Getting Started with dicomtk

Overview

dicomtk is a self-contained, tidyverse-compliant R interface to the OFFIS DICOM Toolkit (DCMTK 3.7.1) with native C++17 / Rcpp bindings, ggplot2 / Shiny / 3D volume visualization, PACS networking, and local/cloud LLM integration (Ollama, OpenAI, Anthropic Claude).

library(dicomtk)
library(dplyr)

# Verify bundled DCMTK 3.7.1 engine & dictionary
dicom_info()
#> # A tibble: 1 × 5
#>   dcmtk_version dictionary_standard dictionary_entries backend codecs_registered
#>   <chr>         <chr>                            <int> <chr>   <lgl>            
#> 1 3.7.1         PS 3.6-2026d / PS …               5277 Bundle… TRUE

Reading DICOM Headers & Tidyverse Workflows

dcm_file <- system.file("extdata", "example.dcm", package = "dicomtk")

# Long tidy tibble of DICOM elements
hdr <- read_dicom(dcm_file)
hdr |>
  filter(group %in% c("0008", "0028")) |>
  select(tag, vr, keyword, name, value) |>
  head(10)
#> # A tibble: 10 × 5
#>    tag         vr    keyword              name                   value          
#>    <chr>       <chr> <chr>                <chr>                  <chr>          
#>  1 (0008,0005) CS    SpecificCharacterSet Specific Character Set "ISO_IR 100"   
#>  2 (0008,0008) CS    ImageType            Image Type             "ORIGINAL\\PRI…
#>  3 (0008,0012) DA    InstanceCreationDate Instance Creation Date "20010101"     
#>  4 (0008,0013) TM    InstanceCreationTime Instance Creation Time "002816"       
#>  5 (0008,0014) UI    InstanceCreatorUID   Instance Creator UID   "1.3.6.1.4.1.5…
#>  6 (0008,0016) UI    SOPClassUID          SOP Class UID          "1.2.840.10008…
#>  7 (0008,0018) UI    SOPInstanceUID       SOP Instance UID       "1.3.6.1.4.1.5…
#>  8 (0008,0020) DA    StudyDate            Study Date             "20010101"     
#>  9 (0008,0021) DA    SeriesDate           Series Date            "20010101"     
#> 10 (0008,0022) DA    AcquisitionDate      Acquisition Date       "20010101"

# Wide format (1 row per file, columns named by DICOM keyword)
read_dicom(dcm_file, wide = TRUE) |>
  select(any_of(c("file", "Modality", "Rows", "Columns", "BitsAllocated")))
#> # A tibble: 1 × 5
#>   file                                      Modality Rows  Columns BitsAllocated
#>   <chr>                                     <chr>    <chr> <chr>   <chr>        
#> 1 /tmp/RtmpJkUnqX/Rinsteb77d2f93ac/dicomtk… CT       512   512     16

Visualizing DICOM Images (`ggplot2`, Clinical Windows, and 3D MPR/MIP)

plot_dicom(dcm_file, window = "auto", palette = "gray", title = "Example DICOM Slice")

Anonymization & Tag Modification

anon_path <- dicom_anonymize(
  dcm_file,
  patient_name = "STUDY^SUBJECT01",
  patient_id = "SUBJ001",
  shift_days = -365
)
dicom_tag_value(anon_path, c("PatientName", "PatientID", "PatientIdentityRemoved"))
#>            PatientName              PatientID PatientIdentityRemoved 
#>      "STUDY^SUBJECT01"              "SUBJ001"                  "YES"

AI / LLM Integration (Ollama, OpenAI, and Anthropic Claude)

# Configure local Ollama or OpenAI/Anthropic provider
# ollama_llm <- dicom_llm("ollama", model = "llava:13b")
# openai_llm <- dicom_llm("openai", model = "gpt-4o")
# claude_llm <- dicom_llm("anthropic", model = "claude-sonnet-4-20250514")

# Deterministic local mode for offline reproducibility
local_ai <- dicom_llm("offline")
dicom_summarize(dcm_file, llm = local_ai)
#> [1] "DICOM Study Summary: CT imaging study (General Study) acquired on GE MEDICAL SYSTEMS system. Series: SmartScore - Gated 0.5 sec (Body Part: Not specified). Image Matrix: 512 x 512 pixels (16-bit MONOCHROME2)."
dicom_report(dcm_file, llm = local_ai)
#> # A tibble: 4 × 2
#>   section     content                                                           
#>   <chr>       <chr>                                                             
#> 1 EXAMINATION CT -  - SmartScore - Gated 0.5 sec                                
#> 2 TECHNIQUE   DICOM Study Summary: CT imaging study (General Study) acquired on…
#> 3 FINDINGS    DICOM Image Analysis (CT, 512x512, Photometric: MONOCHROME2): Res…
#> 4 IMPRESSION  Automated quantitative and metadata review completed. Correlate c…