A　spare　reconstruction　algorithm　for　Cherenkov-excited　luminescence　scanned　tomography　(CELST)　is　developed　to　improve　quantitative　accuracy　when　luminescent　probes　locate　in　deeper　tissue.　The　proposed　algorithm　incorporates　the　sparse　a　priori　information　of　probes　into　the　reconstruction.　A　fast　iterative　shrinkage　threshold　algorithm　(FISTA)　and　3D　block　matching　(BM3D)　algorithm　are　jointly　employed　to　recover　the　distribution　of　luminescent.　Numerical　simulations　demonstrate　that　the　proposed　method　can　result　in　more　accurate　quantitative　reconstruction　and　increase　depth　of　detection　as　compared　with　conventional　Tikhonov　based　reconstruction　algorithm　used　in　CELST.