Medical devices & diagnostics · Brain imaging AI
Darmiyan
BrainSee reads a routine brain MRI and two cognitive test scores and returns a number between 0 and 100. In January 2024 the FDA wrote a new device category around it. As of September 2026 nothing else has entered that category.
A number between 0 and 100
A woman in her seventies keeps losing names. Her doctor orders the brain MRI any memory workup already involves, a standard T1-weighted scan at 1.5 or 3 tesla, with no contrast and no radiation. Two short bedside tests are scored: the Mini-Mental State Examination and the Clinical Dementia Rating sum of boxes.
The scan files and the scores go up to a web portal, along with her age and her sex. The same day, a number comes back.
Below 50, her inputs look more like patients who stayed at mild cognitive impairment for five years. Above 50, more like patients who progressed to Alzheimer’s dementia. Near 50, like both, and the FDA required the labeling to say so.
That is BrainSee. No lumbar puncture, no radioactive tracer, no new hardware. Darmiyan calls the method a virtual microscope, because it grades the space between brain cells rather than hunting one molecule.
The question nobody could answer
Amnestic mild cognitive impairment, aMCI, is the waiting room of dementia. Memory is measurably worse, independence is intact, the future is blank. Some convert. Many do not.
“Once mild cognitive impairment is established, it’s a big question mark in terms of a prognosis and diagnosis of etiology,” co-founder Kaveh Vejdani, a nuclear medicine physician trained at Stanford and NYU Langone, told Being Patient in 2021.
The standard answers are amyloid PET and spinal fluid assays. They detect pathology. They are weaker at forecasting what a person will do next. Darmiyan says roughly 40 percent of amyloid-positive aMCI patients stay stable for at least five years or return to normal cognition, and that gap is what BrainSee was built to fill.
A physicist who cared for her grandmother
Padideh Kamali-Zare grew up in Iran and left for Sweden at 23. “Just two days after my dear grandmother passed away, I started studying biological physics at KTH so I could help patients with brain diseases to enjoy longer, healthier and higher quality lives,” she told KTH in 2023, when the Royal Institute of Technology gave her its Innovation Award. She had helped care for that grandmother, who had mild dementia, since the age of 14.
She finished her PhD at KTH in 2010, then did postdoctoral work in computational neuroscience in New York and became an adjunct assistant professor at NYU School of Medicine, working with Charles Nicholson on modeling brain tissue microstructure.
Darmiyan is the Persian word for in-between. The company was incorporated in Delaware in September 2016, graduated from Y Combinator in Summer 2017, and by 2019 had an automated imaging pipeline and a Canadian research partnership. Vejdani, now chief medical and technology officer, led the algorithm work.
Breakthrough, then De Novo
IT-Farm, a Tokyo and Palo Alto firm, introduced Darmiyan to Eisai at the January 2020 J.P. Morgan conference. Eisai, co-developer of the anti-amyloid antibody Leqembi, led a $6 million seed round that April, with IT-Farm and Y Combinator alongside.
The FDA granted breakthrough device designation in May 2021. The submission went in on September 29, 2022 and was granted on January 11, 2024, the week of the J.P. Morgan conference.
De Novo is worth being precise about. It is not a premarket approval and not a 510(k) clearance. It is the route for a device novel enough to have no predicate but low or moderate enough in risk to skip the PMA process, and granting one creates a classification. The FDA wrote a new regulation, 21 CFR 882.1454, a new product code, QWT, and a new device type: prognostic assessment software of mild cognitive impairment. The classification text says the type “is intended for adjunctive use and not intended as a stand-alone diagnostic or prognostic tool.”
Authorization is not a business
In September 2024 Darmiyan launched a patient-facing platform with remote visits and MRI prescriptions. That December it joined the CMS GUIDE dementia model through the Five Star Home Health Care consortium. In November 2025 the California Association of Health Facilities named it the only preferred provider for cognitive health, covering more than 1,300 member facilities. In August 2026 Filament Health, a subsidiary of the listed Canadian company Rhelion Life Sciences, signed a non-binding letter of intent to explore BrainSee for patient selection in a possible psilocybin trial.
What is proven, and what is still claimed
| Evidence | What the record shows | Source type |
|---|---|---|
| De Novo DEN220066 | Received Sep 29, 2022, granted Jan 11, 2024, decision code DENG. New regulation 21 CFR 882.1454, Class II, product code QWT, Neurology panel. No change control plan authorized. | Public record |
| Authorized use | Aid the assessment of how likely patients aged 55 to 95 already diagnosed with aMCI are to progress to Alzheimer’s dementia within 5 years. Prescription only. The indication states: “BrainSee is not indicated as a stand-alone diagnostic or prognostic tool. BrainSee should not be used to diagnose aMCI.” | Public record |
| Pivotal performance | Retrospective, 198 patients, 27 sites. Specificity 96.3 percent, sensitivity 72.9 percent, PPV 96.6 percent, NPV 70.6 percent. 86 true positives, 3 false positives, 32 false negatives, 77 true negatives. The lower bound for sensitivity, 64.9 percent, fell below the pre-specified goal. FDA: “the device algorithm is tuned to favor specificity over sensitivity, to reduce the false positives.” | Public record |
| Who was studied | 178 of 198, or 89.9 percent, were White. FDA: “This raises uncertainty regarding whether the device performance observed in the clinical validation study is the same in more representative non-Caucasian patients in the U.S.” | Public record |
| Postmarket study | The 2024 decision summary already required a study in under-represented groups. In May 2025 Darmiyan announced FDA sign-off on that 7-year study, naming NACC, HABS-HD, ADNI and OASIS. | Public record |
| Peer-reviewed pivotal | None. The public evidence is a 2019 bioRxiv preprint and CTAD abstracts. | Not found |
| Screening framing | Releases describe “a fully non-invasive, convenient, and globally accessible screening solution, to be prioritized before other tests or treatments.” The authorization is adjunctive, for patients who already carry an aMCI diagnosis. | Differs from the record |
| The word approval | Darmiyan uses FDA Approval and FDA-approved in headlines and on its product page. The FDA database records the decision as granted, and the company’s own consultancy calls it marketing authorization. | Differs from the record |
| Payment | Alzforum reported a $1,500 list price offered at $300 pending Medicare coverage in January 2024. That September the company said services were “all covered under Medicare with just a nominal out-of-pocket booking fee.” No coverage decision or CPT code found. | Claim outruns the record |
| Category exclusivity | An openFDA query of product code QWT, data last updated Sep 14, 2026, returns one record: DEN220066. | Public record |
Read plainly: the regulatory achievement is real and unusual. Darmiyan persuaded the FDA to write a device class around a prognosis rather than a diagnosis, on 198 retrospective cases. The device is very good at telling a patient she will probably be fine: 3 false positives among 80 non-converters. It is much weaker the other way, missing 32 of the 118 who did convert. The marketing reaches past that, and the validation population was nearly nine-tenths White.
What to watch
- Peer-reviewed publication of the pivotal. Everything public traces to a preprint and conference abstracts.
- First readouts from the 7-year postmarket study, where the sensitivity and demographic questions get settled.
- A Medicare coverage decision or a dedicated CPT code. Without one the economics rest on cash pay.
- Whether the Filament Health letter of intent becomes a definitive agreement.
- A second entrant under product code QWT. That class still holds only Darmiyan.
- A new Form D. Nothing has been filed since October 2023.
In their words
“Just two days after my dear grandmother passed away, I started studying biological physics at KTH so I could help patients with brain diseases to enjoy longer, healthier and higher quality lives.”
Padideh Kamali-Zare, founder and CEO, KTH, 2023 · Independent
“Once mild cognitive impairment is established, it’s a big question mark in terms of a prognosis and diagnosis of etiology.”
Kaveh Vejdani, co-founder, to Being Patient, 2021 · Interview
“It is notable that the FDA approval appears to relate to estimating risk of progression in amnestic MCI (aMCI), since hippocampal atrophy has long been recognized as a predictor of progression from aMCI to AD dementia.”
Nick Fox, University College London, to Alzforum, 2024 · Independent
“Sorry to be the skeptic, but 71.6 percent balanced accuracy (hippocampal volume) versus 84.3 percent (BrainSee) does indeed sound like hippocampal volume is a/the primary feature driving prediction”
Stanford University commenter, Alzforum, 2024 · Independent
“These tools are also only as good as the data they are trained on, and we know there is considerable heterogeneity in AD and related dementias, so it will be important for clinicians to know the size and nature of the samples that were used to develop these algorithms.”
Kate Papp, Harvard Medical School, Alzforum, 2024 · Independent
“Despite decades of investment, progress for patients has remained limited for two main reasons. First, we have not been precise enough about whom we study or how we measure meaningful change.”
Padideh Kamali-Zare, Rhelion letter of intent release, August 2026 · Company-stated
Related companies
Sources
- Public recordDe Novo decision summary, DEN220066, BrainSee
- Public recordForm D filings, Darmiyan, Inc., CIK 0001740573
- IndependentFDA approves BrainSee, AI software that purportedly predicts AD
- IndependentFDA clears AI that predicts Alzheimer’s progression from an MRI scan
- IndependentEarly detection of Alzheimer’s thanks to groundbreaking medical technology
- InterviewA virtual microscope for Alzheimer’s earns FDA breakthrough status
- CompanyDarmiyan receives FDA approval for BrainSee
- CompanyEisai leads a $6M series seed round in Darmiyan
- CompanyFDA approval for post-market surveillance study of BrainSee
- PartnerCAHF selects Darmiyan as preferred provider
- PartnerFilament Health signs letter of intent with Darmiyan
- CompanyDarmiyan website: about, products, news and team pages
Profile researched and written by Healthcare Discovery. Last updated September 29, 2026.
