DICOM metadata viewer for Neovim 0.12+. Displays .dcm file headers as an expandable table with cursor-row highlighting.

• Neovim ≥ 0.12
• Python 3 with pydicom (uv tool install pydicom)

vim.pack.add({
    "Gabsha/dcm-nvim",
    config = function()
        require("dcm").setup({
            tag_width = 14,
            name_width = 32,
            auto_open = true,
        })
    end,
})

{
    "Gabsha/dcm-nvim",
    ft = "dcm",
    config = function()
        require("dcm").setup({
            auto_open = true,
        })
    end,
}

Open any .dcm file — the raw/binary buffer is replaced in-place with the rendered metadata table (no binary content is ever shown). If you invoke :DcmOpen <file> for a file that isn't the current buffer, the viewer opens in a right-side split instead:

nvim scan.dcm

┌──────────────┬──────────────────────────────────┬────────────────────┐
│ Tag          │ Name (VR)                        │ Value              │
├──────────────┼──────────────────────────────────┼────────────────────┤
│ (0008,0008)  │ ImageType (CS)                   │ ["ORIGINAL", ...]  │
│ (0010,0010)  │ PatientName (PN)                 │ Doe^John           │
│ (0010,0030)  │ PatientBirthDate (DA)            │ 19700101           │
│ (0020,000E)  │ StudyInstanceUID (UI)            │ 1.2.3.4.5.6...     │
│ ...          │ ...                              │ ...                │
└──────────────┴──────────────────────────────────┴────────────────────┘

Rows with ▶ can be expanded with <CR>. Expanded rows show ▼ and list all contained values. This includes DICOM Sequences (VR SQ), which are nested datasets: expanding a sequence reveals its [n] Item rows, and expanding an Item reveals its own tags -- indented one level further. Since sequences can nest arbitrarily deep (an Item can itself contain another Sequence), this works recursively to any depth, and every row can be expanded/collapsed independently of its siblings and parents.

require("dcm").setup({
    tag_width    = 14,      -- width of the (xxxx,xxxx) column
    name_width   = 32,      -- width of Name (VR) column
    auto_open    = true,    -- auto-open on FileType=dcm
    expand_key   = "<CR>",  -- key to expand/collapse
    close_key    = "q",     -- key to close viewer
})

vim.api.nvim_set_hl(0, "DcmCurrentRow", { bg = "#3a3a5c", bold = true })
vim.api.nvim_set_hl(0, "DcmHeader",     { fg = "#c0caf5", bold = true })
vim.api.nvim_set_hl(0, "DcmExpandable", { underline = true, sp = "#7aa2f7" })
vim.api.nvim_set_hl(0, "DcmChild",      { fg = "#9aa5ce" })

Available groups: DcmHeader, DcmSeparator, DcmFooter, DcmRow, DcmExpandable, DcmExpanded, DcmChild, DcmCurrentRow.

1. ftdetect/dcm.lua sets filetype=dcm for .dcm files
2. An autocmd fires on FileType dcm and calls pydicom via vim.system()
3. Pydicom parses the DICOM header and returns JSON
4. The JSON is rendered into a scratch buffer as a scrollable table
5. Rows with list values, or DICOM Sequences (nested datasets), are expandable via <CR> -- sequences recurse to arbitrary depth
6. Cursor movement triggers row highlighting via extmarks

MIT