to　accurately　fit　the　signals　in　confocal　scanning　Ca2+　signal　images,　facilitating　the　separation　of　noise　from　the　calcium　signals.　We　introduced　two　new　feature　parameters,　Rca2+leak　and　Pca2+leak,　for　the　quantitative　assessment　of　calcium　leakage.　Given　the　subtle　differences　in　calcium　signal　characteristics,　visually　distinguishing　the　states　of　cardiomyocytes　can　be　challenging.　To　address　this,　we　applied　machine　learning　techniques　in　conjunction　with　the　newly　developed　parameters　to　classify　the　states　of　cardiomyocytes,　thereby　inferring　their　pathological　conditions.　Our　approach　utilizes　a　straightforward　and　efficient　methodology　to　propose　two　new　parameters　that　effectively　quantify　the　degree　of　calcium　leakage.　By　integrating　these　new　parameters,　we　achieved　a　classification　accuracy　for　pathological　states　of　cardiomyocytes　that　is　2%　to　6%　higher　than　that　obtained　with　traditional　feature　classification　methods.　©　2024　Copyright　held　by　the　owner/author(s).