WHY TONGUE PHOTOS ARE UNRELIABLE FOR SELF-ASSESSMENT
Original reading record; not a clinical assessment or calibration method.
No photograph, medical history or questionnaire answers need be uploaded to this website.

MY PURPOSE OR QUESTION:
CAPTURE FACTS (record only what is known)
Date / known or uncertain:
Device / known or unknown:
Lighting and any difference from another photo:
Focus, framing and visible area:
Original or edited / known edits:
What is unknown:

IMAGE OBSERVATION IN ORDINARY WORDS:
Do not convert an appearance into a symptom cause or constitution diagnosis.

CLAIM AND METHOD CHECK
App or article and exact claimed output:
Tested task: quality / segmentation / feature label / clinical outcome / unknown
Reference labels or independently assessed outcome:
Study population and capture setting:
Does the studied task match the claimed use?
Study limitations and missing information:

NEXT QUESTION
If a professional requested the photo, what instructions did they give?
What relevant concern needs to be described separately from the image?
What needs an appropriate assessment rather than more photos?

SIX FICTIONAL CASES
Case 1: Your second photograph looks paler after you change from a dim room to a brighter lamp. Does that establish that your constitution changed?
Your answer and reason: ______________________

Case 2: The earlier photograph shows fine lines, but the later one is blurred. Can you conclude the lines disappeared?
Your answer and reason: ______________________

Case 3: An app cites a paper on fissured-tongue image recognition. May it describe that paper’s classification performance as the accuracy of your body-type result?
Your answer and reason: ______________________

Case 4: You use the same camera and room for a week. Does a more consistent setup make the photos sufficient to select a herbal product?
Your answer and reason: ______________________

Case 5: A clinician requests an original photo through an agreed channel. Should you make it look clearer with a colour filter before sending it?
Your answer and reason: ______________________

Case 6: A photo app calls the image normal, but you still have a health concern. Is the app result a reason to delay appropriate assessment?
Your answer and reason: ______________________

ANSWERS AND EXPLANATIONS
Case 1: No. The capture conditions changed, so the images alone cannot establish that conclusion.
Record the different lighting if the photographs are relevant to an appropriate discussion. The 2022 quality study explicitly considered excessive brightness and darkness as image-quality problems. It did not establish a rule that a paler home photo diagnoses a constitution or disease.

Case 2: No. Lost image detail is different from an absent physical feature.
A blurred photograph can conceal texture. Do not treat the less detailed image as evidence that a product worked, or keep retaking pictures until you obtain the appearance you want.

Case 3: No. The cited task and the app’s claimed result would have to match.
The 2026 system study tested segmentation and recognition of fissured-tongue features. Its authors describe preliminary imaging-feature recognition, not independent clinical diagnosis. The label, reference standard, setting and intended use must stay beside any performance claim.

Case 4: No. Reducing capture variation does not establish a diagnosis or a suitable treatment.
Consistency addresses part of the recording process. It does not validate a constitution category, establish a symptom cause or identify product suitability. NCCIH advises discussing complementary approaches with healthcare providers and avoiding replacement or delay of conventional care.

Case 5: Follow the clinician’s instructions and preserve the original; do not silently edit its appearance.
An edited image and an original capture are different records. Ask whether another photo or specific capture information is needed. A filter changes what is shown without supplying missing clinical context or validating the interpretation.

Case 6: No. An image label does not clear a person of a health concern.
Describe the actual concern and its timing to an appropriate health professional. This guide does not assess urgency, diagnose symptoms or provide a safe waiting period. A reassuring picture or constitution category is not a substitute for care.

SOURCE REFERENCES
Xian et al. (2022): Automatic tongue image quality assessment using a multi-task deep learning model
https://pmc.ncbi.nlm.nih.gov/articles/PMC9531121/
Wei et al. (2026): A portable non-contact tongue imaging system with automated analysis for community and home settings
https://pmc.ncbi.nlm.nih.gov/articles/PMC13305718/
NCCIH: Traditional Chinese Medicine — What You Need To Know
https://www.nccih.nih.gov/health/traditional-chinese-medicine-what-you-need-to-know