So You Want to Read MSK Radiography (July 2023)



Donna Magid MD, M.Ed, FAUR   July 2023

Residents and Fellows--The insurance- and Billing-mandated FORMAT is not open to individual preferences or opinions, it has gotten VERY rigid in the last year or so. Billing requirements have to do with insurance rules and the rigidity (unintelligence) of A.I., as well as optimizing the ICD 10 codes. They will often seem …suboptimal to the physician generating a perfectly coherent and precise report based on medical significance rather than on meeting bureaucracy demands. The Department was encountering absurd Billing deficiencies (underpaid or unbilled exams) for years, which has improved as we aggressively pursue/accept the Format details. Make sure your work’s value is recognized! The Formatting rules on which I elaborate are the result of hours of consultation with the Billing personnel, as well as the information disseminated to us from faculty leadership. (Apologies for any errors, likely mine alone, in interpreting/describing these national guidelines.)

Different faculty read differently, as junior residents rapidly learn. (For example, I rarely if ever use the word ‘normal’, whereas another radiography attending may insist on it—we are both correct, this is an art as well as a science). Since I will often be your 1st yr. checker on MSK Rads, the non-negotiable Format guidelines are followed by a more editorial discussion of how I’d suggest you approach reports, language and specific areas. 

Residents-- Please read the MSK handouts on for a beginner’s intro/entrance point to many common MSK radiography studies (fxs, ankle, knee, leg length, scoliosis…). The MSK Rads Curriculum will offer expanded study options once we dive in.

REPORT FORMAT:. Not negotiable, has nothing to do with preferences, medical style, how you were taught, opinion, or precise medical English language usage.  A.I. now scans reports for billing and needs to find certain words, colons, structure. This is like a series of keys searching for very precise locks and ignoring all else. What it cannot ‘see’/comprehend (eg abbreviations) it can not bill for.  

And all dictation must be inside the relevant BRACKETS!! These are annoying [ ] for each header (Indication, Exam, Findings...) and the commentary MUST be inside those brackets or Billing rejects.  If you try to sign off and get that heavenly harp chord alerting you to "Unfilled fields, are you sure you want to sign?" notice -- go back and cut and paste into the brackets. Making the brackets a contrasting color to whatever background field color you select for reporting helps.

EXAM: XR (side)(body part) (#views)Even self-evident “AP and Lateral” (Chest or Scoliosis) needs us to add ‘2 views”. (“AP Chest” does not need a ‘one view’). Number of views VERY important for accurate Billing—often the wrong ICD10 code is attached, and therefore mis-billed, leading to either under-billing or to potentially fraudulent over-billing.

If study is ‘Bilat”, needs to be separated into R and L and entered as two lines eg :Bilat Kneesà

EXAM: XR right knee 4 views                                                                                          

EXAM: XR left knee two views”                                                                                            

Knees are always confusing; if the two studies are ‘3 views R knee and ‘one view Bilat Knees’, it becomes:

EXAM:XR right knee 4 views                                                                                            

EXAM:XR left knee one view

FINGERS, PHALANGES, MCs must be NAMED not numbered—change the INDIC or EXAM if it says ‘2d finger’, to ‘index finger’. Billing requires the long finger be labelled “middle’, not ‘long’.

Pelvis and Bilat Hip, Bilat SI Joints, Bone Survey (eg myeloma), Lower Extremity for Leg Length, Shunt Survey—need not be separated into R/L or body parts

Bilat hips—if there are separate coned laterals of each hip, exam is 3 views.

Multiple exams (“Scoli, Lumbar spine, L Ankle…”) can/should be reported on one report—clinician doesn’t want to have to open 3 separate reports—but each gets its ownEXAM: XR….” Line, ditto IMPRESSION lines. Yes, that means some Rheum reqs—eg ‘bilat hands bilat wrists, bilat hips, bilat SI joints, bilat knees, bilat feet’—will end up having literally a dozen “EXAM:XR..” lines. These rules are driven by Billing/AI, not by common sense or linguistic grace!

And while the Hand and Wrist Findings/Impressions may have overlap (ie one can in Impression say “Right wrist and right hand…”, “Left foot and left 3rd toe…”, ‘see above’ not allowed.

Nor, for multiple body parts like a polytrauma, with shoulders, spine, wrist, ankle, etc, can the Impression roll unassociated parts together ie cannot summarize as ‘No definite abnormality visualized shoulder, wrist, ankle”

Reading tip for Bilats: Newest images show up on L screen.  Drag it also to the R screen.  Then scroll images so same view of L and of R are on L and R screens respectively--( ie L frontal/R frontal, L lateral/R lateral, etc) allows easier compare/contrast of R/L  hands, wrists, feet, ankles since findings often similar/related.

MULTIPLE EXAMS ie sequential views CHEST or Fx being manipulated and re-imaged 3x in clinic: Each gets its own EXAM line with time, or insurance rejects:

                              ‘EXAM: XR Right wrist 1207 hrs…

                               EXAM: XR right wrist 1240 hrs..

                               EXAM: XR right wrist 1257 hrs’… or AI bills for just ONE exam.

(The IMPRESSION can then roll together multiples, but mention # of studies and outcome: “3 sequential studies right wrist 1207 through 1257 hours, fracture reduction in progress, with final image demonstrating…” “5 sequential exams chest 0907 through 1048 hrs show nasogastric tube being advanced, final position with tip and side hole…”)


INDICATION: NO abbreviations ANYWHERE! AI cannot recognize and deems report inadequate to bill. If clinician wrote "ORIF", "CRPP", "BBFF", ‘OOC”, “TTP”, etc –especially Indication-- either spell it out or substitute other words.

”F/U”, “R/O”,”Post-op”, the ubiquitous “Trauma”, or ‘Pain” need expanding to include a body part ie ”F/U wrist fracture’, “Trauma possible ankle fracture”, “Pain left foot”. Type additional info in.

Indication/HX: Spelunk spelunk spelunk! The oft-given ‘pain’ or ‘R/O” are useless--and we cannot bill for “Pain…”Trauma…”Rule out…”Assess…”.  EPIC makes it slightly easier to distinguish EMed or Ortho notes from, say, Ophthal or Phlebotomy; so open and explore—and add to Indications.  “Pain wrist 18 months, no hx of trauma”, ‘Stepped on glass last night, plantar pain foot’, ‘Marathoner, 2 weeks anterior tibia pain”, “Ehlers Danlos, chronic sprains”, “Slipped off curb in high heels”— besides allowing better Billing, all such information make your reading more targeted, more accurate, and improve your clinician/radiologist relations.   If no old images in Hopkins computer, say so.  If outside images in Ambra mentioned, pull them up (yes this takes time)..                                                                              Technologists getting great at adding history, be sure to check for their comments in Epic box where tech name stated. If they added history, add it to Indications as : "Technologist adds, 'Fell 5 days ago, now point tender over 4th and 5th MT heads'.  "Technologist adds,' Surgery 17 yrs ago, chronic pain since, worse since 3 months ago..."  Their comments have to be in quotes. And thank them occasionally (cc me)!.

Number of Indications must match number of exams—and use language likely to be used by the clinician. “LLD and Bilateral Knees”, eg, must be split as Exams into three Exams (the LLD, the R, and the L, Knee); and the Indication must have TWO components- “Gait abnormality’ or “leg length discrepancy” for the LLD, and ‘Knee arthritis” or “Knee pain”. Clinicians poor at providing TWO Indications, add whichever is missing. .  

COMPARISON:Date (or "None"). Sometimes reformatted CT, MR. UGI, abdo XR, provides limited comparison esp. for spine, pelvis. Once a comparison is documented it then needs to be referred to somewhere in Findings or Impression: “Without change from…” “Progressing on the left and not changed on the right since…” “Appearing since..”

FINDINGS:Everything, in a hierarchy from important/major progressing to the trivia. Or, addressing the clinical query first and then other findings. Or, when there are no major findings and the clinical query vague, organizing your observations—proximal to distal, for instance.

ALWAYS start with LIMITATIONS of images/study :

    “C7, T1 cut off; no odontoid view; otherwise unremarkable”, “Limited exam proximal T spine on lateral due to superimposition”, “In cast, markedly limiting detail”, “Habitus markedly limits axillary view exposure”, “Superimposed surgical material limits detail”, “Left hand, one view only”, “Abdomen, supine-only limits assessment for air/fluid levels or free air”; “Fluoroscopy, limited exam”,  “Visualized portions grossly unremarkable…” "Alignment maintained between flex/ex, visualized levels..." alerts clinicians that even if what is seen is unremarkable, we didn’t see it all.

FINDINGS, if simple/brief, may be abbreviated as “See below” (ie IMPRESSION will summarize all important information: “Right forearm distal diametaphyseal fracture radius with progression of healing and no change in alignment”; “Post operative left ankle unchanged”; “Chest with lines and catheters unchanged from 4 hours ago”…). However IMPRESSION can never be ‘See above)

IMPRESSION: Must mention study/side as was given in EXAM; therefore start with "Right shoulder..." Bilateral hands: Right....Left...".

This reiteration of the study/side is a Billing/AI requirement that has nothing to do with report clarity, since it would be obvious to a human that an Impression of 'Left radial head fracture' refers to an elbow; or “Unremarkable except subacromial narrowing “ implies shoulder—but A.I. is not a human.                            

When any bilateral findings very similar, one can (esp in Impression)  say 'Nearly-symmetric bilateral hands, with demineralization, trapezium-thumb MC osteoarthrosis, no evidence of subluxations, erosions, of soft tissue calcifications." Don't force the clinician to read 2 identical paragraphs, one for R and one for L. If significant differences, report R/L separately.

If multiple related/physically contiguous studies (R ankle, R foot, R tib-fib) are all UNREMARKABLE one can roll them together into one line in IMPRESSION but must NAME each one: “Right foot, right ankle, and right tibia-fibula-- all without significant abnormality”.                           If the findings are not identical, separate each on its own Impression line.

‘Headlines’--significant/relevant/current info only—go into the Impression; omit the incidentalomas, extra detail, normal variants, or the ancient history. 

'Advise F/U if clinically indicated’, or 'Further F/U coned and centered to any area of interest", MUST be stated when reading with me. IMHO should be mandatory across board; helps ward off potential litigation or blame-the-radiologist disputes. Whether exam limited or just unremarkable (apparently negative), advising F/U if physical findings, mechanism of injury, other comorbidities, soft tissue clues, etc. (ie all those things the clinician knows but the radiologist cannot) is pragmatic both clinically and medically-legally. Many of our trauma images are initially negative but become more definitely radiographically abnormal 10-14 days later. Negatives before that time may be false.

       “Habitus (or exposure, or rotation, or superimposed foreign matter, or cast, or….)   limits detail, no definite fracture but F/U coned to any area of interest could be obtained if clinically indicated*

          *If advising F/U, add “If clinically indicated”— otherwise some clinicians have said they feel they are legally obligated to get more studies, even if NOT warranted in their judgment (and they do have the pt. at hand—we don’t).

Possibly serious Incidental findings  (lung nodules, neoplasm, foreign matter...)  should be pursued vigorously to see if previously seen/documented/changed (eg—?nodule seen on today's shoulder? Look not only for older shoulders, for example,  but frontal C spines, Chest, CT, upper GI, Scoliosis-- anything that may allow us to document 'unchanged for more than 2 years"). If no prior documentation/assessment, or change, or anything worrisome, Urgent Notification /Critical macros documentation require.


Anatomic position reference points do not change with position—ie ‘anterior’, ‘medial’, ‘distal’, etc remain uniform even if the appendage in question is amputated and hung from the ceiling.  Remember your Gross Anatomy cadaver:  supine, hands supinated, feet plantigrade.  Vague descriptors such as ‘near’, ‘next to’, behind’, above’—meaningless and DANGEROUS. “Apposed” for the two articular surfaces of a joint is fine.

Magid 10 Commandments ( include “LEND NO FOOTHOLD TO LAWYERS” and “BE LITERAL/VERITABLE WITNESSES”.  Choose words carefully and precisely; never ‘assume’ anything unseen or poorly seen can be guesstimated. (example: ”Visualized cervical spine unremarkable” as an Impression would reinforce the Finding’s opening line:”Limited exam of C7 and T1, no odontoid view”.

Report should be structured like newspaper front page: HEADLINES (“Unchanged from last exam…”, “Limited exam…”, “Subtle fracture scaphoid …”, “Cannot r/o lytic lesion proximal diaphyseal…”, followed other important/relevant/focused info, then incidental or miscellaneous detail (“Again seen is bone island…’Unchanged surgical clips…”;”Battery pack again superimposes iliac wing…”). 

Radiographs are 2D representations of a 3D structure.  Unless there are at least 2—for many joints 3-- orthogonal (right angles) views, one can only say that “AA is SUPERIMPOSING BB”, not ‘in BB”; or one can say there is “apparent…”.   Remember: “One View Is No View”.

Don’t express excessive certainty when one cannot be sure— “Apparent…”, “Incompletely seen…”, “Grossly unremarkable on this limited exam…” ,“due to only one view…”, or “poorly seen due to habitus”… , or” incompletely seen at edge of image” are important modifiers. 

Normal” is a word I use far less than most. If the pt. were truly normal, their clinician might not be requesting imaging.  “Unremarkable”, "No significant abnormality visualized areas",  or ‘Radiographically unremarkable” are the best substitute.  “No definite abnormality visualized portions”, “No significant radiographic abnormality” all leave room for nasty future surprises, initially radiographically invisible findings, or currently incomplete/limited exams. Also in an era when patients read our reports, leads to less apparent discrepancy between the clinician's possible perception of abnormalities requiring treatment or follow-up which were not revealed or visible to the radiologist. The report of 'normal' radiograph in a clinically abnormal pt. requiring further assessment may falsely reassure a patient they need not comply.  

 Also with fractures, if date of injury known, “Advise F/U 10-14 days following trauma or onset of sx if clinically indicated”—F/U sooner is NOT a true R/O for subtle fxs. It’s the osteocytes—they need 10-14 days to make a radiographically visible change removing debris to make a fracture line ‘appear’ (osteoclasts) and/or to lay down and start calcifying enough matrix to show healing/reactive change (osteoblasts). When the date of injury/onset is unknown just leave it at ‘F/U as clinically indicated..” since, eg, advising 14 day f/u for a 3 month old injury is meaningless.

In MSK one cannot use the word stable to mean 'unchanged'; it more commonly implies biomechanical or physical stability, which is not an imaging determination. If a finding, alignment, hardware, whatever are unchanged from last exam say so – ‘unchanged’, ‘again seen is’, but NOT  'Stable alignment proximal tibial fracture..."

Our referring clinicians also include rheumatologists, plastic surgeons, internists, nurse practitioners, ambulatory care , Comp Clinic, ob-gyn...our job is to describe and report with non-ambiguous comprehensible clarity. Fancy Latin names for many normal or anatomic variants are counter-productive if the image was requested by non-Orthopaedists.  Describe it as 'normal variant accessory ossicle", not 'os hamuli proprium". Prefacing phrases like 'ulna minus' with 'anatomic variant..." is also helpful if you are not sure the referring clinician is a hand specialist. 

           REMINDERS, SPECIFIC BONES AND JOINTS (see specific Joint handouts)

BILAT KNEES  Knees if quite similar can be read together in Findings and Impression: eg “Early medial compartment narrowing bilaterally left slightly > right with decreasing valgus; left joint effusion, no R lateral.  IMPRESSION: approximately symmetric osteoarthrosis L>R”                                                                                                

Knees if quite different should be split in Findings/Impression:  “R knee arthroplasty unremarkable and unchanged….Left knee medial tibial plateau fracture unchanged except for progression of healing, since last exam…”

SCOLIOSIS Intro/template is on, under MSK handouts. COUNT pairs of ribs and ascertain number of lumbar vertebral bodies—many of our pts. anomalous; hypoplastic 12th ribs, lumbosacral segmentation anomalies common (Don’t miss the PSEUDARTHROSES!); and many lawsuits drifting around based on fact that radiography/CT/MR reports over time may each call the same lumbar abnormal level by different  numbers due to FOV differences.

       Currently (initiated Fall 2017) we are NOT dictating actual measurements for Orthopaedic scoli requests* (insurance issue: if our measurements are not precisely matched to Orthopaedics measurements insurance may deny patient care payment; there is about +/- 5 degrees standard deviation of two reviewers measuring from same landmarks, we are avoiding giving insurance an excuse to deny claims:  "No definite change in right thoracolumbar scoli compared to 6/2017, see clinician measurements. ".

Add "Scoliosis technique, which markedly limits detail to decrease dose", since the markedly low resolution of EOS and other scoli techniques will preclude reliable assessment of subtle trabecular changes, early implant/bone interface changes, fractures, etc. Describe if in brace or out; if supine or sitting; esp if comparing to older exam which may be different.  In-brace can’t be compared to out-of-brace; nor upright to supine; it’s apples-to-oranges.

       *There will be times, here or in future, one *does* measure the scoliosis (for non-Orthopaedists here, or at future employment):  Dictate measurements “From superior endplate of …to inferior endplate of …” If measuring-- Comparisons to older images, or to changes when pt. flexes laterally to L or R, must use same endplates, eg ‘Using the same landmarks, on previous exam Oct. 19, 2015 these curves measured XX and YY respectively…”

       Curves are named ‘right/dextro’ or ‘left’ by direction of shift from midline ie apex of curve; need not add “convex to…” unless you are unclear abt. This convention.   

       Measure-remeasure SD is about +/- 5 degrees between 2 observers using same landmarks, 3 degrees if same observer.  Less than 5 degree change therefore not significant.  IF HOWEVER there is ~5 degree increase since last exam, look for older exams to see if curve is sneaking up 5 degrees each time, therefore stealthily ‘increasing’ -- “L. 31 degree curve superior endplate of X through inferior endplate of Y. On last exam Oct 10, 2017 using same landmarks, this measured 27 degrees; but on older exam May 3, 2014 this measured 16 degrees, suggesting longitudinal increase”

FRACTURES  Read, memorize, and use Fractures 101 handout on, under Elective/MSK handouts. Orthopaedic convention (describe distal relative to proximal, etc) has very precise and well-described descriptors. 

HANDS, WRIST:  NAME, do not NUMBER digits, phalanges, MCs.

     “Thumb, Index, Middle, Ring, Small” for finger, phalanx, or MC;

     “prox-middle-distal” Phalanges, not “1st-2nd…”

 LONG BONE images (humerus, forearm, tib fib) :  start with ‘Limited exam wrist and elbows” (“Joint above/below”) , because beam angle/penumbra is giving you subtle obliques, not true AP/Lat, at edges of such images. (ie magnifying to see ends WILL NOT give same accuracy as actual CONED AND CENTERED view over joints) .

HIPS, PELVIS   Position: obturator foramina perfectly symmetric, or pelvis rotated.  Trochanters thrown off neck, which should project as biconcave femoral neck or LIMITED EXAM. "Hip" means proximal femur and acetabulum; be precise localizing findings.(ie one can report arthritis of the hip, but not 'hip fracture'-- it is either a proximal femoral, or acetabular, fracture)  See TeamRads.

FEET, ANKLES mention if weight-bearing (WB) or non-weight-bearing (NWB),since it changes the longitudinal arch and axes of the talus and calcaneus.  Don’t get too involved trying to dx. pes planus/cavus, esp. on non-WB views. Check TeamRads hand out.

SPINE, Flexion/extension C, T or L spine:  a) DID alignment change and b) DID Pt. ACTUALLY MOVE?? "Alignment maintained between F/E but virtually no range of motion..."  or ‘limited motion in extension, adequate in extension...", or ’.alignment maintained, with unremarkable range of motion between F/E". Did you count vertebral bodies? Is there anomaly/pseudarthrosis at the lumbosacral junction? Any spondylolisthesis between flexion/extension?

Happy to discuss any questions or get feedback on above!