Part 3 · Diagnosis
McDonald criteria calculator
Place MS-type lesions on a schematic brain, spinal cord and optic nerves, add CSF and other results, then watch the 2017 or 2024 McDonald criteria build a verdict step by step.
A learning aid, not a diagnosis. It applies the published rules to the boxes you tick. It cannot diagnose anyone. The last step, "no better explanation", always needs a clinician's judgement after a full workup, so a positive result here reads "criteria met (if no better explanation)". Rules: Thompson et al., Lancet Neurol 2018;17(2):162-173 462 and Montalban et al., Lancet Neurol 2025;24(10):850-865 325. The 2024 paper was paywalled to our research, so its Figures 2 and 3 were read from the official IACCTMS slide deck that reproduces them.
Schematic composite, not to scale and not a real scan. Abbreviations in the steps: PV periventricular, JC cortical/juxtacortical, IT infratentorial, SC spinal cord, ON optic nerve, WM white matter. Deep white-matter lesions are drawn because many people have them, but they are not a DIS site.
Everything else the criteria ask about
κ-FLC index = (CSF κ-FLC ÷ serum κ-FLC) ÷ (CSF albumin ÷ serum albumin). Positive CSF in 2024 means OCB or an index ≥6.1; each laboratory should verify that cut-off 114,254. The 2017 criteria use OCB only 462.
With rigorous quality control, an inter-eye difference of ≥6 μm in peripapillary RNFL or ≥4 μm in GCIPL supports optic nerve involvement; an MRI optic nerve lesion also counts (place it on the map). Any of these needs no better explanation for the optic nerve finding 218,416.
Set "with central vein sign" and "with paramagnetic rim" per site in the lesion details above. Select-6: at least 6 white-matter lesions with a central vein or, if fewer than 10 white-matter lesions are seen, more CVS-positive than CVS-negative lesions. PRL: at least 1 paramagnetic rim lesion 325,418,443. The official slide deck is internally inconsistent here: one slide applies the majority rule below 6 lesions and another below 10. The published figure says fewer than 10, which this calculator uses 218.
"No better explanation" means the presentation is typical of MS and nothing clinical or paraclinical suggests another diagnosis, or the workup for those diagnoses is negative 443. A calculator cannot decide this. See The differential.
The same case under 2017 and 2024
Both rule sets run on every change. The highlighted card is the version the map is showing. Apart from some people with primary progressive MS who lack several spinal cord lesions, everyone who meets 2017 also meets 2024, and a diagnosis made under older criteria is not changed by the revision 443.
2024 dissemination in time follows the paper's Figure 3 footnote: "one or more new T2 lesions or one or more gadolinium-positive lesions or a clinical attack" 325. The companion glossary reads it more strictly, as new T2 lesions over time or enhancing and non-enhancing lesions together on one scan, as in 2017 443. The sources do not settle this; the verdict tells you when the stricter reading would change the answer.
How well the criteria and tests perform
Where experts disagree
Diagnosing without dissemination in time
2024 allows a diagnosis on 4–5 sites alone, or with a central vein sign, to speed diagnosis 325,477. Real-world cohorts found many more diagnoses, specificity of 57.8% against 2017 at follow-up, and low 2017–2024 agreement in one cohort (κ=0.112) 54,178. The London authors note that their "specificity" counts 2017-negative CIS and RIS as false positives. No long-term prospective validation yet.
Calling RIS multiple sclerosis
About 51% of people with RIS develop clinical symptoms within 10 years 218,407, and OCB (HR 5.09) and κ-FLC ≥6.1 (HR 2.79) predict it 148. Critics point to overdiagnosis, the consequences of treatment and labelling, and gaps in implementation 54,254,418. Letters raising concerns were published with an authors' reply 326.
The optic nerve as a fifth site
Adding it raised sensitivity without losing specificity by VEP in one cohort 481 and gave similar accuracy across MRI, OCT and VEP in another 482. Other cohorts found lower specificity, and asymptomatic optic nerve findings in non-optic-neuritis CIS added no patients 53,139. OCT and VEP thresholds depend on quality control, centre and device 416.
Central vein sign and κ-FLC cut-offs
Earlier work used a 40% lesion rule; Select-3 was sensitive but not specific, and 2024 adopted Select-6 63,325,467. Published κ-FLC cut-offs range from 2.4 to 20; 6.1 is a weighted mean, and κ-FLC and OCB disagree in about 10%–13% of cases 114,218.