High-resolution　medical　images　contain　more　detailed　pathological　information　than　low-resolution　medical　images,　and　clarity　of　medical　images　is　critical　for　doctors　in　the　diagnosis　of　disease.　Nevertheless,　previous　deep　learning-based　methods　are　deficient　in　terms　of　capturing　high-frequency　details　and　retaining　edge　information,　so　the　present　paper　puts　forth　a　distillation　network　based　on　the　enhancement　of　frequency　and　spatial　features　as　a　means　of　achieving　super-resolution　reconstruction　of　medical　CT　images.　Specifically,　we　propose　a　distillation　module　for　spatial-frequency　domain　feature　enhancement.　This　module　combines　the　Fast　Fourier　Transform　(FFT)　to　extract　frequency　information　and　utilizes　edge　operators　to　obtain　spatial　information,　which　enables　the　effective　extraction　of　textures　and　details.　Moreover,　it　reduces　the　large　number　of　parameters　brought　by　FFT　through　distillation.　In　addition,　in　order　to　expand　the　receptive　field　of　the　model,　a　spatial　attention　mechanism　module　based　on　large　kernels　is　designed,　which　enables　the　model　to　focus　more　effectively　on　relevant　spatial　regions,　thereby　enhancing　the　extraction　and　utilization　of　spatial　features.　The　experiment　shows　that　the　reconstructed　images　of　the　proposed　method　are　superior　to　the　comparison　algorithms　in　both　objective　evaluation　metrics　and　subjective　perception,　and　the　effect　is　even　better　when　the　scaling　factor　is　large.　©　2025　Elsevier　Inc.