Create conceptual anatomy and disease-mechanism illustrations from descriptions with AI, then convert to editable SVG and edit any label, structure, or layout.
SciFig Team
SciFig editorial team
You need an explanatory anatomy or disease-mechanism figure for a paper. Begin by separating conceptual artwork from clinical evidence: the original patient or experimental image must retain its provenance, while a schematic explains verified information. When choosing an AI route, the target journal’s rules are a useful reference.
AI medical illustration can create conceptual anatomy and disease-mechanism artwork from a description and suitable reference material. This guide covers grounding the figure in your sources, editing it element by element as SVG, and the publication rules you can use as a reference.
Conceptual medical illustration workflow mockup (Figure generated with SciFig)
Transparency note: Illustrations in this article were generated with SciFig AI. Cited claims link to peer-reviewed sources.
Build on labeled cell examples
Compare Nano Banana 2.1 and GPT Image 2.5 for scientific figures, from labels and layouts to focused edits. The comparison guide includes a short prompt structure and selected scientific image examples; the Flare/Sunburst comparison helps you choose a version.
What Medical Illustration Looks Like in Modern Research
Medical illustration is the visual translation of clinical, anatomical, or pathological content into figures appropriate for journal publication, education, and patient communication. Three categories dominate research-paper use: anatomical diagrams (reproductive system, digestive tract, skeletal structures), pathology illustrations (disease mechanisms, inflammation cascades, immune responses), and surgical/procedural schematics (technique overviews, equipment positioning, intervention pathways).
The volume across journals is significant. A 2022 Council of Science Editors audit of clinical journals found that 64% of original research articles included at least one medical illustration beyond data plots and clinical photos. The figures are workhorses — they establish context that text cannot, particularly for readers outside the specific subspecialty. And until recently, producing them required a medical illustrator (Master's-level training, $80–$120 per hour) or careful manual work in Adobe Illustrator with a medical-illustration plugin library.
Three medical illustration types: anatomy, pathology, surgical (Figure generated with SciFig)
SciFig AI Illustration (AI scientific illustration) creates mechanism and concept images from descriptions, sketches, or reference material, and you can convert any image to editable SVG to adjust every element. Use SciFig FlowChart (flowcharts and diagrams) for workflows output as native editable SVG and SciFig DataChart (data-driven charts) for charts generated precisely from your original table.
Why AI Medical Illustration Is Different (And When to Be Careful)
Generic image models (DALL·E, Midjourney, Stable Diffusion) produce plausible-looking medical illustrations with subtle anatomical errors: a heart with the aorta arching the wrong direction, a kidney with the nephron loop on the cortical instead of medullary side, a brain with the cerebellum disproportionately large. The errors are visually undetectable to non-experts and trivially obvious to a reviewer who teaches the relevant anatomy.
SciFig AI Illustration creates illustrations from your instructions and reference material: name the structures, counts, orientations, and labels, and the figure follows them. Every result converts to editable SVG, so any structure or label can be adjusted directly.
Heart anatomy comparison: generic AI vs SciFig reference grounding (Figure generated with SciFig)
Every SciFig figure is editable, so you stay in control of anatomical detail: compare it with your preferred reference (Gray's Anatomy, Netter's Atlas, or prior figures in your target journal) and adjust any structure, label, or orientation directly in the vector canvas.
See Photo-to-Figure Conversion in Action
Use photographs of equipment or a laboratory setup to guide a conceptual schematic.
Step 1: Define a Conceptual Line Drawing from Verified Sources
Choose AI Illustration in SciFig and describe the anatomy or mechanism using your source information. Use reference artwork only when you have the right to do so.
For an educational skin-anatomy schematic, an example prompt:
"Create an educational skin cross-section schematic. Show epidermis and dermis with clearly separated layers and leader-line labels. Use a medical-textbook line style on a white background. Do not depict a real patient, infer a diagnosis, or add measured lesion features."
The output is fully editable: convert it to SVG to adjust any structure or label against your source, mark illustrative content in the caption, and use the target venue’s rules as a reference. Consent and disclosure do not make AI alteration of clinical evidence permissible, and AI rendering is not a reliable anonymization method.
AI-generated conceptual mockup of a reference-to-drawing sequence (Figure generated with SciFig)
For pathology or endoscopy topics, explain the anatomy and process in a conceptual schematic derived from verified information.
Step 2: Generate Anatomical Diagrams from Text Prompts
For anatomy figures without a source photo, the medical illustration generator generates from descriptive prompts. Name the entity, view, and labels you want; the ready-to-use prompts below cover four common needs, and you can edit any element of the result afterwards.
Working prompts across four common anatomical needs:
Anatomical figure
Working prompt
Reproductive system (female)
"Female reproductive system anatomical illustration, anterior view, labeled: uterus, fallopian tubes, ovaries, cervix, vagina. Medical textbook style, soft pastel colors, leader-line labels, white background."
Digestive tract
"Human digestive tract anatomical diagram, anterior view, complete from oral cavity to rectum, labeled: esophagus, stomach, small intestine (duodenum, jejunum, ileum), large intestine (cecum, ascending/transverse/descending colon, sigmoid), rectum. Medical textbook style."
For each, the prompt structure is consistent: entity, view, labels (enumerated), style anchor, background. The framework matches the one used for cellular diagrams (see the animal cell tutorial for E-S-S-V details).
Step 3: Create Pathology Illustrations and Disease Mechanism Diagrams
Pathology illustrations show how disease processes work at the molecular or cellular level. Spell out the compartments and the order of events (which protein activates which, in which compartment, in which order), and the figure follows your mechanism.
For a JAK/STAT signaling cascade — common in inflammation, oncology, and immune-mediated disease research — the working prompt is:
"JAK/STAT signaling pathway molecular illustration. Show: cytokine binding to cell surface receptor (extracellular), JAK phosphorylation (membrane-proximal), STAT recruitment and dimerization (cytoplasm), dimer translocation through nuclear pore, STAT binding to GAS element on DNA (nucleus). Arrows showing direction of signal flow. Label all key proteins. Molecular biology textbook style, layered cell compartments visible."
The output captures the canonical mechanism with correct directionality — STAT phosphorylation precedes dimerization, dimers translocate to nucleus, gene transcription follows. This is where generic image models fail most often: they routinely reverse the order (STAT first translocates, then dimerizes), or omit the GAS element, or label JAK as occupying the wrong subcellular compartment.
JAK/STAT signaling pathway molecular illustration with correct phosphorylation and dimerization order (Figure generated with SciFig)
For other common pathology figures — cancer cell metastasis cascades, immune synapse formation, complement activation, inflammasome assembly — the same prompt structure applies with adjusted entities and compartments.
Convert Lab Photos to Scientific Figures
Use photographs of equipment or a laboratory setup to guide a conceptual schematic.
Step 4: Edit and Add Custom Labels in Vector Canvas
To match your specific layout, edit any label directly. The vector canvas tool opens the figure as a layered SVG: organelles and structures on one layer, labels on a separate layer, leader lines on a third. This lets you reposition labels without rebuilding the underlying anatomy.
Common edits: font size to journal compliance (most journals require 7–8 pt minimum); label background contrast (semi-transparent backgrounds when labels overlay colored anatomy); arrow direction adjustments (signal flow that needs to read left-to-right for your specific layout); terminology localization (changing "renal corpuscle" to "glomerulus" for a clinical audience).
Vector canvas: editable kidney illustration with labels (Figure generated with SciFig)
Label editing takes 2–5 minutes per figure. For figures going into multi-panel compositions (see how to lay these out), the label refinement matters most — alignment across panels is the difference between a clean figure and a flagged one.
Medical Illustration Ethics: Disclosure, Consent, and AI in 2026
The clinical journal’s current policy for the particular figure is the reference when using AI-generated artwork. Some venues restrict it or require editorial permission. Where the workflow is allowed, record the actual tool, model or version, date, and author verification, then disclose these where the journal requests them.
Tool used — "Figures were generated with SciFig (https://scifig.ai)"
Model/version — "using [the model and version used]"
Generation date — "on [the actual generation date]"
Preparation steps — "The authors used [SciFig AI Illustration image generation, SciFig FlowChart AI SVG generation, or SciFig DataChart AI planning followed by Python plotting] and made [actual edits]"
Annotated Methods section AI use disclosure callout: tool, model version, date, human review (Figure generated with SciFig)
For original clinical photographs, document consent and ethics approval where required, preserve the original acquisition, and follow the target journal’s privacy and image-processing instructions. For a separately generated conceptual diagram, you can describe the actual method, using the venue’s policy as a reference.
AI medical illustration creates a new anatomy or mechanism image from a description or reference material. Stock medical images are existing artwork selected from a catalog. Generation can tailor the anatomy and mechanism artwork to your explanation. Keep primary clinical evidence and its provenance as source records.
The answer depends on the journal and figure type. JAMA restricts AI-created or AI-altered clinical images except where they form part of the formally described research method. Keep primary clinical evidence unaltered; the journal's current instructions are the reference for using a conceptual illustration. See the journal AI-figure policy guide.
With SciFig AI Illustration, a structured prompt (anatomical entity, view, labels, and visual style) produces a labeled figure, and because it converts to editable SVG you can compare it with an atlas or an expert's notes and adjust any structure or label directly.
That depends on patient consent, ethics approval, the privacy rules of your institution, and the target journal’s policy. Document patient consent and ethics approval where required, and follow the target journal’s disclosure and image-processing instructions.
SciFig AI Illustration follows the mechanism you describe: name the compartments and the order of events (for JAK/STAT, receptor binding, JAK phosphorylation, STAT dimerization, nuclear translocation), and the figure places each step accordingly. Every protein, arrow, and label converts to editable SVG, so you can adjust any of them directly.
SciFig AI Illustration gives you a conceptual medical figure quickly that you can convert to editable SVG and adjust element by element. A professional medical illustrator may still suit figures that need specialized anatomical judgment, detailed surgical explanation, or a tightly specified house style.
The target journal’s instructions are the reference for the permitted workflow; a disclosure typically records the tool (for example, SciFig), the model or version, the date, and the edits you made, in the location the journal requests. See the journal AI-figure policy guide.
Medical Disclaimer
This article is educational content focused on scientific figure design for research publications. It is not medical advice and should not be used for clinical decisions. The disease mechanisms, anatomical references, and ethical guidance described here are summarized from cited peer-reviewed sources; for clinical practice, consult primary literature, official guidelines, and licensed clinicians. SciFig is a scientific figure tool — it does not diagnose, treat, or advise on patient care.
Keep going
Recommended resources
Templates, tools, and further reading for your research