Skip to contents

Ask a Natural-Language Question About a DICOM File

Source: R/llm.R
dicom_ask.Rd

Answers arbitrary questions about a DICOM file's metadata or image content using the configured LLM (or local metadata inspection).

Usage

dicom_ask(path, question, llm = dicom_llm("ollama"), include_image = FALSE)

Arguments

path

Path to a DICOM file.

question

Character string question (e.g., "What is the pixel spacing and modality?").

llm

A "dicom_llm" object created by dicom_llm().

include_image

Logical. Attach the rendered DICOM image for multimodal vision models. Default is FALSE.

Value

Character string answer.

Examples

dcm_file <- system.file("extdata", "example.dcm", package = "dicomtk")
dicom_ask(dcm_file, "What is the modality and image dimensions?", llm = dicom_llm("offline"))
#> [1] "Image Type ((0008,0008)): ORIGINAL\\PRIMARY\\AXIAL; Modality ((0008,0060)): CT; Image Position (Patient) ((0020,0032)): -72.199997\\-143.000000\\-101.237503; Image Orientation (Patient) ((0020,0037)): 1.000000\\0.000000\\0.000000\\0.000000\\1.000000\\0.000000"