Free account 2:31Chronological narrative
A cited first draft of the medical story, written from the records in date order, with a citation on every statement, for you to verify.
The records that decide severity
TBI imaging means the CT, MRI, susceptibility-weighted and DTI studies used to prove or dispute a brain injury. A normal CT does not rule out mild TBI.
Damage to the brain's long nerve fibers from rapid acceleration, deceleration or rotation; Medrecords AI reads the imaging.
TBI imaging in litigation means the CT, MRI, susceptibility-weighted and diffusion tensor studies used to prove or dispute a brain injury. They are read together with the first 72 hours of records: EMS and emergency Glasgow Coma Scale scores, loss of consciousness, post-traumatic amnesia and any blood biomarker result. A normal CT does not rule out mild TBI, and DTI findings face admissibility challenges under Rule 702.
Most brain injury cases turn on a severity label, and that label comes from records written by paramedics and emergency staff in the first hours, long before anyone orders an MRI or a neuropsychological test. A normal head CT gets read as proof that nothing happened. A DTI report gets offered as proof that something did. Neither side wins the imaging fight without first winning the fight over the first 72 hours.
For brain injury plaintiff and defense attorneys, legal nurse consultants and neuropsychology IME evaluators. You walk away with the severity criteria mapped to the records that prove them, an imaging modality table, a DTI admissibility decision path, a records request and a first 72 hours worksheet.
Traumatic brain injury is graded mild, moderate or severe. The grade drives the guideline that applies, the recovery curve an expert describes, the life care plan and how a jury hears the word "mild." It is assigned from 4 measures, all recorded in the acute chart, not the imaging suite.
The VA/DoD guideline adds a fifth row for structural imaging. It is the only row where a mild injury and a severe injury can look the same: the table allows normal imaging in all 3 grades.
| Criterion | Mild | Moderate | Severe |
|---|---|---|---|
| Structural imaging | Normal | Normal or abnormal | Normal or abnormal |
| Loss of consciousness | 0 to 30 minutes | Over 30 minutes and under 24 hours | Over 24 hours |
| Alteration of consciousness | Up to 24 hours | Over 24 hours; severity based on other criteria | |
| Post-traumatic amnesia | 0 to 1 day | Over 1 and under 7 days | Over 7 days |
| GCS, best available score in first 24 hours | 13 to 15 | 9 to 12 | Under 9 |
Download this guide as a worksheet (CSV), with Done and Notes columns.
Source: VA/DoD Clinical Practice Guideline, Management and Rehabilitation of Post-Acute Mild TBI, version 3.0, 2021, table 1. Its footnote: a patient who meets criteria in more than 1 category is assigned the higher severity.
That footnote is the plaintiff's most useful sentence in the table. A GCS of 15 on arrival does not make an injury mild if the chart documents 3 days of amnesia; under the table, that is moderate. The defense check runs the other way: does a "moderate" label rest on 1 criterion the records do not support, such as an amnesia estimate written weeks later from the patient's own account?
The GCS was described in 1974 by Graham Teasdale and Bryan Jennett. Its developers publish the current structure and guidance at glasgowcomascale.org. Each component is scored on the best response observed:
The components add to a sum score from 3 to 15. The developers' guidance is specific about 1 trap that shows up in almost every trauma chart: when a component cannot be tested, it should be recorded as "NT" (not testable), not assigned a score of 1, and no total should be reported. A GCS of "3T" or "7T" on an intubated patient is an artifact of that problem. It means the verbal score was not testable. It does not mean the patient scored 1 on speech.
The guidance also lists factors that interfere with the assessment: sedation, paralysis, intubation, language barriers, hearing loss and orbital fractures. Alcohol and drugs are the other big confounder in the emergency department, and the ACRM criteria say a clinical sign counts only when it is not better explained by intoxication, pain, psychological trauma or other confounders.
For 30 years, a widely used definition of mild TBI was the 1993 ACRM definition, a 2-page statement in the Journal of Head Trauma Rehabilitation. In 2023 the ACRM published new diagnostic criteria, built from an evidence review and a 3-round Delphi vote. The final agreement rate among the panel was 90.7%. Expect both sides to use whichever version suits their records, and expect older expert reports to cite only the 1993 text.
Step 2 has 3 doors. The second door, symptoms plus a finding, lets a person with no documented LOC and a normal CT meet the criteria, which is the scenario most mild TBI cases are fought over.
The criteria tie each element to a record. Criterion 2 signs include LOC, altered mental status such as slowed or inappropriate responses, and amnesia around the injury. Criterion 3 symptoms may be delayed a few hours but "nearly always appear less than 72 hours from injury," and can be shown by acute care documentation or an interview about the first days. Criterion 4 covers cognitive, balance and eye movement findings on an acute exam, and elevated blood biomarkers, and notes that most of those tests lose sensitivity over the first 72 hours.
The 2023 criteria help plaintiffs whose acute chart shows symptoms and a positive exam finding but no clear LOC. They help defendants through criterion 6, which makes confounders part of the diagnosis. A diagnosis that ignores a charted blood alcohol level or a painful neck injury is open to attack under the very criteria the plaintiff's expert may cite.
Every severity measure in chapter 1 decays fast. Confusion clears, amnesia is only measurable while it lasts, and the ACRM notes that most exam and lab findings lose sensitivity over the first 72 hours. So the records written in that window are the primary evidence, and their gaps are the battleground. An MRI 3 weeks later can show a lesion. It cannot show confusion at the scene.
The measures are spread across at least 6 sources: the 911 call and dispatch log, the police report, the EMS run sheet, ED triage and physician notes, nursing neuro checks, and the first follow-up visits. The EMS run sheet belongs to the ambulance agency, not the hospital, so a hospital request will not produce it. Nursing neuro checks sit in flowsheets that some chart exports leave out. Between them they hold the serial GCS readings that show change over time.
The timeline below is a labeled hypothetical, built to show how a normal CT and a "mild" discharge can sit on top of records that support a more serious grade, and how the same records give the defense its best arguments.
Caller: "The driver isn't answering me, now he's moving." Dispatch log time-stamps the call.
CAD log p. 1"Bystander states pt out approx 1 min." GCS E3 V4 M6 = 13. "Repetitive questioning, asks what happened x4." Neck pain.
EMS run sheet p. 2GCS 14 (V4). Oriented to person and place only.
EMS run sheet p. 3GCS 15. "Denies LOC." Pain 7 of 10 neck. No components recorded.
ED triage note p. 1"Amnestic to event, last memory is stopping at light, next memory is ambulance." CT head and cervical spine ordered.
ED provider note p. 4"No acute intracranial abnormality."
Radiology report p. 7"Concussion. GCS 15, neuro intact." Head injury instructions. No documented reassessment of amnesia or orientation after 18:48.
ED discharge p. 9No call, portal message or visit. Symptoms on this day rest on testimony alone.
Nothing producedHeadache, light sensitivity, nausea, "feels foggy," trouble finding words at work. Referred to a concussion clinic.
Urgent care note p. 2Plaintiff reads: witnessed LOC, confusion at the scene, documented amnesia and new symptoms inside 72 hours. Defense reads: brief LOC, GCS 15 within an hour, normal CT, neck pain as a competing source of symptoms, and a day with no record.
Build the grade from the worst supported criterion
Test every criterion against its source and its confounders
A noncontrast head CT is the emergency test for head trauma. It answers the first hour question: is there bleeding, swelling or a fracture that needs a neurosurgeon. It was never designed to prove the absence of a brain injury. The ACRM 2023 criteria say neuroimaging "is not necessary to diagnose mild TBI" and that its main clinical role is to rule out injuries that might need neurosurgical or other medical intervention.
The ACR Appropriateness Criteria for head trauma rate each imaging test by clinical scenario. For acute mild head trauma, the ACR splits patients by whether a clinical decision rule calls for imaging; its references include the Canadian CT Head Rule and the New Orleans Criteria.
| Scenario | CT head without contrast | MRI head without contrast | MRI head with DTI |
|---|---|---|---|
| Acute mild (GCS 13 to 15), decision rule says no imaging | Usually not appropriate | Usually not appropriate | Usually not appropriate |
| Acute mild, decision rule says image | Usually appropriate | Usually not appropriate | Usually not appropriate |
| Acute moderate (GCS 9 to 12), severe (3 to 8) or penetrating | Usually appropriate | Usually not appropriate | Usually not appropriate |
| Acute, unchanged exam, normal first imaging: short-term follow-up | May be appropriate | May be appropriate | Usually not appropriate |
| Acute, unchanged exam, positive first imaging such as a subdural hematoma: follow-up | Usually appropriate | May be appropriate | Usually not appropriate |
| Acute, new or worsening neurological deficit: follow-up | Usually appropriate | May be appropriate | Usually not appropriate |
| Subacute or chronic, unexplained cognitive or neurological deficits | Usually appropriate | Usually appropriate | Usually not appropriate |
Source: ACR Appropriateness Criteria, Head Trauma, rating tables for variants 1 to 7. The same tables rate functional MRI, SPECT brain perfusion and FDG-PET/CT "usually not appropriate" in each of those scenarios.
2 points carry weight in litigation. A normal ED CT and no early MRI is the expected pattern for acute mild head trauma, so a missing early MRI proves nothing by itself. And MRI becomes "usually appropriate" once a patient has unexplained cognitive or neurological deficits in the subacute or chronic phase, which is where most litigated mild TBI claims sit.
27% of mild TBI patients with a normal admission CT had an abnormal early MRI, done a mean of 12 days after injury. Prospective study of 135 patients at 3 Level I trauma centers. Yuh and colleagues, Annals of Neurology, 2013.
Both sides cite this study. For plaintiffs it shows a normal CT does not mean a normal brain. For defendants it sets limits: trauma center patients, MRI at about 12 days rather than months later, and 73% of CT-negative patients still had a normal MRI.
| Modality | What it shows well | Limits in a TBI case |
|---|---|---|
| CT head without contrast | Acute bleeding, fractures, swelling; fast | Misses small contusions and microscopic axonal injury |
| Conventional MRI (T1, T2, FLAIR) | Contusions, small bleeds, white matter lesions | Often done weeks or months later; lesions must be tied to trauma |
| SWI or gradient echo | Blood products, including microbleeds from axonal injury | A sequence within an MRI; the report should say if it ran |
| Diffusion tensor imaging (DTI) | Group differences in white matter, as values such as fractional anisotropy | No large normative database; not specific to TBI; ACR "usually not appropriate" |
| Functional MRI, SPECT, FDG-PET | Activity, blood flow or metabolism | ACR "usually not appropriate" for head trauma |
Diffuse axonal injury (DAI) is damage to the brain's long nerve fibers, the axons, caused by rapid acceleration, deceleration or rotation of the head. Much of it is microscopic. When small blood vessels tear along with the axons, the injury leaves tiny bleeds, and those bleeds are what imaging can see. The Yuh study calls them foci of hemorrhagic axonal injury.
Conventional sequences miss many of them. Susceptibility-weighted imaging (SWI) is an MRI sequence built to pick up blood products. A 2006 review in Archives of Physical Medicine and Rehabilitation by Ashwal and colleagues describes SWI as "more sensitive in detecting hemorrhagic lesions seen with diffuse axonal injury (DAI) than conventional imaging," and reports that in children with TBI the number and volume of hemorrhagic DAI lesions correlated with neurologic disability and with impairments of intelligence, memory and attention.
Adjusted odds ratios for a poorer Extended Glasgow Outcome Scale score at 3 months. MRI findings kept their predictive value after adjusting for the CT findings and for demographic, clinical and socioeconomic factors.
That is the strongest peer-reviewed support for the claim that early MRI findings in mild TBI mean something for outcome. It is also narrower than it is sometimes presented. The finding was 4 or more foci, not 1. The outcome was measured at 3 months, not years. And odds ratios describe groups; they do not tell an expert how a specific person with 2 microbleeds will do.
Diffusion tensor imaging is an MRI technique that measures how water moves through brain tissue. In healthy white matter, water moves more easily along the fibers than across them. Software turns that into values per region, most often fractional anisotropy (FA), from 0 to 1. Lower FA is read as a sign of disrupted white matter. With a normal CT and a normal conventional MRI, a DTI report showing low FA is often the plaintiff's only imaging evidence, which is why it draws so many challenges.
"A witness who is qualified as an expert by knowledge, skill, experience, training, or education may testify in the form of an opinion or otherwise if the proponent demonstrates to the court that it is more likely than not that: (a) the expert's scientific, technical, or other specialized knowledge will help the trier of fact to understand the evidence or to determine a fact in issue; (b) the testimony is based on sufficient facts or data; (c) the testimony is the product of reliable principles and methods; and (d) the expert's opinion reflects a reliable application of the principles and methods to the facts of the case."
The 2023 amendment, adopted by the Supreme Court in April 2023, took effect on December 1, 2023. It made 2 changes. It spelled out that the proponent must show each requirement is met more likely than not, and it amended (d) to emphasize, in the committee note's words, that each expert opinion "must stay within the bounds of what can be concluded from a reliable application of the expert's basis and methodology." The note also says many courts had treated the sufficiency of an expert's basis and the application of the method as questions of weight for the jury, and calls those rulings "an incorrect application of Rules 702 and 104(a)." For DTI, that matters most at step (d), where group research meets 1 patient. A lab can use a reliable method in general and still apply it unreliably to 1 patient. State courts apply their own versions of the rule, and some still use a general acceptance test, so check the standard in your jurisdiction before building a challenge or a defense of 1.
A widely cited critique is a 2011 article in the Journal of the American Academy of Psychiatry and the Law by Wortzel and colleagues, which anchored a review of the DTI literature to the Daubert factors. Its main points:
The same article gives the plaintiff something too. On whether the technique can be tested, the authors wrote that DTI as applied to mild TBI "fares well." They said their analysis "does not dictate the admissibility of DTI evidence in any given instance," and that with experienced labs and clinically unambiguous mild TBI, DTI may add a quantifiable measure of white matter integrity, though in those cases, they added, it would be superfluous evidence for an otherwise well-established mild TBI. The article is 15 years old. Expect the plaintiff's expert to say the field has moved on, and the defense to answer with the current ACR and VA/DoD positions.
What a proponent usually has to show
What an opponent usually tests
Glial fibrillary acidic protein (GFAP) and ubiquitin C-terminal hydrolase-L1 (UCH-L1) are proteins found in brain cells, GFAP in glial cells and UCH-L1 in neurons. After a brain injury they appear in the blood. The FDA granted De Novo authorization to the first test that measures both, the Banyan Brain Trauma Indicator, under decision number DEN170045. When a GFAP or UCH-L1 result appears in an ED chart, both sides will read it.
The indication is narrow. The test uses serum drawn within 12 hours of a suspected head injury, in patients 18 and older with a GCS of 13 to 15, "along with other available clinical information." A negative result "is associated with the absence of acute intracranial lesions visualized on a head CT." In other words, it helps decide whether a CT is needed. It was not authorized to diagnose a concussion or to measure how bad a brain injury is.
The FDA summary estimated that the test could avoid about one third of unnecessary CT scans. Its required labeling includes this limitation:
"A negative result is generally associated with the absence of acute intracranial lesions. An appropriate neuroimaging method is required for diagnosis of acute intracranial lesions."
ACRM 2023 counts "elevated blood biomarker(s) indicative of intracranial injury" as a criterion 4 finding. The evidence it cites: in 1 study, GFAP was detectable within 1 hour, peaked at about 20 hours and kept its accuracy beyond 72 hours, though it may track CT findings less well in older adults. The VA/DoD guideline's recommendation 4 suggests against serum biomarkers to establish a mild TBI diagnosis in the post-acute phase.
Imaging and biomarkers answer whether there was structural injury. Neuropsychological testing answers a different question: how the person thinks, remembers and processes information now, compared with norms or with their own earlier performance. In most mild TBI cases with normal imaging, the neuropsychologist's report is the main evidence of lasting deficits, and the defense's IME neuropsychologist is the main rebuttal.
The VA/DoD guideline's recommendation 6 suggests against comprehensive neuropsychological testing during the first 30 days after a mild TBI. A full battery done in the first weeks may be challenged as premature. Screening tests done in the ED or at a concussion clinic are a different thing, and they can supply the criterion 4 cognitive finding under ACRM 2023.
Performance validity tests (PVTs) check whether a person's test scores reflect their true ability, for example whether they gave full effort. Symptom validity tests check whether symptom reports are credible. The American Academy of Clinical Neuropsychology updated its consensus statement on validity assessment in 2021, with 20 experts reaching agreement in January of that year. Its conclusion is direct:
"All clinical and forensic evaluations must proactively address the degree to which results of neuropsychological and psychological testing are valid."
This lines up with ACRM criterion 6, which says criterion 4 test findings must not be better explained by "factors influencing the validity of the symptom reporting or test results." A neuropsychology report in a TBI case that has no validity measures, or that reports failed validity measures and then interprets the scores anyway, is open to attack from both directions.
Where the rebuttal report is weakest
Where the deficit report is weakest
Many psychologists will release test materials and raw scores only to another psychologist, citing test security and their professional ethics obligations. Plan early for a psychologist-to-psychologist exchange or a protective order. A report without raw data is hard to test, and a request made late in discovery will be fought on those grounds.
Lawyers and nurse consultants rarely read the images. They read the report, and the report has a structure that tells you more than its impression line. The mock report below is a hypothetical, built from the fields every radiology report carries. The numbered notes say what to check in each.
Overnight emergency CTs are often read first by an on-call or remote radiologist, then finalized later. The 2 versions do not always match. The hypothetical below shows the kind of change that can move a case from "normal CT" to mild TBI "with neuroimaging evidence of structural intracranial injury" under ACRM 2023.
If the patient was discharged at 21:05 on day 0, the treating team saw only the preliminary read. Whether anyone acted on the final read, and whether it was communicated, is its own line of inquiry.
A TBI file suits AI medical record review in 1 way and not another. The severity fight lives in text: EMS narratives, triage notes, neuro checks and discharge templates spread across thousands of pages and 6 or more providers. A large language model (LLM) can read, sort and cite those entries faster than a person. But the imaging is pixels, the neuropsychology is raw data held by a psychologist, and every judgment about severity and cause belongs to an expert.
The fixes are the ones courts now expect. Ground every line in a retrieval-augmented, page-level citation so a person can check it in 1 click. Keep a human-in-the-loop reviewer who checks the cites before anything leaves the firm. Lawyers have been sanctioned under Rule 11 for filing briefs with fabricated AI-generated citations, and an invented clinical fact in an expert disclosure is the same failure in a different document.
Every GCS, LOC and amnesia entry links to its source page. No citation, no use.
Brain injury files hold PHI, psychiatric history and sometimes substance use records. Get the business associate agreement before the first upload.
EMS and scene records must go through OCR with low-confidence pages flagged, not smoothed over.
Imaging should sit on the timeline for the expert to read. A tool that claims to diagnose from DTI is a liability at deposition.
No severity grade, no causation opinion, no admissibility call. Legal AI that claims those shifts risk onto you.
For a wider look at AI document review tools for neuropsychologists and IME evaluators, see the . For the imaging question on its own, see can AI read DICOM imaging, and for the court question, is AI accurate enough for court.
Copy these into your own forms. Adjust the legal basis line to your jurisdiction and to whether you are sending a HIPAA authorization, a subpoena or a discovery request. For the mechanics of each route, see how to get medical records for a lawsuit.
Send to each provider in the first 72 hours chain and to every imaging facility. Ask for each item by name; a general "medical records" request often misses the run sheet, the flowsheets and the images.
RE: [PATIENT NAME], DOB [DATE]
Date of injury: [DATE] Dates of service requested: [DATE] to [DATE]
Legal basis: [HIPAA AUTHORIZATION / SUBPOENA / REQUEST FOR PRODUCTION]
Please produce the complete record, including:
SCENE AND TRANSPORT
[ ] 911 call audio and dispatch (CAD) event log with timestamps
[ ] EMS patient care report, all pages, including narrative,
every Glasgow Coma Scale entry with components, and times
[ ] EMS cardiac monitor or device data, if retained
EMERGENCY AND TRAUMA CARE
[ ] ED triage note and triage flowsheet
[ ] ED physician and trauma team notes, including addenda
[ ] Nursing flowsheets and neuro checks (GCS, pupils,
orientation) for the full stay
[ ] Laboratory results with collection and result times,
including blood alcohol, drug screen, and any GFAP or
UCH-L1 test (device name and reference cutoff)
[ ] Discharge instructions and discharge summary
IMAGING
[ ] Every radiology report version: preliminary, final and
each addendum, with timestamps and reader names
[ ] Images in native DICOM format for every CT and MRI,
all series, not screenshots or a viewer-only disc
[ ] For any DTI study: raw diffusion series, scanner make,
model and software version, acquisition parameters,
analysis software and version, region selection method,
and a description of the normative comparison data
[ ] Any prior brain imaging on file for this patient
FOLLOW-UP
[ ] Urgent care, primary care and concussion clinic notes
for the first 30 days, including symptom scales and
any cognitive, balance or eye movement screening
Please produce electronic records in native electronic
format where available, and certify completeness.
[NAME] [FIRM] [CONTACT]
Fill in 1 row per entry, in time order. The last 2 columns are the ones experts read first. Fill them in as you go.
CASE: [NAME] INJURY TIME: [DATE, HH:MM, SOURCE] CLOCK OFFSETS NOTED: EMS [+/- MIN] ED [+/- MIN] CT [+/- MIN] TIME | HOURS FROM INJURY | SOURCE, PAGE | AUTHOR ROLE | GCS (E/V/M or NT) | LOC | CONFUSION / AOC | AMNESIA (PTA) | SYMPTOMS | EXAM FINDING | CONFOUNDER NOTED | CONFLICT WITH | GAP BEFORE NEXT ENTRY SEVERITY CHECK (VA/DoD table: highest category met wins) Best GCS, first 24 hours: [ ] at [TIME], page [ ] LOC duration: [ ] source: [WITNESS / PATIENT / NONE] AOC duration: [ ] source: [ ] PTA duration: [ ] first documented: [DATE, PAGE] Imaging: CT [NORMAL/ABNORMAL, PAGE] MRI [ ] SWI run [Y/N] ACRM 2023 CHECK Criterion 1 mechanism: [ ] Criterion 2 signs: [ ] Criterion 3 symptoms under 72 hours: [ ] Criterion 4 exam or lab findings: [ ] Criterion 5 imaging: [ ] Criterion 6 confounders to address: [ ] Mild qualifier removed? LOC over 30 min [ ] GCS under 13 after 30 min [ ] PTA over 24 hours [ ]
Medrecords AI is medical chronology software. It works on the records you upload; it does not request or retrieve records from providers. On a TBI file it does 5 things:
What it does not do: it does not read or interpret CT, MRI, SWI or DTI images, does not grade severity, does not decide causation or admissibility, and does not score a case. Flags are signals, not verdicts. It runs under SOC 2 and HIPAA with a signed BAA (see security and HIPAA). Self-Service bills 10 cents a deduplicated page, down to 5 cents at volume, duplicates free; Enterprise On-Prem is an annual license. More on the use case: traumatic brain injury record review. For what happens after severity is settled, see the companion guide on TBI causation and long-term impact.
Book a demo on a brain injury file you are working, then run your first case free on us. Every line comes back cited to its source page. You review, you revise, you sign.
Scheduling only. No records move from a public page.
3 short guides from our help center, recorded on a demo case. They play inside the app: sign in or create a free account, and the video opens at the chapter you picked. First case is free on us.
Free account 2:31A cited first draft of the medical story, written from the records in date order, with a citation on every statement, for you to verify.
Free account 2:35Every medical event on its date, in order, tied to the page it came from. Filter it, search it and export it as PDF or Word.
Free account 2:30Every record the case documents mention that the file does not contain, grouped by provider, with the page that mentions it.
Every figure, cutoff and criterion here was checked against the primary source in September 2026. Quotes are verbatim. The crash timeline, the MRI report and the CT report versions are hypothetical. No court opinion is cited because none was reviewed. Nothing here is legal or medical advice.
Related guides: TBI causation and long-term impact, blunt force trauma medical records, eggshell plaintiff and pre-existing conditions, how to write a medical chronology. Related solutions: sports concussion and CTE record review, motor vehicle accident record review, personal injury defense, legal nurse consultants.